Communication system, base station, and communication terminal
The integration of AI/ML in communication systems dynamically configures settings for effective processing across various scenarios, enhancing KPIs like RA processing and energy efficiency by leveraging trained models for optimized PRACH transmission and backoff in mobile communication systems.
Patent Information
- Application Number
- PCT/JP2024/018428
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2023-05-30
- Filing Date
- 2024-05-20
- Publication Date
- 2024-12-05
AI Technical Summary
Current mobile communication systems face challenges in optimizing various Key Performance Indicators (KPIs) such as delay, reliability, connection density, user experience, and energy efficiency across different communication situations, particularly due to limited use cases of Artificial Intelligence (AI)/Machine Learning (ML) techniques.
Implementing AI/ML in communication systems to dynamically configure settings for communication processing based on acquired information from base stations, enabling effective processing in various communication scenarios, including Random Access (RA) processing, by using trained models for optimizing PRACH transmission and backoff processing.
This approach allows for more effective RA processing and overall system optimization, improving KPIs like connection efficiency and energy consumption by adapting to specific communication situations.
Smart Images

Figure JP2024018428_05122024_PF_FP_ABST
Abstract
Description
Communication system, base station and communication terminal
[0001] The present disclosure relates to wireless communication technology.
[0002] The 3rd Generation Partnership Project (3GPP), a standardization organization for mobile communication systems, is considering a fifth-generation (hereinafter sometimes referred to as "5G") wireless access system as a successor to Long Term Evolution (LTE) and Long Term Evolution Advanced (LTE-A), one of the fourth-generation wireless access systems (see Non-Patent Document 1) (for example, Non-Patent Document 2). The technology for the wireless section of 5G is called "New Radio Access Technology" ("New Radio" is abbreviated as "NR"). The NR system is being studied based on the LTE system and the LTE-A system.
[0003] For example, in Europe, an organization called METIS has compiled 5G requirements (see Non-Patent Document 3). The requirements for a 5G wireless access system are that it must have 1000 times the system capacity, 100 times the data transmission speed, one-fifth the data processing delay, and 100 times the number of simultaneous connections of communication terminals compared to an LTE system, while also achieving further reductions in power consumption and cost reductions for the equipment (see Non-Patent Document 3).
[0004] In order to meet such demands, 3GPP is currently studying 5G standards (see Non-Patent Documents 4 to 23).
[0005] The NR access method uses OFDM (Orthogonal Frequency Division Multiplexing) in the downlink direction and OFDM and DFT-s-OFDM (Discrete Fourier Transform-spread-OFDM) in the uplink direction. Also, like LTE and LTE-A, the 5G system does not include circuit switching and is only a packet communication method.
[0006] NR allows the use of higher frequencies than LTE in order to improve transmission speeds and reduce processing delays.
[0007] In NR, which may use higher frequencies than LTE, cell coverage is ensured by forming a narrow beam-shaped transmission and reception range (beam forming) and changing the direction of the beam (beam sweeping).
[0008] The decisions made by 3GPP regarding the frame structure in an NR system, as described in Non-Patent Document 1 (Chapter 5), are explained using Figure 1. Figure 1 is an explanatory diagram showing the structure of a radio frame used in an NR communication system. In Figure 1, one radio frame is 10 ms. The radio frame is divided into 10 equally sized subframes. The NR frame structure supports one or more numerologies, i.e., one or more subcarrier spacings (SCSs). In NR, one subframe is 1 ms long, and one slot consists of 14 symbols, regardless of the subcarrier spacing. Furthermore, the number of slots included in one subframe is one when the subcarrier spacing is 15 kHz, and the number of slots at other subcarrier spacings increases in proportion to the subcarrier spacing (see Non-Patent Document 11 (3GPP TS38.211)).
[0009] 3GPP's decisions regarding channel configuration in NR systems are described in Non-Patent Document 2 (Chapter 5) and Non-Patent Document 11.
[0010] A physical broadcast channel (PBCH) is a channel for downlink transmission from a base station device (hereinafter sometimes simply referred to as a "base station") to a communication terminal device (hereinafter sometimes referred to as a "communication terminal" or "terminal") such as a mobile terminal device (hereinafter sometimes simply referred to as a "mobile terminal"). The PBCH is transmitted together with a downlink synchronization signal.
[0011] Downlink synchronization signals in NR include a primary synchronization signal (P-SS) and a secondary synchronization signal (S-SS). Synchronization signals are transmitted from base stations as synchronization signal bursts (hereinafter sometimes referred to as SS bursts) at predetermined intervals and for a predetermined duration. SS bursts are composed of synchronization signal blocks (hereinafter sometimes referred to as SS blocks) for each beam of the base station.
[0012] The base station transmits the SS block of each beam by changing the beam during the duration of the SS burst. The SS block consists of the P-SS, S-SS, and PBCH.
[0013] The Physical Downlink Control Channel (PDCCH) is a channel for downlink transmission from a base station to a communication terminal. The PDCCH carries downlink control information (DCI). The DCI includes resource allocation information for a Downlink Shared Channel (DL-SCH), which is one of the transport channels described below, resource allocation information for a Paging Channel (PCH), which is one of the transport channels described below, and Hybrid Automatic Repeat reQuest (HARQ) information related to the DL-SCH. The DCI may also include an uplink scheduling grant. The DCI may also include an acknowledgement (Ack) / negative acknowledgement (Nack), which is a response signal to the uplink transmission. In addition, in order to flexibly switch between DL and UL within a slot, the DCI may include a slot format indication (SFI). The PDCCH or the DCI is also called an L1 / L2 control signal.
[0014] In NR, a time-frequency region that is a candidate for including a PDCCH is provided. This region is called a control resource set (CORESET). A communication terminal monitors the CORESET and acquires a PDCCH.
[0015] The Physical Downlink Shared Channel (PDSCH) is a channel for downlink transmission from a base station to a communication terminal. The Downlink Shared Channel (DL-SCH), which is a transport channel, and the PCH, which is a transport channel, are mapped to the PDSCH.
[0016] The Physical Uplink Control Channel (PUCCH) is a channel for uplink transmission from a communication terminal to a base station. The PUCCH carries uplink control information (UCI). The UCI includes Ack / Nack, which is a response signal to a downlink transmission, CSI (Channel State Information), a scheduling request (SR), and the like. The CSI is composed of a Rank Indicator (RI), a Precoding Matrix Indicator (PMI), and a CQI (Channel Quality Indicator) report. The RI is rank information of a channel matrix in MIMO (Multiple Input Multiple Output). The PMI is information on a precoding weight matrix used in MIMO. The CQI is quality information indicating the quality of received data or the quality of a communication path. The UCI may be carried by the PUSCH, which will be described later. The PUCCH or UCI is also called an L1 / L2 control signal.
[0017] The Physical Uplink Shared Channel (PUSCH) is a channel for uplink transmission from a communication terminal to a base station. The Uplink Shared Channel (UL-SCH), which is one of the transport channels, is mapped to the PUSCH.
[0018] A physical random access channel (PRACH) is a channel for uplink transmission from a communication terminal to a base station. The PRACH carries a random access preamble.
[0019] A downlink reference signal (RS) is a known symbol in an NR communication system. The following four types of downlink reference signals are defined: a demodulation reference signal (DM-RS), which is a UE-specific reference signal, a phase tracking reference signal (PT-RS), a positioning reference signal (PRS), and a channel state information reference signal (CSI-RS). Measurements of the physical layer of a communication terminal include reference signal received power (RSRP) measurement and reference signal received quality (RSRQ) measurement.
[0020] Similarly, the uplink reference signal is a known symbol in an NR communication system. The following three types of uplink reference signals are defined: a data demodulation reference signal (DM-RS), a phase tracking reference signal (PT-RS), and a sounding reference signal (SRS).
[0021] The transport channels described in Non-Patent Document 2 (Chapter 5) will be described below. Among the downlink transport channels, a broadcast channel (BCH) is broadcast to the entire coverage of the base station (cell). The BCH is mapped to a physical broadcast channel (PBCH).
[0022] Retransmission control using HARQ is applied to the Downlink Shared Channel (DL-SCH). The DL-SCH can be broadcast to the entire coverage of the base station (cell). The DL-SCH supports dynamic or semi-static resource allocation. Semi-static resource allocation is also called semi-persistent scheduling. The DL-SCH supports discontinuous reception (DRX) in communication terminals to reduce power consumption of the communication terminals. The DL-SCH is mapped to the Physical Downlink Shared Channel (PDSCH).
[0023] The Paging Channel (PCH) supports DRX of communication terminals to enable low power consumption of the communication terminals. The PCH is required to broadcast to the entire coverage of the base station (cell). The PCH is mapped to physical resources such as the Physical Downlink Shared Channel (PDSCH) that can be dynamically used for traffic.
[0024] Among the uplink transport channels, the uplink shared channel (UL-SCH) is subject to retransmission control using HARQ. The UL-SCH supports dynamic or semi-static resource allocation. Semi-static resource allocation is also called configured grant. The UL-SCH is mapped to the physical uplink shared channel (PUSCH).
[0025] The Random Access Channel (RACH) is limited to control information, has a risk of collision, and is mapped to the Physical Random Access Channel (PRACH).
[0026] The following describes HARQ. HARQ is a technology that improves the communication quality of a transmission path by combining Automatic Repeat reQuest (ARQ) and Forward Error Correction. HARQ has the advantage that error correction functions effectively through retransmission even for transmission paths whose communication quality varies. In particular, by combining the reception result of the initial transmission and the reception result of the retransmission when retransmitting, it is possible to obtain further quality improvement.
[0027] An example of a retransmission method will be described below. If the receiving side is unable to decode the received data correctly, in other words, if a CRC (Cyclic Redundancy Check) error occurs (CRC=NG), the receiving side will send a "Nack" to the transmitting side. The transmitting side, having received the "Nack", will retransmit the data. If the receiving side is able to decode the received data correctly, in other words, if no CRC error occurs (CRC=OK), the receiving side will send an "Ack" to the transmitting side. The transmitting side, having received the "Ack", will send the next data.
[0028] Another example of a retransmission method will be described. If a CRC error occurs on the receiving side, the receiving side requests a retransmission from the transmitting side. The retransmission request is made by toggling an NDI (New Data Indicator). The transmitting side, upon receiving the retransmission request, retransmits the data. If no CRC error occurs on the receiving side, no retransmission request is made. If the transmitting side does not receive a retransmission request for a predetermined period of time, it assumes that no CRC error occurred on the receiving side.
[0029] The logical channels described in Non-Patent Document 1 (Chapter 6) are explained below. The Broadcast Control Channel (BCCH) is a downlink channel for broadcasting system control information. The BCCH, which is a logical channel, is mapped to the broadcast channel (BCH) or the downlink shared channel (DL-SCH), which are transport channels.
[0030] The Paging Control Channel (PCCH) is a downlink channel for transmitting paging information and changes to system information. The PCCH, which is a logical channel, is mapped to the Paging Channel (PCH), which is a transport channel.
[0031] The Common Control Channel (CCCH) is a channel for transmitting control information between a communication terminal and a base station. The CCCH is used when the communication terminal does not have an RRC connection with the network. In the downlink direction, the CCCH is mapped to the Downlink Shared Channel (DL-SCH), which is a transport channel. In the uplink direction, the CCCH is mapped to the Uplink Shared Channel (UL-SCH), which is a transport channel.
[0032] A Dedicated Control Channel (DCCH) is a channel that transmits dedicated control information between a communication terminal and a network in a one-to-one relationship. The DCCH is used when the communication terminal has an RRC connection with the network. The DCCH is mapped to an uplink shared channel (UL-SCH) in the uplink and to a downlink shared channel (DL-SCH) in the downlink.
[0033] A Dedicated Traffic Channel (DTCH) is a point-to-point communication channel for transmitting user information to a communication terminal. DTCH exists in both uplink and downlink. In uplink, DTCH is mapped to an uplink shared channel (UL-SCH) and in downlink, it is mapped to a downlink shared channel (DL-SCH).
[0034] The location of a communication terminal is tracked in units of an area consisting of one or more cells. Location tracking is performed to track the location of the communication terminal even when it is in standby mode and to call the communication terminal, in other words, to enable the communication terminal to receive calls. The area for tracking the location of this communication terminal is called a Tracking Area (TA).
[0035] In NR, paging of communication terminals within a range that is smaller than a tracking area is supported. This range is called a RAN Notification Area (RNA). Paging of communication terminals in the RRC_INACTIVE state, which will be described later, is performed within this range.
[0036] In NR, in order to support wide frequency bandwidths (transmission bandwidths), carrier aggregation (CA) is being considered, which aggregates (also referred to as "aggregation") two or more component carriers (CCs). CA is described in Non-Patent Document 1.
[0037] When CA is configured, a communication terminal (UE) has only one RRC connection with the network (NW). In the RRC connection, one serving cell provides NAS (Non-Access Stratum) mobility information and security input. This cell is called a primary cell (PCell). Depending on the UE's capabilities, a secondary cell (SCell) is configured to form a serving cell set together with the PCell. A serving cell set consisting of one PCell and one or more SCells is configured for one UE.
[0038] In addition, in 3GPP, in order to further increase communication capacity, there is a dual connectivity (abbreviated as DC) in which a UE connects to two base stations to communicate. DC is described in Non-Patent Documents 1 and 22.
[0039] Of the base stations performing dual connectivity (DC), one may be referred to as a "master base station (Master Node: MN)" and the other as a "secondary base station (Secondary Node: SN)." Serving cells configured by the master base station may be collectively referred to as a master cell group (MCG), and serving cells configured by secondary base stations may be collectively referred to as a secondary cell group (SCG). In DC, the primary cell in the MCG or SCG is referred to as a special cell (SpCell or SPCell). The special cell in the MCG is referred to as a PCell, and the special cell in the SCG is referred to as a primary SCG cell (PSCell).
[0040] In addition, in NR, the base station pre-sets a portion of the carrier frequency band (hereinafter sometimes referred to as the Bandwidth Part (BWP)) for the UE, and the UE transmits and receives data to and from the base station using this BWP, thereby reducing power consumption in the UE.
[0041] Additionally, 3GPP is considering supporting services (or applications) using side link (SL) communication (also referred to as PC5 communication) in both the Evolved Packet System (EPS) (described later) and the 5G core system (see Non-Patent Documents 1, 2, 26 to 28). SL communication involves communication between terminals. Services using SL communication include, for example, vehicle-to-everything (V2X) services and proximity services. In SL communication, not only direct communication between terminals but also communication between a UE and a network via a relay has been proposed (see Non-Patent Documents 26 and 28).
[0042] The physical channels used for SL (see Non-Patent Documents 2 and 11) are as follows: The physical sidelink broadcast channel (PSBCH) carries information related to the system and synchronization and is transmitted from the UE.
[0043] The physical sidelink control channel (PSCCH) carries control information from the UE for sidelink and V2X sidelink communications.
[0044] The physical sidelink shared channel (PSSCH) carries data from the UE for sidelink and V2X sidelink communications.
[0045] The physical sidelink feedback channel (PSFCH) carries HARQ feedback on the sidelink from a UE that received a PSSCH transmission to the UE that transmitted the PSSCH.
[0046] The transport channel used for SL (see Non-Patent Document 1) will be described. The sidelink broadcast channel (SL-BCH) has a predetermined transport format and is mapped to the PSBCH, which is a physical channel.
[0047] The Sidelink Shared Channel (SL-SCH) supports broadcast transmissions. The SL-SCH supports both UE autonomous resource selection and base station scheduled resource allocation. UE autonomous resource selection involves a collision risk, whereas when the UE is allocated dedicated resources by the base station, there is no collision. The SL-SCH also supports dynamic link adaptation by changing transmit power, modulation, and coding. The SL-SCH is mapped to the PSSCH, which is a physical channel.
[0048] The logical channels used for SL (see Non-Patent Document 2) will be described. The Sidelink Broadcast Control Channel (SBCCH) is a sidelink channel for broadcasting sidelink system information from one UE to other UEs. The SBCCH is mapped to the SL-BCH, which is a transport channel.
[0049] The Sidelink Traffic Channel (STCH) is a point-to-multipoint traffic channel for transmitting user information from one UE to other UEs. The STCH is used only by UEs with sidelink communication capability and UEs with V2X sidelink communication capability. Point-to-point communication between two sidelink-capable UEs is also realized by the STCH. The STCH is mapped to the SL-SCH, a transport channel.
[0050] The Sidelink Control Channel (SCCH) is a control channel for transmitting control information from one UE to another UE. The SCCH is mapped to the SL-SCH, which is a transport channel.
[0051] In LTE, only broadcast was supported for SL communication. In NR, support for unicast and groupcast as SL communication in addition to broadcast is being considered (see Non-Patent Document 27 (3GPP TS23.287)).
[0052] In unicast communication and groupcast communication in SL, HARQ feedback (Ack / Nack), CSI reporting, etc. are supported.
[0053] In addition, 3GPP is considering integrated access and backhaul (IAB), which performs both the access link between a UE and a base station and the backhaul link between base stations wirelessly (see Non-Patent Documents 2, 20, and 29).
[0054] Several new technologies are required for mobile communication systems. For example, the introduction of artificial intelligence (AI) / machine learning (ML) into RAN is required to improve communication performance. Discussions on such new technologies have begun in 3GPP (Non-Patent Documents 30, 31).
[0055] 3GPP TS36.300 V17.3.03GPP TS38.300 V17.3.0“Scenarios, requirements and KPIs for 5G mobile and wireless system”、ICT-317669-METIS / D1.13GPP TR23.799 V14.0.03GPP TR38.801 V14.0.03GPP TR38.802 V14.2.03GPP TR38.804 V14.0.03GPP TR38.912 V16.0.03GPP RP-1721153GPP TS23.501 V18.0.03GPP TS38.211 V17.4.03GPP TS38.212 V17.4.03GPP TS38.213 V17.4.03GPP TS38.214 V17.4.03GPP TS38.321 V17.3.03GPP TS38.322 V17.2.03GPP TS38.323 V17.3.03GPP TS37.324 V17.0.03GPP TS38.331 V17.3.03GPP TS38.401 V17.3.03GPP TS38.413 V17.3.03GPP TS37.340 V17.3.03GPP TS38.423 V17.3.03GPP TS38.305 V17.3.03GPP TS23.273 V18.0.03GPP TR23.703 V12.0.03GPP TS23.287 V17.5.03GPP TS23.303 V17.0.03GPP TS38.340 V17.3.03GPP TR37.817 V17.0.03GPP RP-2136023GPP TS38.314 V17.2.0
[0056] Mobile communication systems require an increasing number of key performance indicators (KPIs), such as latency, reliability, connection density, user experience, and energy efficiency, and consistent optimization of these is required. For this reason, studies have begun on technologies to introduce AI / ML into mobile communication systems and improve various KPIs by collecting and analyzing data (Non-Patent Documents 30, 31). However, the use cases for which AI / ML is being considered are limited to network energy saving, load balancing, and mobility optimization, and other use cases have not yet been considered. This creates a problem in that various KPIs cannot be optimized in various communication situations.
[0057] In view of the above-mentioned problems, one of the objectives of the present disclosure is to realize a communication system that introduces AI / ML in various communication situations and can perform effective processing according to the communication situation.
[0058] The communication system according to the present disclosure includes a base station compatible with a fifth-generation wireless access system and a communication terminal connected to the base station, wherein the base station determines settings to be used in communication processing performed by the communication terminal using a model that has been trained using information obtained from communication terminals already connected to the base station and other adjacent base stations, and notifies the communication terminals connected to the base station of the determined settings. After receiving notification of the settings from a first base station, which is the base station that determined the settings, the communication terminal terminates its connection with the first base station and then uses the settings to perform communication processing for the first base station or communication processing for a second base station that is a base station different from the first base station.
[0059] According to the present disclosure, it is possible to realize a communication system capable of performing effective processing according to the communication situation.
[0060] The objects, features, aspects, and advantages of the present disclosure will become more apparent from the following detailed description and the accompanying drawings.
[0061] 1 is an explanatory diagram showing the configuration of a radio frame used in an NR communication system. FIG. 2 is a block diagram showing the overall configuration of an NR communication system 210 being discussed in 3GPP. FIG. 3 is a configuration diagram of DC by a base station connected to an NG core. FIG. 4 is a block diagram showing the configuration of a mobile terminal 202 shown in FIG. 2. FIG. 5 is a block diagram showing the configuration of a base station 213 shown in FIG. 2. FIG. 6 is a block diagram showing the configuration of a 5GC unit. FIG. 7 is a flowchart showing an overview of processes from cell search to standby operation performed by a communication terminal (UE) in an NR communication system. FIG. 8 is a diagram showing an example of a cell configuration in an NR system. FIG. 9 is a connection configuration diagram showing an example of a connection configuration of a terminal in SL communication. FIG. 10 is a connection configuration diagram showing an example of a connection configuration of a base station that supports access / backhaul integration. FIG. 11 is a diagram showing an example of a configuration of a learning device related to a communication system for embodiment 1. FIG. 12 is a flowchart related to learning processing of the learning device for embodiment 1. FIG. 13 is a diagram showing an example of a configuration of an inference device related to a communication system for embodiment 1. FIG. 14 is a flowchart related to inference processing of the inference device for embodiment 1. FIG. 15 is a diagram showing an example of a sequence of RA processing using AI / ML for RA processing for embodiment 1. FIG. 16 is a diagram showing an example of a sequence of RA processing using AI / ML for RA processing for embodiment 1 variant 1. Fig. 10 is a diagram showing an example of a sequence of SDT processing using an AI / ML for SDT processing, according to embodiment 2. Fig. 11 is a diagram showing an example of a sequence of IAB processing using an AI / ML for IAB processing, according to embodiment 4. Fig. 12 is a diagram showing the first half of an example of a sequence of BH RLF processing using an AI / ML for BH RLF processing, according to modified example 1 of embodiment 4. Fig. 13 is a diagram showing the second half of an example of a sequence of BH RLF processing using an AI / ML for BH RLF processing, according to modified example 1 of embodiment 4.
[0062] Embodiment 1. Figure 2 is a block diagram showing the overall configuration of an NR communication system 210 being discussed in 3GPP. Figure 2 will be explained. The radio access network is called an NG-RAN (Next Generation Radio Access Network) 211. A mobile terminal device (hereinafter referred to as a "mobile terminal (User Equipment: UE)") 202, which is a communication terminal device, is capable of wireless communication with a base station device (hereinafter referred to as an "NR base station (NG-RAN NodeB: gNB)") 213, and transmits and receives signals via wireless communication. The NG-RAN 211 is composed of one or more NR base stations 213.
[0063] Here, the term "communication terminal device" includes not only mobile terminal devices such as mobile cell phone terminal devices, but also stationary devices such as sensors. In the following description, the term "communication terminal device" may be simply referred to as a "communication terminal."
[0064] An access stratum (AS) protocol is terminated between the UE 202 and the NG-RAN 211. Examples of AS protocols include radio resource control (RRC), service data adaptation protocol (SDAP), packet data convergence protocol (PDCP), radio link control (RLC), medium access control (MAC), and physical layer (PHY). RRC is used in the control plane (hereinafter sometimes referred to as the C-plane, C-Plane, or CP), SDAP is used in the user plane (hereinafter sometimes referred to as the U-plane, U-Plane, or UP), and PDCP, MAC, RLC, and PHY are used in both the C-plane and the U-plane.
[0065] The control protocol RRC (Radio Resource Control) between the UE 202 and the NR base station 213 performs broadcasting, paging, RRC connection management, etc. The states of the NR base station 213 and the UE 202 in RRC include RRC_IDLE, RRC_CONNECTED, and RRC_INACTIVE.
[0066] In RRC_IDLE, PLMN (Public Land Mobile Network) selection, system information (SI) broadcast, paging, cell re-selection, mobility, etc. are performed. In RRC_CONNECTED, the mobile terminal has an RRC connection and can transmit and receive data with the network. In addition, in RRC_CONNECTED, handover (HO), measurement of neighbor cells, etc. are performed. In RRC_INACTIVE, the connection between the 5G core unit 214 and the NR base station 213 is maintained, and system information (SI) broadcast, paging, cell re-selection, mobility, etc. are performed.
[0067] The gNB 213 is connected to a 5G core unit (hereinafter sometimes referred to as the "5GC unit") 214, which includes an Access and Mobility Management Function (AMF), a Session Management Function (SMF), or a User Plane Function (UPF), via an NG interface. Control information and / or user data is communicated between the gNB 213 and the 5GC unit 214. The NG interface is a collective term for the N2 interface between the gNB 213 and the AMF 220, the N3 interface between the gNB 213 and the UPF 221, the N11 interface between the AMF 220 and the SMF 222, and the N4 interface between the UPF 221 and the SMF 222. Multiple 5GC units 214 may be connected to one gNB 213. The gNBs 213 are connected via an Xn interface, and control information and / or user data are communicated between the gNBs 213.
[0068] The 5GC unit 214 is a higher-level device, specifically a higher-level node, and controls the connection between the NR base station 213 and the mobile terminal (UE) 202, distributes paging signals to one or more NR base stations (gNB) 213 and / or LTE base stations (E-UTRAN NodeB: eNB), and performs other functions. The 5GC unit 214 also performs mobility control in the idle state. The 5GC unit 214 manages the tracking area list when the mobile terminal 202 is in the idle state, in the inactive state, and in the active state. The 5GC unit 214 initiates a paging protocol by transmitting a paging message to a cell belonging to the tracking area in which the mobile terminal 202 is registered.
[0069] The gNB 213 may configure one or more cells. When one gNB 213 configures multiple cells, each of the cells is configured to be able to communicate with the UE 202.
[0070] The gNB 213 may be divided into a central unit (hereinafter, sometimes referred to as CU) 215 and a distributed unit (hereinafter, sometimes referred to as DU) 216. One CU 215 is configured within the gNB 213. One or more DUs 216 are configured within the gNB 213. One DU 216 configures one or more cells. The CU 215 is connected to the DU 216 via an F1 interface, and control information and / or user data is communicated between the CU 215 and the DU 216. The F1 interface consists of an F1-C interface and an F1-U interface. The CU 215 is responsible for the functions of the RRC, SDAP, and PDCP protocols, and the DU 216 is responsible for the functions of the RLC, MAC, and PHY protocols. One or more TRPs (Transmission Reception Points) 219 may be connected to the DU 216. The TRP 219 transmits and receives radio signals to and from the UE.
[0071] The CU 215 may be divided into a C-plane CU (CU-C) 217 and a U-plane CU (CU-U) 218. One CU-C 217 is configured within the CU 215. One or more CU-Us 218 are configured within the CU 215. The CU-C 217 is connected to the CU-U 218 via an E1 interface, and control information is communicated between the CU-C 217 and the CU-U 218. The CU-C 217 is connected to the DU 216 via an F1-C interface, and control information is communicated between the CU-C 217 and the DU 216. The CU-U 218 is connected to the DU 216 via an F1-U interface, and user data is communicated between the CU-U 218 and the DU 216.
[0072] In a 5G communication system, a Unified Data Management (UDM) function and a Policy Control Function (PCF) described in Non-Patent Document 10 (3GPP TS23.501) may be included. The UDM and / or PCF may be included in the 5GC unit 214 in FIG. 2 .
[0073] In a 5G communication system, a Location Management Function (LMF) described in Non-Patent Document 24 (3GPP TS 38.305) may be provided. The LMF may be connected to a base station via an AMF as disclosed in Non-Patent Document 25 (3GPP TS 23.273).
[0074] A 5G communication system may include a Non-3GPP Interworking Function (N3IWF) described in Non-Patent Document 10 (3GPP TS23.501). The N3IWF may terminate an Access Network (AN) between the UE and the N3IWF in non-3GPP access between the UE and the N3IWF.
[0075] FIG. 3 is a diagram showing a DC (dual connectivity) configuration connected to an NG core. In FIG. 3, solid lines indicate U-Plane connections, and dashed lines indicate C-Plane connections. In FIG. 3, the master base station 240-1 may be a gNB or an eNB. Furthermore, the secondary base station 240-2 may be a gNB or an eNB. For example, in FIG. 3, a DC configuration in which the master base station 240-1 is a gNB and the secondary base station 240-2 is an eNB may be referred to as NG-EN-DC. In FIG. 3, an example is shown in which the U-Plane connection between the 5GC unit 214 and the secondary base station 240-2 is performed via the master base station 240-1, but it may also be performed directly between the 5GC unit 214 and the secondary base station 240-2. 3, an EPC (Evolved Packet Core), which is a core network connected to the LTE system and the LTE-A system, may be connected to the master base station 240-1 instead of the 5GC unit 214. A U-Plane connection may be directly established between the EPC and the secondary base station 240-2.
[0076] FIG. 4 is a block diagram showing the configuration of mobile terminal 202 shown in FIG. 2. The transmission process of mobile terminal 202 shown in FIG. 4 will be described. First, control data from control unit 310 and user data from application unit 302 are sent to protocol processing unit 301. Buffering of the control data and user data may be performed. Buffers for the control data and user data may be provided in control unit 310, application unit 302, or protocol processing unit 301. Protocol processing unit 301 performs protocol processing such as SDAP, PDCP, RLC, and MAC, for example, determining a destination base station in DC, and adding a header for each protocol. The protocol-processed data is passed to encoder unit 304, where it is subjected to encoding such as error correction. Some data may be output directly from protocol processing unit 301 to modulation unit 305 without being encoded. The data encoded by encoder unit 304 is modulated by modulation unit 305. Precoding in MIMO may be performed by modulation unit 305. The modulated data is converted into a baseband signal, and then output to frequency conversion section 306, where it is converted into a radio transmission frequency. Then, the transmission signal is transmitted from antennas 307-1 to 307-4 to base station 213. Although the example in FIG. 4 shows a case where the number of antennas is four, the number of antennas is not limited to four.
[0077] Furthermore, the reception process of the mobile terminal 202 is performed as follows. Radio signals from the base station 213 are received by the antennas 307-1 to 307-4. The received signals are converted from a radio reception frequency to a baseband signal by the frequency conversion unit 306, and demodulated by the demodulation unit 308. The demodulation unit 308 may also perform weight calculation and multiplication processing. The demodulated data is passed to the decoder unit 309, where decoding processes such as error correction are performed. The decoded data is passed to the protocol processing unit 301, where protocol processing such as MAC, RLC, PDCP, and SDAP is performed, for example, operations such as removing headers in each protocol. Of the data that has undergone protocol processing, the control data is passed to the control unit 310, and the user data is passed to the application unit 302.
[0078] A series of processes in the mobile terminal 202 is controlled by a control unit 310. Therefore, the control unit 310 is also connected to each of the units 302, 304 to 309, although this is omitted in FIG.
[0079] Each unit of the mobile terminal 202, such as the control unit 310, protocol processing unit 301, encoder unit 304, and decoder unit 309, is implemented by a processing circuit including, for example, a processor and memory. For example, the control unit 310 is implemented by a processor executing a program describing a series of processes performed by the mobile terminal 202. The program describing the series of processes performed by the mobile terminal 202 is stored in memory. Examples of memory include non-volatile or volatile semiconductor memory such as RAM (Random Access Memory), ROM (Read Only Memory), and flash memory. Each unit of the mobile terminal 202, such as the control unit 310, protocol processing unit 301, encoder unit 304, and decoder unit 309, may be implemented by a dedicated processing circuit such as an FPGA (Field Programmable Gate Array), an ASIC (Application Specific Integrated Circuit), or a DSP (Digital Signal Processor). In FIG. 4, the number of antennas used by the mobile terminal 202 for transmission and the number of antennas used for reception may be the same or different.
[0080] 5 is a block diagram showing the configuration of the base station 213 shown in FIG. 2. The transmission process of the base station 213 shown in FIG. 5 will be described. The EPC communication unit 401 transmits and receives data between the base station 213 and the EPC. The 5GC communication unit 412 transmits and receives data between the base station 213 and the 5GC (such as the 5GC unit 214). The other base station communication unit 402 transmits and receives data with other base stations. The EPC communication unit 401, the 5GC communication unit 412, and the other base station communication unit 402 each exchange information with the protocol processing unit 403. Control data from the control unit 411, and user data and control data from the EPC communication unit 401, the 5GC communication unit 412, and the other base station communication unit 402 are sent to the protocol processing unit 403. Buffering of the control data and user data may be performed. Buffers for control data and user data may be provided in the control unit 411, the EPC communication unit 401, the 5GC communication unit 412, or the other base station communication unit 402.
[0081] The protocol processing unit 403 performs protocol processing such as SDAP, PDCP, RLC, MAC, etc., such as routing transmission data in DC, etc., and adding headers for each protocol. The protocol-processed data is passed to the encoder unit 405, where it is subjected to encoding processing such as error correction. Some data may be output directly from the protocol processing unit 403 to the modulation unit 406 without being encoded. Data may also be sent from the protocol processing unit 403 to the other base station communication unit 402. For example, in DC, data sent from the 5GC communication unit 412 or the EPC communication unit 401 may be sent to another base station, such as a secondary base station, via the other base station communication unit 402. The encoded data is modulated by the modulation unit 406. The modulation unit 406 may also perform MIMO precoding. The modulated data is converted into a baseband signal, and then output to the frequency conversion unit 407, where it is converted into a radio transmission frequency. Thereafter, the transmission signals are transmitted from antennas 408-1 to 408-4 to one or more mobile terminals 202. Although the number of antennas is four in the example shown in Fig. 5, the number of antennas is not limited to four.
[0082] Furthermore, the reception process of the base station 213 is performed as follows. Radio signals from one or more mobile terminals 202 are received by antennas 408-1 to 408-4. The received signals are converted from a radio reception frequency to a baseband signal by a frequency conversion unit 407, and demodulated by a demodulation unit 409. The demodulated data is passed to a decoder unit 410, where decoding processes such as error correction are performed. The decoded data is passed to a protocol processing unit 403, where protocol processes such as MAC, RLC, PDCP, and SDAP are performed, for example, operations such as removing headers in each protocol. Of the data that has undergone protocol processing, control data is passed to the control unit 411, 5GC communication unit 412, EPC communication unit 401, or other base station communication unit 402, and user data is passed to the 5GC communication unit 412, EPC communication unit 401, or other base station communication unit 402. Data sent from the other base station communication unit 402 may be sent to the 5GC communication unit 412 or the EPC communication unit 401. The data may be, for example, uplink data sent to the 5GC communication unit 412 or the EPC communication unit 401 via another base station in DC.
[0083] A series of processes in the base station 213 is controlled by a control unit 411. Therefore, the control unit 411 is also connected to each of the units 401, 402, 405 to 410, and 412, although this is omitted in FIG.
[0084] Each unit of the base station 213, for example, the control unit 411, the protocol processing unit 403, the 5GC communication unit 412, the EPC communication unit 401, the other base station communication unit 402, the encoder unit 405, and the decoder unit 410, is realized by a processing circuit including a processor and a memory, or a dedicated processing circuit such as an FPGA, an ASIC, or a DSP, similar to the above-mentioned mobile terminal 202. In Fig. 5, the number of antennas used by the base station 213 for transmission and the number of antennas used for reception may be the same or different.
[0085] As an example of the configuration of the CU 215 shown in Fig. 2, a configuration in which a DU communication unit is provided is sometimes used, excluding the encoder unit 405, modulation unit 406, frequency conversion unit 407, antennas 408-1 to 408-4, demodulation unit 409, and decoder unit 410 shown in Fig. 5. The DU communication unit is connected to a protocol processing unit 403. The protocol processing unit 403 in the CU 215 performs protocol processing such as PDCP and SDAP.
[0086] As an example of the configuration of the DU 216 shown in Fig. 2, a configuration in which a CU communication unit is provided may be used, excluding the EPC communication unit 401, other base station communication unit 402, and 5GC communication unit 412 shown in Fig. 5. The CU communication unit is connected to a protocol processing unit 403. The protocol processing unit 403 in the DU 216 performs protocol processing such as PHY, MAC, and RLC.
[0087] FIG. 6 is a block diagram showing the configuration of the 5GC unit. FIG. 6 shows the configuration of the 5GC unit 214 shown in FIG. 2 described above. FIG. 6 shows a case where the 5GC unit 214 shown in FIG. 2 includes an AMF configuration, an SMF configuration, and a UPF configuration. In the example shown in FIG. 6, the AMF may have the function of the control plane control unit 525, the SMF may have the function of the session management unit 527, and the UPF may have the functions of the user plane communication unit 523 and the Data Network communication unit 521. The Data Network communication unit 521 transmits and receives data between the 5GC unit 214 and the Data Network. The base station communication unit 522 transmits and receives data via the NG interface between the 5GC unit 214 and the base station 213. User data sent from the Data Network is passed from the Data Network communication unit 521 to the base station communication unit 522 via the user plane communication unit 523, and then transmitted to one or more base stations 213. User data sent from the base station 213 is passed from the base station communication unit 522 to the Data Network communication unit 521 via the user plane communication unit 523, and then transmitted to the Data Network.
[0088] The control data sent from the base station 213 is passed from the base station communication unit 522 to the control plane control unit 525. The control plane control unit 525 may pass the control data to the session management unit 527. The control data may be sent from the Data Network. The control data sent from the Data Network may be sent from the Data Network communication unit 521 to the session management unit 527 via the user plane communication unit 523. The session management unit 527 may send the control data to the control plane control unit 525.
[0089] The user plane communication unit 523 includes a PDU processing unit 523-1, a mobility anchoring unit 523-2, etc., and performs general processing for the user plane (hereinafter sometimes referred to as U-Plane). The PDU processing unit 523-1 processes data packets, for example, transmitting and receiving packets to and from the Data Network communication unit 521, and transmitting and receiving packets to and from the base station communication unit 522. The mobility anchoring unit 523-2 is responsible for anchoring the data path during UE mobility.
[0090] The session management unit 527 manages the PDU session established between the UE and the UPF. The session management unit 527 includes a PDU session control unit 527-1, a UE IP address allocation unit 527-2, etc. The PDU session control unit 527-1 manages the PDU session between the mobile terminal 202 and the 5GC unit 214. The UE IP address allocation unit 527-2 assigns an IP address to the mobile terminal 202, etc.
[0091] The control plane control unit 525 includes a NAS security unit 525-1, an idle state mobility management unit 525-2, etc., and performs overall processing for the control plane (hereinafter sometimes referred to as the C-Plane). The NAS security unit 525-1 performs security for NAS (Non-Access Stratum) messages, etc. The idle state mobility management unit 525-2 performs mobility management in the standby state (idle state: also referred to as RRC_IDLE state or simply idle), generation and control of paging signals in the standby state, addition, deletion, update, search, tracking area list management, etc. for one or more mobile terminals 202 under its control.
[0092] A series of processes of the 5GC unit 214 is controlled by a control unit 526. Therefore, although the control unit 526 is omitted in Fig. 6, it is connected to each unit 521 to 523, 525, and 527. Like the control unit 310 of the mobile terminal 202 described above, each unit of the 5GC unit 214 is realized by, for example, a processing circuit configured to include a processor and memory, or a dedicated processing circuit such as an FPGA, ASIC, or DSP.
[0093] Next, an example of a cell search method in a communication system is shown. Fig. 7 is a flowchart showing an outline of the process from cell search to standby operation performed by a communication terminal (UE) in an NR communication system. When the communication terminal starts a cell search, in step ST601, it synchronizes slot timing and frame timing using a primary synchronization signal (P-SS) and a secondary synchronization signal (S-SS) transmitted from a surrounding base station.
[0094] P-SS and S-SS are collectively called a synchronization signal (SS). A synchronization code is assigned to the synchronization signal (SS) in one-to-one correspondence with a PCI (Physical Cell Identifier) assigned to each cell. 1008 different PCIs are being considered. A communication terminal synchronizes using these 1008 different PCIs and detects (identifies) the PCI of the synchronized cell.
[0095] In step ST602, the communication terminal receives the PBCH for the next synchronized cell. A master information block (MIB) including cell configuration information is mapped to the BCCH on the PBCH. Therefore, the MIB can be obtained by receiving the PBCH and obtaining the BCCH. Examples of MIB information include a system frame number (SFN), scheduling information for system information block (SIB) 1, subcarrier spacing for SIB1 and the like, and information on the DM-RS position.
[0096] Furthermore, the communication terminal acquires an SS block identifier from the PBCH. A part of the bit string of the SS block identifier is included in the MIB. The remaining bit string is included in an identifier used to generate a sequence of DM-RS associated with the PBCH. The communication terminal acquires the SS block identifier using the MIB included in the PBCH and the sequence of DM-RS associated with the PBCH.
[0097] Next, in step ST603, the communication terminal measures the received power of the SS block.
[0098] Next, in step ST604, the communication terminal selects the cell with the best reception quality, for example, the cell with the highest reception power, that is, the best cell, from among the one or more cells detected up to step ST603. The communication terminal also selects the beam with the best reception quality, for example, the beam with the highest reception power of the SS block, that is, the best beam. The reception power of the SS block for each SS block identifier is used, for example, to select the best beam.
[0099] Next, in step ST605, the communication terminal receives DL-SCH based on the scheduling information of SIB1 included in the MIB, and obtains SIB (System Information Block) 1 in the broadcast information BCCH. SIB1 includes information on access to the cell, cell configuration information, and scheduling information of other SIBs (SIBk: k is an integer greater than or equal to 2). SIB1 also includes a tracking area code (TAC).
[0100] Next, in step ST606, the communication terminal compares the TAC of SIB1 received in step ST605 with the TAC portion of the tracking area identity (TAI) in the tracking area list already held by the communication terminal. The tracking area list is also called a TAI list. The TAI is identification information for identifying a tracking area, and is composed of an MCC (Mobile Country Code), an MNC (Mobile Network Code), and a TAC (Tracking Area Code). The MCC is a country code. The MNC is a network code. The TAC is a tracking area code number.
[0101] If the comparison in step ST606 shows that the TAC received in step ST605 is the same as the TAC included in the tracking area list, the communication terminal enters standby mode in the cell. If the comparison shows that the TAC received in step ST605 is not included in the tracking area list, the communication terminal requests a core network (EPC) including an MME and the like to change the tracking area through the cell in order to perform a Tracking Area Update (TAU).
[0102] An apparatus constituting a core network (hereinafter sometimes referred to as a "core network side apparatus") updates the tracking area list based on the identification number (UE-ID, etc.) of the communication terminal sent from the communication terminal together with a TAU request signal. The core network side apparatus transmits the updated tracking area list to the communication terminal. The communication terminal rewrites (updates) the TAC list held by the communication terminal based on the received tracking area list. Thereafter, the communication terminal enters standby mode in the cell.
[0103] Next, examples of random access methods in a communication system are shown. Four-step random access and two-step random access are used for random access. For each of the four-step and two-step random access methods, there is contention-based random access, i.e., random access in which timing collisions with other mobile terminals may occur, and contention-free random access.
[0104] An example of a collision-based four-step random access method is shown below. In the first step, the mobile terminal transmits a random access preamble to the base station. The random access preamble may be selected by the mobile terminal from a predetermined range, or may be individually assigned to the mobile terminal and notified by the base station.
[0105] In the second step, the base station transmits a random access response to the mobile terminal, which includes uplink scheduling information used in the third step, a terminal identifier used in the uplink transmission in the third step, and the like.
[0106] In the third step, the mobile terminal performs uplink transmission to the base station. The mobile terminal uses the information acquired in the second step for uplink transmission. In the fourth step, the base station notifies the mobile terminal whether or not the collision has been resolved. If the mobile terminal is notified that there is no collision, it ends the random access process. If the mobile terminal is notified that there is a collision, it starts the process over from the first step.
[0107] The contention-free four-step random access method differs from the contention-based four-step random access method in the following points: Prior to the first step, the base station pre-assigns a random access preamble and uplink scheduling to the mobile terminal, and the notification of whether or not contention has been resolved in the fourth step is not required.
[0108] An example of a collision-based two-step random access method is shown below. In the first step, the mobile terminal transmits a random access preamble and performs uplink transmission to the base station. In the second step, the base station notifies the mobile terminal whether there is a collision. If the mobile terminal is notified that there is no collision, it terminates the random access process. If the mobile terminal is notified that there is a collision, it restarts the process from the first step.
[0109] The contention-free two-step random access method differs from the contention-based two-step random access method in the following points: prior to the first step, the base station pre-assigns a random access preamble and uplink scheduling to the mobile terminal, and in the second step, the base station transmits a random access response to the mobile terminal.
[0110] FIG. 8 shows an example of a cell configuration in NR. In an NR cell, narrow beams are formed and transmitted in different directions. In the example shown in FIG. 8, at certain times, base station 750 transmits and receives signals to and from a mobile terminal using beam 751-1. At other times, base station 750 transmits and receives signals to and from a mobile terminal using beam 751-2. In a similar manner, base station 750 transmits and receives signals to and from a mobile terminal using one or more of beams 751-3 to 751-8. In this way, base station 750 forms a wide-area cell 752.
[0111] 8 shows an example in which the number of beams used by the base station 750 is 8, but the number of beams may be different from 8. Also, in the example shown in FIG. 8, the number of beams used simultaneously by the base station 750 is 1, but it may be multiple.
[0112] The concept of Quasi-CoLocation (QCL) is used to identify beams (see Non-Patent Document 14 (3GPP TS38.214)). That is, the beam is identified by information indicating which reference signal (e.g., SS block, CSI-RS) the beam can be considered to be the same as. The information may include types of information regarding the aspects of the beams that can be considered to be the same, such as Doppler shift, Doppler shift spread, mean delay, mean delay spread, and spatial Rx parameters (see Non-Patent Document 14 (3GPP TS38.214)).
[0113] In 3GPP, a side link (SL) is supported for D2D (Device to Device) communication and V2V (Vehicle to Vehicle) communication (see Non-Patent Document 1 and Non-Patent Document 16). The SL is defined by the PC5 interface.
[0114] In order to support unicast and groupcast in addition to broadcast in SL communication, support for PC5-S signaling is being considered (see Non-Patent Document 27 (3GPP TS23.287)). For example, PC5-S signaling is implemented to establish a link for implementing SL, i.e., PC5 communication. The link is implemented in the V2X layer and is also called a Layer 2 link.
[0115] Furthermore, support for RRC signaling in SL communication is being considered (see Non-Patent Document 27 (3GPP TS23.287)). RRC signaling in SL communication is also referred to as PC5 RRC signaling. For example, it has been proposed to notify UE capabilities between UEs performing PC5 communication, or to notify AS layer settings for performing V2X communication using PC5 communication.
[0116] An example of a connection configuration of mobile terminals in SL communication is shown in Fig. 9. In the example shown in Fig. 9, UE 805 and UE 806 exist within the coverage 803 of base station 801. UL / DL communication 807 is performed between base station 801 and UE 805. UL / DL communication 808 is performed between base station 801 and UE 806. SL communication 810 is performed between UE 805 and UE 806. UE 811 and UE 812 exist outside the coverage 803. SL communication 814 is performed between UE 805 and UE 811. In addition, SL communication 816 is performed between UE 811 and UE 812.
[0117] As an example of communication between a UE and a NW via a relay in SL communication, a UE 805 shown in FIG. 9 relays communication between a UE 811 and a base station 801.
[0118] A configuration similar to that shown in FIG. 4 may be used for a UE that performs relaying. The relaying process in the UE will be described using FIG. 4. The relaying process by UE 805 in communication from UE 811 to base station 801 will be described. A radio signal from UE 811 is received by antennas 307-1 to 307-4. The received signal is converted from a radio reception frequency to a baseband signal by frequency conversion unit 306, and demodulated by demodulation unit 308. Weight calculation and multiplication processing may also be performed by demodulation unit 308. The demodulated data is passed to decoder unit 309, where decoding processing such as error correction is performed. The decoded data is passed to protocol processing unit 301, where protocol processing such as MAC and RLC used for communication with UE 811 is performed, such as removing headers in each protocol. Protocol processing such as RLC and MAC used for communication with base station 801 is also performed, such as adding headers in each protocol. Protocol processing of PDCP and SDAP may be performed in protocol processing unit 301 of UE 811. The protocol-processed data is passed to encoder unit 304, where encoding such as error correction is performed. Some data may be output directly from protocol processing unit 301 to modulation unit 305 without being encoded. The data encoded by encoder unit 304 is modulated by modulation unit 305. Precoding in MIMO may be performed by modulation unit 305. The modulated data is converted into a baseband signal, and then output to frequency conversion unit 306, where it is converted into a radio transmission frequency. Thereafter, a transmission signal is transmitted to base station 801 from antennas 307-1 to 307-4.
[0119] In the above, an example of relaying by UE 805 in communication from UE 811 to base station 801 has been shown, but similar processing is also used in relaying communication from base station 801 to UE 811.
[0120] 5G base stations can support integrated access and backhaul (IAB) (see Non-Patent Documents 2 and 20). A base station supporting IAB (hereinafter sometimes referred to as an IAB base station) is composed of an IAB donor CU, which is a CU of the base station operating as an IAB donor that provides IAB functions, an IAB donor DU, which is a DU of the base station operating as an IAB donor, and an IAB node connected to the IAB donor DU and to a UE via a radio interface. An F1 interface is provided between the IAB node and the IAB donor CU (see Non-Patent Document 2).
[0121] An example of IAB base station connections is shown in Figure 10. IAB donor CU 901 is connected to IAB donor DU 902. IAB node 903 is connected to IAB donor DU 902 using a wireless interface. IAB node 903 is connected to IAB node 904 using a wireless interface. In other words, IAB nodes may be connected in cascade. UE 905 is connected to IAB node 904 using a wireless interface. UE 906 may be connected to IAB node 903 using a wireless interface, and UE 907 may be connected to IAB donor DU 902 using a wireless interface. Multiple IAB donor DUs 902 may be connected to an IAB donor CU 901, multiple IAB nodes 903 may be connected to an IAB donor DU 902, and multiple IAB nodes 904 may be connected to an IAB node 903.
[0122] A BAP (Backhaul Adaptation Protocol) layer is provided in the connection between the IAB donor DU and the IAB node and in the connection between the IAB nodes (see Non-Patent Document 29). The BAP layer performs operations such as routing received data to the IAB donor DU and / or the IAB node, and mapping to an RLC channel (see Non-Patent Document 29).
[0123] As an example of the configuration of the IAB donor CU, a configuration similar to that of CU 215 is used.
[0124] An example of the configuration of the IAB donor DU is the same as that of the DU 216. The protocol processing unit of the IAB donor DU performs BAP layer processing, such as adding a BAP header to downstream data, routing to an IAB node, and removing the BAP header from upstream data.
[0125] As an example of the configuration of an IAB node, a configuration excluding the EPC communication unit 401, other base station communication unit 402, and 5GC communication unit 412 shown in Figure 5 may be used.
[0126] The transmission and reception processing at the IAB node will be described using FIGS. 5 and 10 . The transmission and reception processing at the IAB node 903 in communication between the IAB donor CU 901 and the UE 905 will be described. In uplink communication from the UE 905 to the IAB donor CU 901, a radio signal from the IAB node 904 is received by the antenna 408 (some or all of the antennas 408-1 to 408-4). The received signal is converted from a radio reception frequency to a baseband signal by the frequency conversion unit 407, and demodulated by the demodulation unit 409. The demodulated data is passed to the decoder unit 410, where decoding processing such as error correction is performed. The decoded data is passed to the protocol processing unit 403, where protocol processing such as MAC and RLC used for communication with the IAB node 904 is performed, such as removing headers in each protocol. In addition, routing to the IAB donor DU 902 is performed using a BAP header, and protocol processing such as RLC and MAC used for communication with the IAB donor DU 902, such as adding headers for each protocol, is performed. The protocol-processed data is passed to the encoder unit 405, where it is subjected to encoding processing such as error correction. Some data may be output directly from the protocol processing unit 403 to the modulation unit 406 without being encoded. The encoded data is modulated by the modulation unit 406. The modulation unit 406 may also perform precoding in MIMO. The modulated data is converted into a baseband signal, and then output to the frequency conversion unit 407, where it is converted into a radio transmission frequency. Then, a transmission signal is transmitted to the IAB donor DU 902 from antennas 408-1 to 408-4. Similar processing is performed in downlink communication from the IAB donor CU 901 to the UE 905.
[0127] The IAB node 904 also performs the same transmission and reception processing as the IAB node 903. The protocol processing unit 403 of the IAB node 903 performs BAP layer processing, such as adding a BAP header and routing to the IAB node 904 in upstream communication, and removing the BAP header in downstream communication.
[0128] In the 3GPP mobile communication system standard, a UE must perform a random access (RA) process to connect to a cell, for example, during cell selection or cell reselection. In the RA process, the UE transmits a PRACH to the cell. If the UE cannot receive a PRACH response from the base station, it increases the transmission power by a predetermined amount and retransmits the PRACH. While significantly increasing the transmission power enables connection with fewer transmissions, the transmission power also causes interference with other UL communications. Thus, unless an appropriate PRACH transmission power is derived, it is impossible to improve the various KPIs required for a mobile communication system.
[0129] In this embodiment, a method is disclosed that enables more effective RA processing according to the communication situation.
[0130] In this embodiment, AI / ML is introduced into RA processing. RA processing includes, for example, PRACH transmission processing, RA problem (Random Access Problem) processing, and backoff processing (see Non-Patent Document 15). By using AI / ML in RA processing, for example, information used in PRACH transmission processing, RA problem processing, backoff processing, and the like is optimized.
[0131] Learning of a model (sometimes referred to as training in this specification) will now be described. In this specification, a model (including a trained model) may be referred to as an AI / ML model.
[0132] The communication system according to this embodiment employs a learning device 1100 having the configuration shown in FIG. 11 . FIG. 11 is a diagram illustrating an example of the configuration of the learning device 1100 related to the communication system. The learning device 1100 includes a data acquisition unit 1110 and a model generation unit 1120. The trained model generated by the model generation unit 1120 is stored in a trained model storage unit 1130. While FIG. 11 illustrates an example of a configuration in which the trained model storage unit 1130 exists outside the learning device 1100, the learning device 1100 may also be configured to include the trained model storage unit 1130. The learning device 1100 shown in FIG. 11 may be applied to a base station.
[0133] The data acquisition unit 1110 acquires input information and feedback information as learning data.
[0134] The model generation unit 1120 learns output information based on learning data including input information and feedback information, i.e., generates a learned model that infers output information from input information of a communication system.
[0135] The learning algorithm used by the model generation unit 1120 may be a known algorithm such as supervised learning, unsupervised learning, or reinforcement learning. As an example, a case where reinforcement learning is applied will be described. In reinforcement learning, an agent (acting subject) in a certain environment observes the current state (environmental parameters) and decides on an action to take. The environment changes dynamically depending on the agent's actions, and the agent is given a reward according to the environmental changes. The agent repeats this process and learns the course of action that will obtain the most reward through a series of actions. Q-learning and TD-learning are known as representative reinforcement learning methods. For example, in the case of Q-learning, the general update formula for the action value function Q(s, a) is expressed as Equation (1).
[0136]
[0137] In formula (1), s t represents the state of the environment at time t, and a t represents the action at time t. t Therefore, the state is s t+1 It changes to r t+1 represents the reward that can be obtained depending on the change in state, γ represents the discount rate, and α represents the learning coefficient. Note that γ is in the range of 0<γ≦1, and α is in the range of 0<α≦1. Using the input information, the action a t , state s t and the state s at time t t Best action in a t Learn.
[0138] The update formula expressed by equation (1) increases the action value Q if the action value Q of the action a with the highest Q value at time t+1 is greater than the action value Q of the action a executed at time t, and decreases the action value Q in the opposite case. In other words, the action value function Q(s, a) is updated so that the action value Q of the action a at time t approaches the best action value at time t+1. As a result, the best action value in a certain environment is propagated sequentially to the action values in previous environments.
[0139] As described above, when generating a trained model by reinforcement learning, the model generation unit 1120 includes a reward calculation unit 1121 and a function update unit 1122.
[0140] The reward calculation unit 1121 calculates a reward based on the input information and feedback information. The reward calculation unit 1121 calculates a reward r based on a reward criterion. For example, in the case of a reward increase criterion, the reward r is increased (for example, a reward of "1"), while in the case of a reward decrease criterion, the reward r is decreased (for example, a reward of "-1" is given).
[0141] The function update unit 1122 updates the function for determining the output in accordance with the reward calculated by the reward calculation unit 1121, and outputs the updated function to the trained model storage unit 1130. For example, in the case of Q-learning, the action value function Q(s t , a t ) is used as a function to calculate the output information.
[0142] The learning device 1100 repeatedly performs the above-described learning. The learned model storage unit 1130 stores the action-value function Q(s t , a t ), i.e., stores the trained model.
[0143] Next, the learning process of the learning device 1100 will be described with reference to Fig. 12. Fig. 12 is a flowchart showing the learning process of the learning device.
[0144] In step b1, the data acquisition unit 1110 acquires input information and feedback information as learning data.
[0145] In step b2, the model generation unit 1120 calculates a reward based on the input information and feedback information. Specifically, the reward calculation unit 1121 acquires the input information and feedback information and determines whether to increase the reward (step b3) or decrease the reward (step b4) based on a predetermined reward standard.
[0146] If the remuneration calculation unit 1121 determines that the remuneration should be increased, it increases the remuneration in step b3. On the other hand, if the remuneration calculation unit 1121 determines that the remuneration should be decreased, it decreases the remuneration in step b4.
[0147] In step b5, the function update unit 1122 updates the action value function Q(s) represented by equation (1) stored in the trained model storage unit 1130 based on the reward calculated by the reward calculation unit 1121. t , a t ) to update.
[0148] The learning device 1100 repeatedly executes steps b1 to b5 above to generate the action-value function Q(s t , a t ) is stored as a trained model in the trained model storage unit 1130.
[0149] This section explains inference using a trained model.
[0150] 13 is a diagram showing the configuration of an inference device 1200 related to a communication system. The inference device 1200 includes a data acquisition unit 1210 and an inference unit 1220.
[0151] The data acquisition unit 1210 acquires input information.
[0152] The inference unit 1220 infers output information using the trained model. That is, by inputting the input information acquired by the data acquisition unit 1210 into this trained model, output information suitable for the input information can be inferred.
[0153] In this embodiment, the output information is described as being output using a trained model trained by the model generation unit 1120 of the communication system, but it is also possible to obtain a trained model from another communication system and output the output information based on this trained model.
[0154] 13 shows an example of a configuration in which the learned model storage unit 1130 exists outside the inference device 1200, but the learned model storage unit 1130 may also be included in the inference device 1200. The inference device 1200 shown in FIG. 13 may be applied to a base station. The inference device 1200 may be applied to the same node as the learning device 1100, or may be applied to a different node.
[0155] Next, the process for obtaining output information using the inference device 1200 will be described with reference to FIG.
[0156] In step c1, the data acquisition unit 1210 acquires input information.
[0157] In step c2, the inference unit 1220 acquires a trained model from the trained model storage unit 1130, inputs input information into the acquired trained model, and acquires output information.
[0158] In step c3, the inference unit 1220 outputs the output information obtained by the trained model to the communication system.
[0159] In step c4, the communication system executes communication processing using the output information, thereby enabling more effective communication processing using the trained model.
[0160] In this embodiment, a case where reinforcement learning is applied to the learning algorithm used by the inference unit 1220 has been described, but the present invention is not limited to this. As for the learning algorithm, other than reinforcement learning, supervised learning, unsupervised learning, semi-supervised learning, or the like can also be applied.
[0161] Furthermore, the learning algorithm used in the model generation unit 1120 may be deep learning, which learns to extract the features themselves, or machine learning may be performed according to other known methods, such as neural networks, genetic programming, inductive logic programming, or support vector machines.
[0162] The learning device 1100 and the inference device 1200 may be connected to a communication system via a network, for example. The learning device 1100 and the inference device 1200 may also be built into the communication system. Furthermore, the learning device 1100 and the inference device 1200 may exist on a cloud server.
[0163] The model generation unit 1120 may also learn output information using learning data acquired from multiple nodes. The model generation unit 1120 may acquire learning data from multiple nodes used in the same area, or may learn output information using learning data collected from multiple nodes operating independently in different areas. It is also possible to add or remove nodes that collect learning data from the target during the process. Furthermore, the learning device 1100 that has learned output information for a certain node may be applied to another node, and the output information for the other node may be re-learned and updated.
[0164] In this embodiment, an RA processing AI / ML model, which is an AI / ML model used for RA processing, is provided in a base station. A learning (sometimes referred to as model training) device and an inference device may also be provided in a base station. The RA processing AI / ML model may also be provided in a RAN node. For example, it may be provided in a CU or a DU. The RA processing AI / ML model may also be provided in an IAB donor. For example, it may be provided in an IAB donor CU or an IAB donor DU. The RA processing AI / ML model may also be provided in an IAB node. The IAB node constitutes an IAB-DU with DU functionality. For example, it may be provided in the IAB-DU. Configuration information required for RA processing can be derived by the base station and each node using the AI / ML model, making configuration easier.
[0165] The base station may transmit information indicating whether or not it has an AI / ML model for RA processing (which may be information indicating whether or not it has the model) to the UE. RRC signaling may be used for this transmission. The base station may transmit information indicating whether or not it has an AI / ML model for RA processing to a neighboring base station. Xn signaling may be used for this transmission. The base station may transmit information indicating whether or not it has an AI / ML model for RA processing to a CN (Core Network) node (which may be a function). The CN node may be, for example, an NWDAF (Network Data Analysis Function). The base station may transmit information indicating whether or not it has an AI / ML model for RA processing to a management node. The management node may be, for example, an MnS (Management Service) node or an OAM (Operation, Administration, and Management). In this way, the UE or node that has been notified of the information can recognize whether or not the base station that notified it has an AI / ML model for RA processing. It becomes possible to recognize whether or not AI / ML can be introduced in the RA process using the base station.
[0166] Input information for the RA processing AI / ML model is provided. Thirty examples of input information for the RA processing AI / ML model are disclosed below.
[0167] (1) Information about the UE status. (2) Information about the trajectory of the UE. (3) Information about the base station connected to the UE. (4) Information about the base station connected to the UE from neighboring base stations. (5) Information about cell selection and cell reselection of the UE. (6) Information about cell selection and cell reselection of the UE from neighboring base stations. (7) Information about RA processing of the UE. (8) Information about RA processing of the UE from neighboring base stations. (9) Information about RA problems of the UE. (10) Information about RA problems of the UE from neighboring base stations. (11) Information about RLF (Radio Link Failure) due to RA problems of the UE. (12) Information about RLF due to RA problems of the UE from neighboring base stations. (13) Information about radio measurements of the UE. (14) Information about MDT (Minimization of Drive Tests) of the UE. (15) CLI (Cross-Link Interference Management) information at the UE. (16) Information about RACH configuration. (17) Information about base station resources. (18) Information about UE traffic. (19) Information about RRC_Idle UEs. (20) Information about RRC_Inactive UEs. (21) Information about base station load. (22) Information about UL interference. (23) RIM (Remote Interference Management) information at the base station. (24) RIM information at neighboring base stations. (25) Information about L2 (Layer 2) measurements of the UE. (26) Information about L2 measurements from the base station. (27) Information about UE performance. (28) Information about UE QoE (Quality of Experience). (29) Information about propagation delay between the UE and the cell. (30) A combination of (1) to (29).
[0168] (1) may be, for example, the location, movement speed, or movement direction of the UE. The location of the UE may be, for example, coordinates. For example, it may be an area. For example, it may be information identifying the TA, RNA, base station, cell, CU, DU, or TRP in which the UE is located or connected. For example, it may be information identifying the TA, RNA, base station, cell, CU, DU, or TRP with the best reception quality (which may be reception power). It may also be information related to time. The information related to time is not limited to hours, minutes, and seconds, but may also be year, month, date, day of the week, etc. It may also be information related to time used in the network, such as HFN (Hyper Frame Number) or SFN. These pieces of information may be associated with the time. It may also be the remaining battery level of the UE. It may also be information related to the power consumption of the UE. These pieces of information may be combined. The state of the UE can be used as input information for training or inference.
[0169] The information on the UE trajectory in (2) may be, for example, the information on the UE state disclosed in (1) as appropriate. For example, it may be information on the connected base station disclosed in (3). For example, it may be information on cell selection and cell reselection disclosed in (5). It may be information on past and current UE trajectories. The information on past and current UE trajectories may be historical information. The information on trajectories may be information on future UE trajectories. The information on future trajectories may be predicted information. These pieces of information may be made into a list. Information on time may also be included. The information on the UE trajectory and the information on time may be associated. The information on the UE trajectory can be used as input information for training or inference.
[0170] (3) may be, for example, information about a RAN node (which may be, for example, a base station) to which the UE has previously connected. The RAN node may be, for example, a base station, a cell, a cell group, a CU, a DU, a TRP, etc. The base station information may be, for example, information that identifies a base station. For example, it may be setting information about a base station to which the UE has previously connected. For example, it may be history information about base stations to which the UE has previously connected. For example, it may be a list of information about the base stations. It may also include information about the location of the UE. It may also include information about time. The information about base stations to which the UE has previously connected may be associated with information about the location and time. The base stations to which the UE has previously connected may be limited to information about base stations to which the UE has previously connected through RA processing.
[0171] (4) is, for example, information of a neighboring base station about a base station to which the UE has previously connected. The information about the base station to which the UE has previously connected may be the information disclosed in (3).
[0172] (5) may be, for example, information about base stations from which the UE has previously selected and / or reselected a cell. For example, it may be information identifying the base station. For example, it may be configuration information about the base station. It may be history information about base stations from which the UE has previously selected and / or reselected a cell. For example, it may be a list of information about the base stations. It may also include information about the location of the UE. It may also include information about time. Information about base stations from which the UE has previously selected and / or reselected a cell may be associated with information about the location and time. Information when the UE is in an RRC_Idle or RRC_Inactive state can be used as input information for training and inference. The information about base stations from which the UE has previously selected and / or reselected a cell may be limited to information about base stations that performed RA processing.
[0173] (6) is, for example, information on base stations that the UE has in the neighboring base station and / or has previously selected a cell and / or reselected a cell. The information on the UE's cell selection and / or reselection may be the information disclosed in (5).
[0174] (7) may be, for example, information about an RA process previously performed by the UE. For example, it may be information about whether the RA process was successful. For example, it may be information about the base station that performed the RA process. For example, it may be information identifying the base station. For example, it may be setting information used by the UE for the RA process. It may also be history information about RA processes previously performed by the UE. For example, it may be information derived by the UE to perform the RA process (hereinafter referred to as UE derived information). For example, it may be the received power (RSRP) of the SSB (Synchronization Signal Block) used by the UE to determine the PRACH transmission power, the path loss derived based on that, the number of PRACH retransmissions, the PRACH transmission power determined using those, the selected RA resource, the time from PRACH transmission to RAR (Random Access Response) reception, the history of RA problems, etc. It may also be history information about past UE derived information. For example, it may be a list of that information. It may also include information about the UE's location. It may also include information about time. The information about the RA process may be associated with information about the location and time. For example, information about which base stations the UE has performed RA processing for in the past can be used as input information for training or inference.
[0175] (8) is, for example, information about the RA process of the UE that a neighbor base station has. The information about the RA process of the UE may be the information disclosed in (7).
[0176] (9) is, for example, information about a past RA problem of the UE. For example, it may be information about the UE's past RA problem occurrence history. For example, it may be information about a base station where the UE has previously experienced an RA problem. For example, it may be information about the settings used by the UE for the RA process where the UE has previously experienced an RA problem. For example, it may be information about the RACH settings when the RA problem occurred. For example, it may be PRACH transmission power when the RA problem occurred. For example, it may be a list of such information. It may also include information about the UE's location. It may also include information about time. Information about the RA problem may be associated with information about the location or time. For example, information about which base station the UE has previously experienced an RA problem with and which RA settings the RA problem occurred with can be used as input information for training or inference.
[0177] (10) is, for example, information about the UE's past RA problem that a neighbor base station has. The information about the UE's RA problem may be the information disclosed in (9).
[0178] (11) is, for example, information about RLF caused by a past RA problem of the UE. For example, it may be information about the history of RLF occurrence caused by a past RA problem of the UE. For example, it may be information about a base station where the UE previously experienced an RLF caused by an RA problem. For example, it may be information about the settings used for RA processing when the UE previously experienced an RLF caused by an RA problem. For example, it may be a list of such information. It may also include information about the location of the UE. It may also include information about time. Information about RLF caused by an RA problem may be associated with information about location and time. For example, information about which base station the UE previously experienced an RLF caused by an RA problem at, and which RA settings experienced an RLF caused by an RA problem at can be used as input information for training or inference.
[0179] (12) is information about the RLF caused by the UE's past RA problem that a neighbor base station has. The information about the RLF caused by the UE's past RA problem may be the information disclosed in (11).
[0180] (13) is, for example, a radio measurement result by a UE. For example, it may be RSRP, RSRQ, or SINR (Signal to Interference Noise Ratio). It may include information about the location of the UE. It may also include information about time. The radio measurement result may be associated with information about the location or time. The reception quality depending on the location or state of the UE can be used as input information for training or inference.
[0181] (14) is, for example, a measurement result by MDT. For example, it may be a measurement result by immediate MDT or a measurement result by logged MDT. It may be a measurement result by management-based MDT. It may be a measurement result by signaling-based MDT. It may be a measurement result by a trace function. The measurement result by MDT may be associated with information about the UE's location and time. The information about the UE's location and time may be information measured by MDT. By configuring MDT for the UE, the measurement result by MDT can be obtained as input information for the AI / ML model for RA processing. More UE states and more information can be used as input information for training and inference.
[0182] (15) may be, for example, a CLI measurement result at the UE (see Chapter 17.2 of Non-Patent Document 2). The CLI measurement result may be associated with information about the UE's location and time. By configuring the UE for CLI measurement, the CLI measurement result can be obtained as input information for training and inference. Information about uplink interference from other UEs can be used as input information for training and inference.
[0183] (16) is, for example, information about the RACH setting of a base station. For example, it may be the target received power of the base station, the RS transmission power of the base station, a ramping factor, the maximum number of PRACH transmissions, an RAR window, or backoff indicator information. It may be information about the current RACH setting or information about past RACH settings. For example, it may be a list of such information. It may also include information about the location of the base station. It may also include information about the time when the RACH setting was performed. It may also include information about the load of the base station. Information about the RACH setting may be associated with information about the location, time, and load. For example, what kind of RACH setting was performed at which base station can be used as input information for training or inference.
[0184] (17) is, for example, the resource usage of a base station. It may be the current resource usage or the predicted resource usage. For example, it may be the resource usage per base station, per cell group, per cell, per DU, or per TRP. The resource may be a frequency resource or a time resource. The frequency resource usage may be the number of carriers (which may be the number of component carriers (CCs)), the number of RBs, the number of subcarriers, etc. For example, it may be the BWP usage. It may be per carrier, per physical resource block (PRB), or per subcarrier. The time resource usage may be the number of subframes, the number of slots, the number of symbols, etc. It may be per subframe, per slot, or per symbol. The amount of use of the base station resources, etc., can be used as input information for training or inference.
[0185] (18) is, for example, UE traffic at a base station. It may be current traffic or predicted traffic. UE traffic may be, for example, the number of UEs connected to a base station, or the number of UEs using a base station. For example, it may be the number of UEs connected or using per base station, per cell group, per cell, per DU, or per TRP. How many UEs are connected to a base station can be used as input information for training or inference.
[0186] (19) is, for example, the number of UEs in RRC_Idle. It may be the current number of UEs or the predicted number of UEs. The number of UEs in RRC_Idle may be limited to UEs that have completed location registration (sometimes referred to as registration). For example, it may be the number of UEs obtained by subtracting the number of UEs in the RRC connected state and the RRC_Inactive state from the number of UEs that have completed location registration. For example, it may be the number of UEs in RRC_Idle for each PLMN and TA. For example, it may be the number of UEs in RRC_Idle within a paging notification area. For example, the number of UEs in RRC_Idle that may perform RA processing can be used as input information for training or inference.
[0187] (20) is, for example, the number of RRC_Inactive UEs. It may be the current number of UEs or the predicted number of UEs. The number of RRC_Inactive UEs may be, for example, the number of UEs that hold an RRC_Inactive UE context in a base station. For example, it may be the number of UEs to which a base station has notified RRC_Release to RRC_Inactive. For example, it may be the number of RRC_Inactive UEs per PLMN, per TA, or per RNA. For example, the number of RRC_Inactive UEs that may perform RA processing can be used as input information for training or inference.
[0188] (21) is, for example, load information of a base station. It may be current load information or predicted load information. The load information may be, for example, the number of DRBs (Data Radio Bearers) set in the base station, buffer usage, etc. The load information may be information on resources disclosed above, information on UE traffic, or load information other than the number of UEs in RRC_Idle or RRC_Inactive. The load on the base station, etc., may be input information for training or inference.
[0189] (22) is, for example, information about UL interference at a base station. For example, it may be UL interference power derived from an UL signal received by the base station from a UE. For example, it may be UL interference power from a UE that is not connected to the base station. The UL interference power may be SINR. It may be current UL interference power or predicted UL interference power. The level of UL interference occurring at the base station can be used as input information for training or inference.
[0190] (23) may be, for example, the RIM measurement result at the base station (see Non-Patent Document 2, Chapter 17.1). Information about interference from other base stations can be used as input information for training and inference.
[0191] (24) may be, for example, the RIM measurement result at a neighboring base station.
[0192] (25) may be, for example, a measurement result by a UE of L2 measurement information. It may be a current L2 measurement result or a predicted L2 measurement result. The L2 measurement information may be, for example, information measured by a UE as described in Non-Patent Document 32. For example, it may be a packet delay amount. The L2 measurement result by a UE may be an L2 measurement result by a UE connected to a base station through an RA process. For example, it may be an L2 measurement result by a UE for a PCell, PSCell, SPCell, or SCell connected through an RA process. It may also include information about the location of the UE. It may also include information about time. Information about the L2 measurement by the UE may be associated with information about the location and time. Information about communication performance with a base station can be used as input information for training or inference.
[0193] (26) may be, for example, an L2 measurement result by a base station. It may also be an L2 measurement result possessed by a neighboring base station. It may also be a current L2 measurement result or a predicted L2 measurement result. The L2 measurement information by a base station may be, for example, information measured by a base station as described in Non-Patent Document 32. For example, the information may include the number of received RA preambles, packet delay, number of UEs in a connected state, number of stored contexts of UEs in an inactive state, packet loss rate, PRB information used for MIMO (Multiple Input Multiple Output), number of PDCP packets in split DRB, total RAN (Radio Access Network) delay in split DRB, etc. The L2 measurement result by a base station may also be an L2 measurement result by a base station regarding a UE. The L2 measurement result by a base station regarding a UE may include information regarding the UE's location. It may also include information regarding time. Information regarding the L2 measurement by a base station regarding a UE may be associated with information regarding the location or time. Information about communication performance at the base station can be used as input information for training and inference.
[0194] (27) may be, for example, the QoS of the UE. It may be a packet loss rate or a delay time. It may be, for example, data communication quality. It may be, for example, the reception quality of SSB by the UE. It may be, for example, the reception quality of PDCCH and PDSCH by the UE. The reception quality may be RSRP, RSRQ, SINR, or BER. The information about the UE performance may be information after connection with the base station by RA processing. It may be information about the performance required of the UE. It may also include information about the location of the UE. It may also include information about time. The information about the UE performance may be associated with information about the location and time. The information about the UE performance can be used as input information for training or inference.
[0195] (28) may be, for example, a QoE measurement result (sometimes referred to as an application layer measurement result). It may be information after connecting to a base station through RA processing. It may be information about the QoE required by the UE. It may also include information about the location of the UE. It may also include information about time. Information about QoE may be associated with information about the location of the UE or information about time. Measurement results at the application layer of the UE can be used as input information for training or inference.
[0196] (29) may be, for example, propagation delay information between a UE and a cell. The propagation delay information may be, for example, timing advance information. For example, it may be RTT (Round Trip Time) information. The RTT may be, for example, a measurement value of the transmission and reception time difference. For example, it may be information measured by the UE or by a base station. It is not limited to a cell, and may also be a RAN node. For example, by using the propagation delay information, it is possible to obtain a more accurate distance between the UE and the base station or the position of the UE.
[0197] The input information disclosed above can be current information or predicted information, which can be used as input to the AI / ML model to derive more effective output information in the future.
[0198] The input information disclosed above may be associated with a base station, which may be, for example, a base station, a cell, a cell group, a CU, a DU, a TRP, etc. By associating with a base station, it is possible to take into account which base station the input information is for.
[0199] The input information disclosed above may be associated with information about the state of the UE. The information about the state of the UE may be appropriately applied to the information disclosed above. By associating the input information with the state of the UE, it is possible to take into account the state of the UE in which the input information is input.
[0200] When performing RA processing, there is information that only the UE can know. For example, as disclosed in (7) of the example input information above, there is the received power (RSRP) of the SSB by the UE used to determine the PRACH transmission power, the path loss derived based on that, the number of PRACH retransmissions, the PRACH transmission power determined using those, the selected RA resource, the time from PRACH transmission to RAR reception, the history of RA problems, etc. These are not set by the base station, but are determined or selected by the UE. By using such information as input information for the AI / ML model, the processing performed by the UE can be performed more effectively.
[0201] A RAN node transmits input information for an AI / ML model for RA processing to a RAN node having an AI / ML model for RA processing. The notification may include information indicating what the information is. For example, it may include information indicating that it is input information. This makes it easier to distinguish it from input information and feedback information. The notification may include information indicating what the input information is. For example, it may include information indicating that it is input information for training. This makes it easier to distinguish it from other input information.
[0202] A base station having an RA processing AI / ML model inputs the input information for the RA processing AI / ML model disclosed above to the RA processing AI / ML model for training or inference, thereby enabling more effective RA processing to be performed using the RA processing AI / ML model.
[0203] A base station transmits input information for an RA processing AI / ML model to a neighboring base station. A base station may transmit input information for an RA processing AI / ML model to a base station having an RA processing AI / ML model. A base station having an RA processing AI / ML model may transmit a request for input information for an RA processing AI / ML model to a neighboring base station. The request may include information about a UE to be subjected to RA processing. The information about the UE may be a UE identifier. A base station that receives the request transmits input information for an RA processing AI / ML model to the requesting base station. Inter-base station signaling may be used to transmit input information for an RA processing AI / ML model or a request for the information between base stations. For example, Xn signaling may be used. For example, an Xn message may be provided for the transmission. For example, current information and predicted information may be transmitted in separate messages.
[0204] The DU transmits input information for the RA processing AI / ML model to the CU. The DU may transmit input information for the RA processing AI / ML model to a CU having an RA processing AI / ML model. The CU may be a CU-CP (Control Plane) or a CU-UP (User Plane). The CU having the RA processing AI / ML model may transmit a request for input information for the RA processing AI / ML model to the DU. The request may include information about the UE to be subjected to RA processing. The information about the UE may be a UE identifier. The DU that receives the request transmits the input information for the RA processing AI / ML model to the requesting CU. CU-DU signaling may be used to transmit the input information for the RA processing AI / ML model and the request for the information between the CU and DU. For example, F1 signaling may be used. For example, an F1 message may be provided for this transmission. For example, the current information and the predicted information may be sent in separate messages.
[0205] The CU-UP may transmit input information for the RA processing AI / ML model to the CU-CP. The CU-UP may transmit input information for the RA processing AI / ML model to a CU-CP having an RA processing AI / ML model. The CU-CP having an RA processing AI / ML model may transmit a request for input information for the RA processing AI / ML model to the CU-UP. The request may include information about the UE to be subjected to RA processing. The information about the UE may be a UE identifier. The CU-UP that receives the request transmits the input information for the RA processing AI / ML model to the requesting CU-CP. Signaling between the CU-UP and CU-CP may be used to transmit the input information for the RA processing AI / ML model between the CU-UP and CU-CP or a request for the information. For example, E1 signaling may be used. For example, an E1 message may be provided for this transmission. For example, the current information and the predicted information may be transmitted in separate messages.
[0206] The UE transmits input information for the AI / ML model for RA processing to the base station. The UE may transmit the input information for the AI / ML model for RA processing to a base station having an AI / ML model for RA processing. The base station having the AI / ML model for RA processing may transmit a request for input information for the AI / ML model for RA processing to the UE. The request may include information about the UE that is the target of RA processing. The information about the UE may be a UE identifier. The UE that receives the request transmits the input information for the AI / ML model for RA processing to the requesting base station. Signaling between the UE and the base station may be used to transmit the input information for the AI / ML model for RA processing between the UE and the base station or the request for the information. For example, RRC signaling may be used. As another method, MAC signaling may be used. Alternatively, L1 / L2 signaling may be used. The input information may be included in UCI. PUCCH may be used. This allows the information or a request for the information to be sent quickly.
[0207] The transmission of the information or the request for the information between the UE and the base station may be performed, for example, after the RA procedure is successful, or may be performed during the RRC connection procedure, thereby reducing the signaling required for the information or the request for the information.
[0208] The information may be transmitted between the UE and the base station even if a request for the information is not received. For example, the information may be transmitted after the RA process is successful. For example, the information may be transmitted after the RRC connection process is completed. This reduces the signaling required to request the information.
[0209] When the base station receives input information from the UE, the base station may transmit the input information to a neighboring base station. The base station may also transmit the input information even if it does not receive a request for input information from a neighboring base station. This reduces the signaling required to request the information.
[0210] The above-mentioned transmission methods may be combined for each request for information between the UE and the base station or for each piece of information, allowing for flexible setting methods.
[0211] Output information of the AI / ML model for RA processing is disclosed. Information used in RA processing may be used as output information. Fourteen examples of output information of the AI / ML model for RA processing are disclosed below.
[0212] (1) Information about UE trajectory. (2) Information about RACH configuration. (3) Priority of RACH configuration. (4) Information about base station candidates performing RA processing. (5) Probability of reaching base station candidates performing RA processing. (6) Connection time due to RA processing. (7) Information about base station resources. (8) Information about UE traffic. (9) Information about RRC_Idle UE. (10) Information about RRC_Inactive UE. (11) Information about base station load. (12) Information about UL interference. (13) Information about propagation delay between the UE and the cell. (14) A combination of (1) to (13).
[0213] (1) may be, for example, predicted future UE trajectory information. The information on the UE trajectory may be the information disclosed in the example of input information (2) for the AI / ML model for RA processing. Information on past and current trajectories may be included. Historical information on past and current UE trajectories may be included together with the predicted future UE trajectory information. This information may be presented as a list. Information on time may also be included. The information on the UE trajectory and the information on time may be associated with each other.
[0214] (2) is information about the RACH configuration predicted by the AI / ML model. The information about the RACH configuration may be the information disclosed in example (16) of input information for the AI / ML model for RA processing. It may also include information about the predicted UE location. It may also include information about the predicted time. The information about the RACH configuration may be associated with information about the location and time. The information about the RACH configuration is not limited to one piece, and may be multiple pieces. It may also include information about the prediction accuracy of the information about the RACH configuration predicted by the AI / ML model. The UE can use the RACH configuration derived by the AI / ML model. This enables more effective RA processing.
[0215] (3) is the priority of the information related to the RACH configuration indicated by the information disclosed in (2). Priorities may be set for the RACH configurations. The predicted RACH configurations may be associated with their priorities. Priorities are assigned to the multiple output RACH configurations. For example, if an RA problem occurs with the first RACH configuration, RA processing may be performed using the RACH configuration with the next highest priority. This enables more flexible and effective RA processing.
[0216] (4) is information about the base station candidate that will perform the predicted RA process. Base station candidate that will perform the RA process may be provided. The number of base station candidate that will perform the RA process is not limited to one, and may be multiple. Information about the predicted accuracy of the base station candidate that will perform the predicted RA process may be included. For example, the UE can recognize which base station it is more effective to perform the RA process for.
[0217] (5) is information on the predicted probability of arrival at a candidate base station that performs RA processing. It may include information on the confidence interval of the probability of arrival at the candidate base station predicted by the AI / ML model. For example, the UE can recognize which base station is more effective for performing RA processing.
[0218] (6) is information on the connection time by the predicted RA process. The connection time by the predicted RA process may be set for each RACH configuration candidate, for example. For example, it may be set for each base station candidate that performs the RA process. For example, the UE can recognize which base station it is more effective to perform the RA process for.
[0219] (7) is information on the predicted resource status of the base station. The information on the resource status of the base station may be the information disclosed in the example of input information (17) for the AI / ML model for RA processing. The resource status of the base station may be the predicted resource status of the candidate base station. For example, a neighboring base station can recognize the predicted resource status of the base station for which RA processing has been performed.
[0220] (8) is information about predicted UE traffic. The information about UE traffic may be the information disclosed in the example of input information for the AI / ML model for RA processing (18). The information about predicted UE traffic may be predicted UE traffic at the candidate base station. For example, a neighboring base station can recognize the predicted UE traffic of the base station where the RA processing is performed.
[0221] (9) may be, for example, information about predicted RRC_Idle UEs. The information about RRC_Idle UEs may be the information disclosed in the example of input information for the AI / ML model for RA processing (19). For example, a neighboring base station can recognize the number of RRC_Idle UEs when RA processing is performed.
[0222] (10) may be, for example, information about predicted RRC_Inactive UEs. The information about RRC_Inactive UEs may be the information disclosed in the example of input information for the AI / ML model for RA processing (20). For example, a neighboring base station can recognize the number of RRC_Inactive UEs when RA processing is performed.
[0223] (11) may be, for example, information about the predicted load of the base station. The information about the load of the base station may be the information disclosed in the example of input information for the AI / ML model for RA processing (21). For example, a neighboring base station can recognize the load situation of the base station when RA processing is performed.
[0224] (12) may be, for example, information about predicted UL interference. The information about UL interference may be the information disclosed in the example of input information for the AI / ML model for RA processing (22). The information about predicted UL interference may be information about predicted UL interference in a candidate base station. For example, a UE or a neighboring base station can recognize information about predicted UL interference of a base station for which RA processing has been performed.
[0225] (13) may be, for example, information about a predicted propagation delay between the UE and the cell. The information about the propagation delay between the UE and the cell may be the information disclosed in the example of input information (29) for the AI / ML model for RA processing. Information about past and current propagation delays between the UE and the cell may be included. Historical information about the propagation delay between the UE and the cell may be included together with information about the predicted future propagation delay between the UE and the cell. This information may be listed. Information about time may also be included. The information about the propagation delay between the UE and the cell may be associated with the information about time. For example, the UE or the base station may recognize information about the predicted propagation delay between the UE and the cell from the base station where the RA processing was performed.
[0226] An offset may be set in the output information of the AI / ML model for RA processing. An offset may be set in some or all of the output information disclosed above. An offset value is derived by the AI / ML model for RA processing. The offset may be for each UE. By adding a UE-specific offset to the output information for each cell, it becomes possible to set it for each UE. The output information can be optimized for each UE. The offset may be for each UE group. An offset may be set for each predetermined group. For example, it may be a group of UEs that provide the same service. For example, it may be a group of UEs that exist in the same area. The output information can be optimized for each UE group. The offset may be, for example, for each base station, cell group, RNA, or TA. A predicted offset may be set. A predicted offset value may be derived by the AI / ML model for RA processing. In this way, more flexible setting of the output information is possible.
[0227] For example, an offset for deriving the PRACH transmission power is provided. For example, a UE-specific offset is provided in the target received power of the base station. For example, a UE-specific offset is provided in the ramping factor. Not only positive values but also 0 or negative values may be set as the offset value. In this way, it becomes possible to add an offset value for each UE to the RACH configuration set for each cell, making it possible to set the RACH for each UE. By providing an offset, the RACH configuration for each cell can be used as before.
[0228] The offset value may be used as input information for the AI / ML model for RA processing, making it possible to obtain an AI / ML model and inference results that take the offset value into account.
[0229] As another method, RA processing AI / ML model output information may be provided for each UE. Part or all of the output information disclosed above may be output information for each UE. As disclosed above, output information may be for each UE group, rather than for each UE. For example, UE-specific output information may be set by RRC, and the output information notified by SIB may be updated. This allows flexible setting of output information.
[0230] Information regarding the relationship between the UE state and the output information may be provided. The information may be a list. The information regarding the UE state may be the information regarding the UE state disclosed in the above-mentioned input information, as appropriate. For example, it may be a list indicating the relationship between the UE location and RACH configuration information. For example, it may be a list indicating the relationship between the RA procedure start time and RACH configuration information. The information regarding the UE state is not limited to one piece of information, and may be a combination of multiple pieces of information disclosed above. In this way, it is possible to indicate the relationship between the UE state and the output information.
[0231] Information regarding the relationship between the state of the UE and the output information may be notified in advance from a node having an AI / ML model to a node that requires the output information. The node that requires the output information may be, for example, a UE. For example, it may be a base station. In this way, for example, when the UE performs RA processing, notification from the base station to the UE is not required. Even if the UE is in RRC_Idle or RRC_Inactive, the UE itself can determine which RACH configuration to use depending on the UE state.
[0232] The output information of the AI / ML model for RA processing disclosed above is notified to the RAN node or UE performing RA processing as appropriate. A node having the AI / ML model may notify the UE of the information via the RAN node to which the UE is most recently connected. A node having the AI / ML model may broadcast the information to UEs being served by it. In this way, it is possible to perform more effective RA processing using the AI / ML model for RA processing. The notification may include information indicating what kind of information it is. For example, it may include information indicating that it is output information. This makes it easier to distinguish it from input information and feedback information.
[0233] A base station transmits RA processing AI / ML model output information to a neighboring base station. A base station having an RA processing AI / ML model may transmit RA processing AI / ML model output information to a base station involved in RA processing. A base station involved in RA processing may transmit a request for RA processing AI / ML model output information to a base station having an RA processing AI / ML model. A base station that receives the request transmits RA processing AI / ML model output information to the requesting base station. Inter-base station signaling may be used to transmit RA processing AI / ML model output information and a request for the information between base stations. For example, Xn signaling may be used. For example, an Xn message may be provided for the transmission.
[0234] The CU transmits RA processing AI / ML model output information to the DU. A CU having an RA processing AI / ML model may transmit RA processing AI / ML model output information to the DU. The DU may transmit a request for RA processing AI / ML model output information to a CU having an RA processing AI / ML model. The CU that receives the request transmits RA processing AI / ML model output information to the requesting DU. CU-DU signaling may be used to transmit RA processing AI / ML model output information and requests for such information between the CU and DU. For example, F1 signaling may be used. For example, an F1 message may be provided for such transmission.
[0235] The base station transmits RA processing AI / ML model output information to the UE. A base station having an RA processing AI / ML model may transmit RA processing AI / ML model output information to the UE. The UE may transmit a request for RA processing AI / ML model output information to a base station having an RA processing AI / ML model. The base station receiving the request transmits RA processing AI / ML model output information to the requesting UE. Signaling between the UE and the base station may be used to transmit the RA processing AI / ML model output information or a request for that information between the UE and the base station. For example, RRC signaling may be used. As another method, MAC signaling or L1 / L2 signaling may be used. This makes it possible to transmit the information or the request for that information early.
[0236] The base station may transmit per-cell configuration information among the output information to the UE for each cell. For example, the information may be included in an SIB. The SIB may be broadcast. For example, the information may be transmitted to the UE being served by the cell by RRC signaling. The per-cell configuration information can be easily changed, making it possible to optimize the RA process for each cell.
[0237] The base station may transmit setting information for each UE, which is included in the output information, to each UE individually. For example, the information may be transmitted using individual RRC signaling, MAC signaling, or L1 / L2 signaling. This makes it possible to transmit setting information for each UE, thereby optimizing the RA process for each UE.
[0238] These methods of transmitting output information may be combined, which allows for flexible setting of output information and further optimization of RA processing.
[0239] Feedback information resulting from RA processing using an AI / ML model for RA processing (hereinafter, this feedback information will be referred to as "RA processing feedback information using an AI / ML model for RA processing"). Information regarding the performance of each node or communication due to RA processing may also be used as the feedback information. Fifteen examples of RA processing feedback information using an AI / ML model for RA processing are disclosed below.
[0240] (1) Information about UE performance. (2) Information about UE QoE. (3) Information about UE L2 measurements. (4) Information about L2 measurements from the base station. (5) Information about UE RA processing. (6) Information about UE RA problems. (7) Information about RLF due to UE RA problems. (8) Information about base station resources. (9) Information about UE traffic. (10) Information about RRC_Idle UEs. (11) Information about RRC_Inactive UEs. (12) Information about base station load. (13) Information about UL interference. (14) Information about propagation delay between the UE and the cell. (15) A combination of (1) to (14).
[0241] (1) may be, for example, (27) in the above example of input information for the AI / ML model for RA processing.
[0242] (2) may be, for example, (28) in the above example of input information for the AI / ML model for RA processing.
[0243] (3) may be, for example, (25) in the above example of input information for the AI / ML model for RA processing.
[0244] (4) may be, for example, (26) in the above example of input information for the AI / ML model for RA processing.
[0245] (5) may be, for example, (7) or (8) in the above example of input information for the AI / ML model for RA processing.
[0246] (6) may be, for example, (9) or (10) in the above example of input information for the AI / ML model for RA processing.
[0247] (7) may be, for example, (11) or (12) in the above example of input information for the AI / ML model for RA processing.
[0248] (8) may be, for example, (17) in the above example of input information for the AI / ML model for RA processing.
[0249] (9) may be, for example, (18) in the above example of input information for the AI / ML model for RA processing.
[0250] (10) may be, for example, (19) in the above example of input information for the AI / ML model for RA processing.
[0251] (11) may be, for example, (20) in the above example of input information for the AI / ML model for RA processing.
[0252] (12) may be, for example, (21) in the above example of input information for the AI / ML model for RA processing.
[0253] (13) may be, for example, (22) in the above example of input information for the AI / ML model for RA processing.
[0254] (14) may be, for example, (29) in the above example of input information for the AI / ML model for RA processing.
[0255] The RA processing node appropriately notifies the node having the RA processing AI / ML model of the RA processing feedback information using the RA processing AI / ML model disclosed above. This makes it possible to evaluate the reward calculation of the RA processing AI / ML model, etc., and to update the model. This makes it possible to execute more effective RA processing. The notification may include information indicating what kind of information it is. For example, it may include information indicating that it is feedback information. This makes it easier to distinguish it from input information and output information.
[0256] A base station transmits RA processing feedback information using the AI / ML model for RA processing to a neighboring base station. A base station involved in RA processing may transmit RA processing feedback information using the AI / ML model for RA processing to a base station having an AI / ML model for RA processing. A base station having an AI / ML model for RA processing may transmit a request for RA processing feedback information using the AI / ML model for RA processing to a base station involved in RA processing. A base station that receives the request transmits RA processing feedback information using the AI / ML model for RA processing to the requesting base station. Inter-base station signaling may be used to transmit RA processing feedback information using the AI / ML model for RA processing between base stations or a request for such information. For example, Xn signaling may be used.
[0257] The DU transmits RA processing feedback information using the AI / ML model for RA processing to the CU. A DU involved in RA processing may transmit RA processing feedback information using the AI / ML model for RA processing to a CU. A DU involved in RA processing may transmit RA processing feedback information using the AI / ML model for RA processing to a CU having an AI / ML model for RA processing. A CU having an AI / ML model for RA processing may transmit a request for RA processing feedback information using the AI / ML model for RA processing to the DU. The DU that receives the request transmits RA processing feedback information using the AI / ML model for RA processing to the requesting CU. CU-DU signaling may be used to transmit RA processing feedback information using the AI / ML model for RA processing between the CU and DU, or a request for such information. For example, F1 signaling may be used.
[0258] The UE transmits RA processing feedback information using the AI / ML model for RA processing to the base station. The UE may transmit RA processing feedback information using the AI / ML model for RA processing to a base station having an AI / ML model for RA processing. The base station having an AI / ML model for RA processing may transmit a request for RA processing feedback information using the AI / ML model for RA processing to the UE. The UE, having received the request, transmits RA processing feedback information using the AI / ML model for RA processing to the requesting base station. Signaling between the UE and the base station may be used to transmit RA processing feedback information using the AI / ML model for RA processing or a request for this information between the UE and the base station. For example, RRC signaling may be used. As another method, MAC signaling or L1 / L2 signaling may be used. This makes it possible to transmit the information or the request for this information early.
[0259] 15 is a diagram showing an example of a sequence of RA processing using an AI / ML for RA processing according to the first embodiment. This example shows an example in which a RAN node has an AI / ML model for RA processing and performs training and inference. The RAN node may be, for example, a base station.
[0260] In steps ST1501 and ST1502, RAN nodes #1 and #2 are provided with an AI / ML model for RA processing. An AI / ML model for RA processing may be provided in all RAN nodes that can execute AI / ML for RA processing. The same AI / ML model may be provided within a PLMN. The same model may be provided within an RAN. The same model may be provided within a predetermined area. In step ST1503, the UE is in an RRC connected state with RAN node #1. RAN node #1, which has an AI / ML model for RA processing, transmits an input information request for model training to the UE in step ST1511 and to a neighboring RAN node (RAN node #2 in this example) in step ST1512. Similarly, in step ST1513, RAN node #2 having an AI / ML model for RA processing transmits a request for input information for model training to a surrounding RAN node (RAN node #1 in this example) and to a UE connected to RAN node #2 (not shown in the figure).
[0261] The request may be triggered by receiving a request for input information for the AI / ML model for RA processing from a neighboring RAN node. As another method, RAN node #1 may send a model training implementation request to a neighboring RAN node. For example, when the RAN node itself decides to implement model training, it may send a model training implementation request to a neighboring RAN node. In this way, model training using the AI / ML model for RA processing can be implemented not only by RAN node #1 but also by neighboring RAN nodes.
[0262] The request may include information indicating the required information. The request may include information indicating the purpose of the input information for the AI / ML model for RA processing. For example, the request may include information indicating that it is for RA processing. For example, the request may include information indicating that it is for model training. This is effective, for example, when the input information for the AI / ML model for RA processing includes information for model training. For example, when the information held by the UE includes information for model training, information for inference, and information for feedback, the UE can quickly ascertain the information that should be notified to the RAN node, and as a result, quick notification from the UE to the RAN node is possible.
[0263] The request may be transmitted between the RAN node and the UE using, for example, RRC signaling. The request for input information for the AI / ML model for the RAN node may be included in a measurement configuration transmitted from the RAN node to the UE, thereby reducing the amount of signaling.
[0264] For example, Xn signaling may be used to transmit the request between RAN nodes.
[0265] In steps ST1514 and ST1516, the UE and RAN node #2 transmit model training input information to RAN node #1. RAN node #2 may transmit model training input information acquired from a connected UE to RAN node #1. RAN node #1 is then able to input information about UEs connected to RAN node #2 into the AI / ML model. In step ST1515, RAN node #1 transmits model training input information to RAN node #2. Also, although not shown in the figure, a UE connected to RAN node #2 transmits model training input information to RAN node #2. RAN node #1 may transmit model training input information acquired from the UE to RAN node #2. RAN node #2 is then able to input information about UEs connected to RAN node #1 into the AI / ML model.
[0266] Measurement results by the UE may be transmitted to the RAN node in a measurement report. The measurement results may be radio measurements, L2 measurements, or MDT measurements. Logged MDT measurements may be measurements made by the UE in RRC_Inactive or RRC_Idle state. These measurements may be transmitted via the same signaling or via different signaling. The UE measurements may be used as input for model training.
[0267] For example, RRC signaling may be used to transmit input information for model training between the RAN node and the UE. For example, a UEInformationResponse message may be used. For example, a MeasurementReport message may be used to transmit radio measurement results and L2 measurement results. Using conventional messages can avoid complex processing.
[0268] For example, Xn signaling may be used to transmit the input information between RAN nodes.
[0269] In steps ST1517 and ST1518, RAN node #1 and RAN node #2 train the RA processing AI / ML model using input information for the RA processing AI / ML model held by the own node and input information for the RA processing AI / ML model acquired from the UE and adjacent RAN nodes. The RA processing AI / ML model is updated through training. These processes may be performed periodically or each time RA processing is performed in the own node or in a neighboring RAN node.
[0270] In step ST1521, RAN node #1 determines to perform RRC release for the UE. In steps ST1531 and ST1532, RAN node #1 transmits an inference input information request to the UE and neighboring RAN nodes. RAN node #2 having an RA processing AI / ML model similarly transmits an inference input information request to neighboring RAN nodes and to a UE (not shown) connected to RAN node #2 in step ST1533. This request may be triggered by receiving a request for input information for the RA processing AI / ML model from a neighboring RAN node. As an alternative method, RAN node #1 may transmit an inference execution request to a neighboring RAN node. For example, RAN node #1 that has determined RRC release for the UE may transmit an inference execution request to a neighboring RAN node. In this way, inference using the RA processing AI / ML model can be performed not only by RAN node #1 but also by neighboring RAN nodes.
[0271] The request may include information indicating the required information. The request may include information indicating the use of the input information for the RA processing AI / ML model. For example, the request may include information indicating that the input information is for inference. This is effective, for example, when the input information for the RA processing AI / ML model includes information for inference.
[0272] In steps ST1534 and ST1536, the UE and RAN node #2 transmit inference input information to RAN node #1. RAN node #2 may transmit inference input information acquired from a connected UE to RAN node #1. RAN node #1 becomes able to input information of a UE connected to RAN node #2 into the AI / ML model. In step ST1535, RAN node #1 transmits inference input information to RAN node #2. Also, although not shown in the figure, a UE connected to RAN node #2 transmits inference input information to RAN node #2. RAN node #1 may transmit inference input information acquired from the UE to RAN node #2. RAN node #2 becomes able to input information of a UE connected to RAN node #1 into the AI / ML model.
[0273] The method of transmitting an input information request for an AI / ML model for RA processing (corresponding to the input information request for model training and the input information request for inference shown in Figure 15) and input information for an AI / ML model for RA processing (corresponding to the input information for model training and the input information for inference shown in Figure 15) may be any of the methods disclosed above, as appropriate.
[0274] In steps ST1537 and ST1538, RAN nodes #1 and #2 perform inference using the RA processing AI / ML model, using the RA processing AI / ML model input information held by the own nodes and the RA processing AI / ML model input information acquired from the UE and adjacent base stations. The RAN nodes derive RA processing AI / ML model output information through this inference.
[0275] In step ST1541, RAN node #1 transmits RA processing AI / ML model output information to the UE. For example, UE-specific output information may be transmitted. For example, UE-specific RACH configuration may be transmitted. For example, UE-specific offset to be added to the cell's RACH configuration may be transmitted. Before the UE transitions to the RRC_Idle or RRC_Inactive state, for example, before performing RRC release processing, or during RRC release processing, it is preferable to perform transmission / reception processing of RA processing AI / ML model output information. In this way, the UE can obtain UE-specific output information.
[0276] The RA processing AI / ML model output information may be transmitted between RAN nodes. RAN node #1, which performed the inference in step ST1537, may transmit output information (inference result) to RAN node #2. RAN node #2, which performed the inference in step ST1538, may transmit output information (inference result) to RAN node #1. The RA processing AI / ML model output information can be shared between RAN nodes. RAN node #1 may transmit the output information received from RAN node #2 to the UE. For example, this may be transmitted in step ST1541. The UE can acquire the RA processing AI / ML model output information of each RAN node. This is also effective, for example, when the RA processing AI / ML model output information of RN node #1 and the RA processing AI / ML model output information of RAN node #2 differ.
[0277] In Step ST1551, the RAN node #1 performs an RRC release procedure with the UE. The UE transitions to the RRC_Idle or RRC_Inactive state.
[0278] In steps ST1561 and ST1562, RAN nodes #1 and #2 transmit output information derived using the AI / ML model for RA processing. For example, they transmit RACH configuration. The output information may be broadcast, for example, by being included in an SIB. These processes may be performed before the process of step ST1551. They may be performed after steps ST1537 and ST1538. In this way, RAN node #1 and surrounding RAN nodes can transmit the AI / ML model output information for RA processing, and the UE can perform RA processing using the AI / ML model output information for RA processing.
[0279] The UE determines an RRC connection with a RAN node. The UE may be moving. In step ST1571, the UE performs RA processing between the UE and RAN node #2 using the RA processing AI / ML model output information received from RAN node #2 and / or the RA processing AI / ML model output information received from RAN node #1. FIG. 15 illustrates a case where the UE selects RAN node #2. In this way, RA processing is possible between the UE and RAN node #2 using the output information derived by the RA processing AI / ML model.
[0280] In step ST1572, the UE performs an RRC connection process with the RAN node #2. Part of the RRC connection process may be performed by the RA process. This allows the UE to transition to an RRC connected state with the RAN node #2. For example, data communication between the UE and the network becomes possible.
[0281] In step ST1581, the UE transmits RA processing feedback information using the AI / ML model for RA processing to RAN node #2. The feedback information may include information about the RAN node to which the UE was last RRC connected. For example, it may include information to identify the RAN node. The RAN node that receives this information can recognize the RAN node to which the UE was last RRC connected. The feedback information can then be transmitted to the RAN node.
[0282] As another method, the UE may transmit information to RAN node #2 requesting the transmission of feedback information, either included in the feedback information or separately. Receipt of this information may trigger RAN node #2 to transmit feedback information to neighboring RAN nodes. RAN node #2 may transmit feedback information to RAN node #1 using information received from the UE regarding the RAN node to which the UE was last RRC connected. The UE can request that feedback information be transmitted from RAN node #2 to RAN node #1 as appropriate. Flexible control is possible.
[0283] The transmission of the feedback information in step ST1581 may be performed when the UE performs the RA procedure. Alternatively, the UE may perform the RA procedure and transmit the feedback information to a RAN node to which it has an RRC connection. In this way, the RAN node can obtain the feedback information every time the UE performs the RA procedure.
[0284] As another method, the RAN node #2 to which the UE has connected after performing the RA process may transmit a request for feedback information to the UE. The UE that has received the request may transmit feedback information to the RAN node #2. In this way, it is possible to transmit feedback information from the UE to the RAN node #2 as needed, thereby reducing the amount of signaling.
[0285] In step ST1581, RAN node #2 that has received feedback information from the UE transmits the UE feedback information to RAN node #1. In step ST1582, RAN node #2 transmits the feedback information that its own RAN node has to RAN node #1. In step ST1583, RAN node #1 may transmit the feedback information that its own RAN node has to RAN node #2.
[0286] The method for transmitting the RA processing feedback information request using the AI / ML model for RA processing (not shown in Figure 15) and the RA processing feedback information used for the AI / ML model for RA processing (corresponding to the feedback information shown in Figure 15) may be appropriately applied by applying the method for input information disclosed above.
[0287] There may be no need to request input information or feedback information from the AI / ML model for RA processing. Each node may transmit input information or feedback information from the AI / ML model for RA processing to a node having the AI / ML model for RA processing as appropriate. For example, the transmission may be performed periodically. For example, the input information or feedback information may be transmitted when new input information or feedback information is obtained. This reduces the amount of signaling. When a node having an AI / ML model for RA processing receives the information from another node or from its own node, it may transmit an input information request to other nodes excluding the node. This reduces unnecessary signaling, such as transmitting the same input information.
[0288] As another method, one RAN node may perform AI / ML model training. The RAN node that performed the model training may transmit updated information of the AI / ML model to other RAN nodes. The other RAN nodes may update their AI / ML models using the received model update information. The base station to which the UE last established an RRC connection may perform AI / ML model inference and transmit output information to the UE in an RRC Release process. This simplifies the processing using the AI / ML model as a system.
[0289] In this way, the RAN node can acquire information on the performance of RA processing using the AI / ML model for RA processing. The RAN node may evaluate the effectiveness of RA processing when the AI / ML model for RA processing is used, using the RA processing feedback information using the AI / ML model for RA processing that the RAN node holds and the RA processing feedback information using the AI / ML model for RA processing that is acquired from the UE and neighboring RAN nodes.
[0290] In this way, AI / ML is introduced into the RA processing, enabling more effective RA processing.
[0291] By using the method disclosed in this embodiment, it is possible to perform training and inference in RA processing using an AI / ML model. By deriving output information of the AI / ML model using various types of input information, it is possible to optimize RA processing. Furthermore, by providing UE-specific RACH settings, it is possible to optimize RACH settings suitable for the UE's state. Since more appropriate RACH settings are possible, it is possible to shorten the time required for UEs to connect to a cell through RA processing. Furthermore, it is possible to optimize various KPIs in RA processing, such as reducing UL interference, reducing UE transmission power, and reducing RA processing failures.
[0292] First Modification of First Embodiment In order to solve the problem disclosed in the first embodiment, another method that enables more effective RA processing according to the communication situation will be disclosed.
[0293] In this modification, an AI / ML model for RA processing is provided in the UE. A learning device and an inference device are provided in the UE. The UE can derive the setting information required for RA processing using the AI / ML model. The UE may be an IAB node. The IAB node constitutes an IAB-MT having terminal functions. For example, the IAB-MT may be provided with an AI / ML model for RA processing. The IAB node can derive the setting information required for RA processing using the AI / ML model.
[0294] The input information, output information, and feedback information of the AI / ML model for RA processing may be appropriately applied to the information disclosed in the first embodiment, and the same effects can be obtained.
[0295] The UE may transmit information on whether or not it has an AI / ML model for RA processing (which may be information on whether or not it has the model) to a base station. RRC signaling may be used for this transmission. The base station that receives the information from the UE may transmit the information on whether or not it has an AI / ML model for RA processing to a neighboring base station. Xn signaling may be used for this transmission. The UE may transmit the information on whether or not it has an AI / ML model for RA processing to a CN (Core Network) node (which may be a function). The base station that receives the information from the UE may transmit the information to the CN node. The CN node may be, for example, a NWDAF (Network Data Analysis Function). The UE may transmit the information on whether or not it has an AI / ML model for RA processing to a management node. The base station or CN node that receives the information from the UE may transmit the information to the management node. The management node may be, for example, a MnS (Management Service) node or an OAM (Operation, Administration and Management). In this way, the base station, CN node, or management node that has received the information can recognize whether the UE that sent the notification has an AI / ML model for RA processing, and can recognize whether AI / ML can be introduced in the RA processing of the UE.
[0296] The UE receives the AI / ML model input information for RA processing from a RAN node. The RAN node may be, for example, a base station or a serving cell. The UE may receive the input information when in an RRC connected state. To receive the input information, the UE may transition to an RRC connected state. The UE may perform an RRC connection procedure with the RAN node. Alternatively, the UE may receive the input information using an SDT (Small Data Transmission) procedure. This allows the UE to receive the input information without transitioning to an RRC connected state. This reduces the amount of signaling and avoids complicated processing.
[0297] To train the AI / ML model for RA processing, the UE may transition to an RRC connected state. For example, input information for model training may be received via an RRC-connected base station. Alternatively, to train the AI / ML model for RA processing, the UE may not transition to an RRC connected state. For example, input information for model training may be received using an SDT procedure.
[0298] The AI / ML model training for RA processing is performed as appropriate. For example, it may be performed periodically. The model is updated periodically. The period may be statically determined in advance by a standard or the like, or may be set by the NW node. As another method, the AI / ML model training for RA processing may be performed when the UE performs RA processing. When a UE in RRC_Inactive or RRC_Idle establishes an RRC connection with the cell in which the RA processing has been performed, it can receive input information for model training. As another method, the AI / ML model training for RA processing may be performed when the UE performs cell selection or cell reselection. For example, the model training may be performed when an RNA update process is performed. For example, the model training may be performed when a TA update process is performed. When a UE in RRC_Inactive or RRC_Idle establishes an RRC connection with a new cell, it can receive input information for model training.
[0299] Inference using the AI / ML model for RA processing is performed when the UE activates RA processing. This may be performed when the UE is in RRC_Idle or RRC_Inactive. Input information for inference may be limited to information related to the UE's RA processing and / or information related to the UE's state. This may eliminate the need for input information from other nodes for inference in the UE.
[0300] The NW node may transmit a request to the UE to perform model training. The NW node may transmit a request to perform inference to the UE. Paging processing may be used as a method for transmitting the request. The request may be included in paging information and transmitted from the base station. As another method, the request may be included in an SIB and broadcast by the base station. In this way, the NW node can transmit a request to perform model training or inference to the UE. The UE can perform model training or inference upon request from the NW node. The NW node can determine whether the UE can perform model training or inference.
[0301] The number of UEs to be subjected to model training or inference is not limited to one, and may be multiple. It may be a specific UE. For example, it may be a predetermined UE group. For example, it may be a UE group that provides the same service. For example, it may be a UE under the control of a base station. For example, it may be a UE in an RNA. For example, it may be a UE in a TA.
[0302] Figure 16 is a diagram showing an example of the sequence of RA processing using an AI / ML for RA processing, according to Modification 1 of Embodiment 1. This figure discloses an example in which a UE has an AI / ML model for RA processing and performs training and inference. Steps common to those in Figure 15 are given the same step numbers, and common explanations will be omitted. In step ST1601, an AI / ML model for RA processing is provided in the UE. In step ST1503, the UE is in an RRC-connected state with RAN node #1.
[0303] In step ST1611, a UE having an AI / ML model for RA processing transmits a request for input information for model training to a connected RAN node (RAN node #1 in this example). In step ST1612, RAN node #1 transmits a request for input information for model training to a neighboring RAN node (RAN node #2 in this example). This request may be triggered by receiving a request for input information for the AI / ML model for RA processing from the UE.
[0304] The information to be included in the request may be the same as that disclosed in the first embodiment. The request may include information indicating that the UE has an AI / ML model or information identifying the UE. This allows the RAN node (including neighboring RAN nodes) to quickly grasp the information to be notified to the UE, thereby enabling the RAN node to quickly notify the UE.
[0305] For example, RRC signaling may be used to request and transmit the input information between the RAN node and the UE. The request for input information for the AI / ML model for the RAN node may be included in a measurement configuration transmitted from the RAN node to the UE. This reduces the amount of signaling.
[0306] For example, Xn signaling may be used to request and transmit the input information between RAN nodes.
[0307] In step ST1615, the UE trains the RA processing AI / ML model using the input information for the RA processing AI / ML model held by the UE itself and the input information for the RA processing AI / ML model acquired from the RAN node. The RA processing AI / ML model is updated through the training. These processes may be performed as appropriate, or may be performed periodically. By performing the training as appropriate, the AI / ML model is updated to one that is suitable for more effective RA processing.
[0308] In step ST1551, the RAN node #1 performs an RRC release procedure for the UE. Unlike the first embodiment, for example, the RAN node #1 may start the RRC release procedure immediately after deciding to release the RRC for the UE. The UE transitions to the RRC_Idle or RRC_Inactive state.
[0309] Input information for the AI / ML for RA processing may be transmitted and received prior to the RRC release process in step 1551. For example, RAN node #1 may transmit input information for the AI / ML for RA processing to the UE. For example, RAN node #1 may transmit a request for input information for the AI / ML for RA processing to RAN node #2. RAN node #2, having received the request, transmits input information for the AI / ML for RA processing to RAN node #1. RAN node #1 transmits the input information received from RAN node #2 to the UE. RAN node #1 may transmit the input information of its own RAN node and that of RAN node #2 together to the UE. The input information may be used for training the AI / ML model or for inference. In this way, the UE can input the input information received from the RAN node immediately before RRC release to the AI / ML model. The time from receiving input information from the RAN node to the UE performing RA processing can be shortened, which results in more effective output information of the AI / ML model for RA processing, especially when used for inference.
[0310] In steps ST1561 and ST1562, RAN nodes #1 and #2 transmit RACH configuration.
[0311] The UE determines an RRC connection with a RAN node. The UE may be moving. In Step ST1631, the UE collects input information for the AI / ML model for RA processing in the UE itself. For example, the UE collects information related to RA processing of the UE and information related to the UE state.
[0312] In step ST1632, the UE performs inference using the RA processing AI / ML model using its own UE's RA processing AI / ML model input information and / or the RACH configuration acquired from RAN node #1. The UE derives RA processing AI / ML model output information through this inference. As the inference input information, the UE may use the most recent model training input information acquired from RAN node #1 or RAN node #2. A wider variety of input information can be input to the inference.
[0313] In step ST1571, the UE performs RA processing with RAN node #2. The UE performs RA processing using output information derived using the AI / ML model for RA processing. In this way, RA processing using AI / ML becomes possible.
[0314] In step ST1572, the UE performs an RRC connection process with the RAN node #2. Part of the RRC connection process may be performed by the RA process. This allows the UE to transition to an RRC connected state with the RAN node #2. For example, data communication between the UE and the NW becomes possible.
[0315] In step ST1681, the RAN node #2 transmits RA processing feedback information using the AI / ML model for RA processing to the UE.
[0316] The transmission of the feedback information in step ST1681 may be performed when the UE performs the RA procedure. Alternatively, the UE may perform the RA procedure and the feedback information may be transmitted from a RAN node to which the UE has an RRC connection. In this way, the UE can obtain the feedback information every time it performs the RA procedure.
[0317] The UE may send a request for feedback information to RAN node #2. RAN node #2 can explicitly recognize that the UE is requesting feedback information. The request may include information about the RAN node to which the UE had last RRC connection. For example, it may include information identifying the RAN node. RAN node #2 receiving this information can recognize the RAN node to which the UE had last RRC connection. RAN node #2 may request feedback information from the RAN node to which the UE had last RRC connection (RAN node #1 in this example). The request may include information about the UE that performed the RA process. It may also include information identifying the UE. RAN node #1 receiving the request for feedback information transmits the feedback information to RAN node #2. RAN node #2 may transmit the feedback information obtained from RAN node #1 to the UE. In this way, the UE can obtain feedback information not only from RAN node #2 but also from RAN node #1. RAN node #2 may request feedback information from neighboring RAN nodes, not limited to RAN node #1. The RAN node that receives the request transmits feedback information to RAN node #2. RAN node #2 may transmit feedback information acquired from neighboring RAN nodes to the UE. In this way, the UE can obtain feedback information from neighboring RAN nodes of RAN node #2.
[0318] In this way, the UE can acquire information on the performance of RA processing using the AI / ML model for RA processing. The UE may evaluate the effectiveness of RA processing when the AI / ML model for RA processing is used, using the RA processing feedback information using the AI / ML model for RA processing that the UE holds and the RA processing feedback information using the AI / ML model for RA processing that is acquired from the RAN node.
[0319] There may be no need to request input information or feedback information from the AI / ML model for RA processing. Each node may transmit input information or feedback information from the AI / ML model for RA processing to a UE having the AI / ML model for RA processing as appropriate. For example, the transmission may be performed periodically. For example, the input information or feedback information may be transmitted when new input information or feedback information is obtained. This reduces the amount of signaling. When a node having an AI / ML model for RA processing receives the information from another node or the node itself, it may transmit an input information request to other nodes excluding the node. This reduces unnecessary signaling, such as transmitting the same input information.
[0320] In this way, AI / ML is introduced into the RA processing, enabling more effective RA processing.
[0321] By using the method disclosed in this modification, it is possible to obtain the same effects as in the first embodiment. Furthermore, since it is possible for the UE to perform inference when activating the RA procedure, it is possible to execute a more appropriate RA procedure according to the status of the UE. Furthermore, since the UE performs the inference, it is possible to reduce the amount of signaling required for inference and simplify the inference procedure.
[0322] For a node (in this specification, a UE may also be considered a node) in which an AI / ML model for RA processing is provided, the method disclosed in the first embodiment and the method disclosed in this modification may be combined. For example, a learning device that trains an AI / ML model for RA processing may be provided in the RAN node, and an inference device may be provided in the UE. Model training may be performed using the method disclosed in the first embodiment, and inference may be performed using the method disclosed in the first modification of the first embodiment. Input information required for model training may be transmitted to the RAN node, and input information required for inference may be transmitted to the UE. For example, model training may be performed based on input information from many RAN nodes, thereby improving the accuracy of the model. Performing inference in the UE may derive output information closer to the RA processing, thereby enabling more effective RA processing.
[0323] For example, an AI / ML model for RA processing may be provided in the RAN node and the UE. Input information and feedback information for the model may be transmitted between the RAN node and the UE. For example, applying the AI / ML model to RA processing in a UE in an RRC_Connected state may enable more effective RA processing.
[0324] The method disclosed in the first embodiment and this modification may be applied to the RA process when the UE is connected not only to the PCell of the MCG but also to the SCell, the PSCell of the SCG, or the SCell. It may be applied to the RA process when connected to a cell. It may also be applied to the RA process performed by the IAB node. It may also be applied to the RA process performed by a UE in the RRC_Idle or RRC_Inactive state, as well as the RA process performed by a UE in the RRC_Connected state. It may also be applied to the RA process when the UE is connected to the HO destination RAN node or the DC destination RAN node. Similar effects can be obtained.
[0325] Embodiment 2. 3GPP supports SDT (Small Data Transmission) (see Chapter 18 of Non-Patent Document 2). SDT is a function that enables an RRC_INACTIVE UE to transmit and receive data without transitioning to RRC_CONNECTED. This allows the UE to transmit and receive data quickly. For example, when data communication is required intermittently, SDT is effective because it enables data communication with low latency. However, SDT can be activated only when the UL data volume is equal to or less than a predetermined threshold and the received power is greater than a predetermined threshold. The UL data volume threshold and the received power threshold are set by the base station. It is necessary to optimize the UL data volume threshold and the received power threshold so that an RRC_INACTIVE UE can use SDT as much as possible.
[0326] In this embodiment, a method is disclosed that enables more effective SDT processing according to the communication situation.
[0327] In this embodiment, AI / ML is introduced into SDT processing. SDT processing includes, for example, setting a UL data volume threshold, a received power threshold, and a timer value for SDT. AI / ML is used in SDT processing to optimize such information.
[0328] The methods disclosed in the first embodiment may be applied as appropriate to the AI / ML model, learning device, and inference device.
[0329] In this embodiment, an AI / ML model for SDT processing, which is an AI / ML model used for SDT processing, is provided in a base station. A learning device and an inference device may also be provided in a base station. The AI / ML model for SDT processing may also be provided in a RAN node. For example, it may be provided in a CU or a DU. Configuration information required for SDT processing can be derived by the base station and each node using the AI / ML model, making configuration easier.
[0330] Regarding the information on whether or not an AI / ML model for SDT processing is available, the method disclosed in the first embodiment may be applied as appropriate.
[0331] Provide input information for the AI / ML model for SDT processing. 25 examples of input information for the AI / ML model for RA processing are disclosed below.
[0332] (1) Information about the UE state. (2) Information about the UE trajectory. (3) Information about the UE's connected base station. (4) Information about the UE's connected base station from neighboring base stations. (5) Information about the UE's cell selection and cell reselection. (6) Information about the UE's cell selection and cell reselection from neighboring base stations. (7) Information about the UE's SDT processing. (8) Information about the UE's SDT processing from neighboring base stations. (9) Information about the UE's SDT timer expiry. (10) Information about the UE's SDT timer expiry from neighboring base stations. (11) Information about the UE's radio measurements. (12) Information about the UE's MDT. (13) Information about SDT settings. (14) Information about base station resources. (15) Information about the UE's traffic. (16) Information about RRC_Idle UEs. (17) Information about RRC_Inactive UEs. (18) Information about the load of the base station. (19) Information about L2 (Layer 2) measurements of the UE. (20) Information about L2 measurements from the base station. (21) Information about the performance of the UE. (22) Information about the QoE (Quality of Experience) of the UE. (23) Information about the performance of the base station. (24) Information about the propagation delay between the UE and the cell. (25) A combination of (1) to (24).
[0333] For (1) to (6), it is preferable to appropriately apply (1) to (6) of the example of input information for the AI / ML model for RA processing disclosed in embodiment 1. For (3) and (5) of the example of input information disclosed in embodiment 1, it is preferable to replace the description of RA processing with SDT processing.
[0334] (7) may be, for example, information about SDT processing previously performed by the UE. For example, it may be information about whether the SDT processing was successful. For example, it may be information about the base station that performed the SDT processing. For example, it may be information that identifies the base station. For example, it may be setting information used by the UE for the SDT processing. It may be history information about SDT processing previously performed by the UE. For example, it may be information derived by the UE to perform SDT processing (hereinafter referred to as UE derived information). For example, UL data volume, received power, etc. It may also be history information about past UE derived information. For example, it may be a list of such information. It may also include information about the location of the UE. It may also include information about time. Information about SDT processing may be associated with information about location and time. For example, the input information may be which base stations the UE has previously performed SDT processing for.
[0335] (8) is, for example, information about the SDT processing of the UE that a neighbor base station has. The information about the SDT processing of the UE may be the information disclosed in (7).
[0336] (9) is, for example, information about the UE's past SDT timer expiry. For example, it may be information about the UE's past SDT timer expiry history. For example, it may be information about the base station where the UE's SDT timer expiry occurred in the past. For example, it may be information about the settings used by the UE for the SDT process where the SDT timer expiry occurred in the past. For example, it may be a list of such information. It may also include information about the UE's location. It may also include information about time. Information about SDT timer expiry may be associated with information about location and time. For example, the input information may be information about which base station the UE's SDT timer expiry problem occurred in in the past, or which SDT setting the SDT timer expiry occurred in.
[0337] (10) is, for example, information about the past SDT timer expiration of the UE held by a neighbor base station. The information about the SDT timer expiration of the UE may be the information disclosed in (9).
[0338] (11) and (12) may be appropriately applied to (13) and (14), respectively, which are examples of input information for the AI / ML model for RA processing disclosed in the first embodiment.
[0339] (13) is, for example, information about the SDT setting of the base station. For example, it is an UL data volume threshold, a received power threshold, an SDT timer value, etc. It may be information about the current SDT setting, or information about past SDT settings. For example, it may be a list of the information. For example, it may be an SDT setting included in an SIB and transmitted, or an SDT setting included in an RRCRelease message and transmitted. For example, it may be an SDT setting set for each cell, or an SDT setting set for each UE. It may also include information about the location of the base station. It may also include information about the time when the SDT setting was performed. It may also include information about the load of the base station. Information about the SDT setting may be associated with information about the location, time, and load. For example, the input information may be which SDT setting was performed at which base station.
[0340] (14) to (18) may be appropriately applied to (17) to (21) of the example of input information for the AI / ML model for RA processing disclosed in the first embodiment.
[0341] For (19) to (22), (25) to (28) of the input information examples for the AI / ML model for RA processing disclosed in embodiment 1 may be applied as appropriate. Regarding (19), "It may be an L2 measurement result by a UE connected to a base station by RA processing. For example, it may be an L2 measurement result by a UE for a PCell, PSCell, SPCell, or SCell connected by RA processing" in the input information (25) for the AI / ML model for RA processing disclosed in embodiment 1 may be changed to "It may be an L2 measurement result by a UE that communicates with a base station by SDT processing. For example, it may be an L2 measurement result by a UE for a PCell, PSCell, SPCell, or SCell that communicates by SDT processing." Regarding (21), in the input information (27) for the AI / ML model for RA processing disclosed in the first embodiment, “may be information after connecting with a base station by RA processing” may be changed to “may be information after communication with a base station by SDT processing.” Regarding (22), in the input information (28) for the AI / ML model for RA processing disclosed in the first embodiment, “may be information after connecting with a base station by RA processing” may be changed to “may be information after communication with a base station by SDT processing.”
[0342] (23) may be, for example, information about the throughput of a base station. For example, it may be information about the delay time of a base station. For example, it may be information about the packet loss rate of a base station. It may be the total amount at the base station, an average, a maximum value, or a minimum value. For example, it may be information limited to connected UEs, information limited to UEs that have performed SDT processing, or a combination of these. It may also include information about the location of the base station. It may also include information about time. Information about the performance of the base station may be associated with information about the location and time. Information about the performance of the base station can be used as input information for training and inference.
[0343] (24) may be appropriately applied to (29) of the example of input information for the AI / ML model for RA processing disclosed in the first embodiment.
[0344] The input information for the AI / ML model for RA processing disclosed above can be current information or predicted information, which can be used as input information for the AI / ML model to derive more effective output information for the future.
[0345] The input information for the AI / ML model for RA processing disclosed above may be associated with a base station, which may be, for example, a base station, a cell, a cell group, a CU, a DU, a TRP, etc. By associating with a base station, it is possible to take into account which base station the input information is for.
[0346] The input information for the RA processing AI / ML model disclosed above may be associated with information about the state of the UE. The information about the UE state may be appropriately applied to the information disclosed above. By associating the input information with the UE state, it is possible to take into account the state of the UE in which the input information is input.
[0347] When performing SDT processing, there is information that only the UE knows. For example, as disclosed in (7) of the input information example, the UL data amount and received power. The UE uses this information to determine whether to perform SDT processing. By using such information as input information for the AI / ML model, the processing performed by the UE can be performed more effectively.
[0348] A base station having an SDT processing AI / ML model inputs the input information for the SDT processing AI / ML model disclosed above to the SDT processing AI / ML model for training or inference, thereby enabling more effective SDT processing to be performed using the SDT processing AI / ML model.
[0349] The method of transmitting input information for the AI / ML model for SDT processing between base stations, between CU and DU, and between UE and base station may be appropriately applied by applying the input information transmission method disclosed in embodiment 1.
[0350] Output information of the AI / ML model for SDT processing is disclosed. Information used in SDT processing may be used as output information. Thirteen examples of output information of the AI / ML model for SDT processing are disclosed below.
[0351] (1) Information about the UE's trajectory. (2) Information about SDT configuration. (3) Priority of SDT configuration. (4) Information about base station candidates that perform SDT processing. (5) Probability of reaching a base station candidate that performs SDT processing. (6) Communication time due to SDT processing. (7) Information about base station resources. (8) Information about UE traffic. (9) Information about RRC_Idle UEs. (10) Information about RRC_Inactive UEs. (11) Information about base station load. (12) Information about propagation delay between the UE and the cell. (13) A combination of (1) to (12).
[0352] For (1), it is advisable to appropriately apply (1) of the example of output information of the AI / ML model for RA processing disclosed in the first embodiment.
[0353] (2) is information about the SDT setting predicted by the AI / ML model. The information about the SDT setting may be the information disclosed in the example of input information (13) for the AI / ML model for SDT processing. It may also include information about the predicted location of the UE. It may also include information about the predicted time. Information about the SDT setting may be associated with information about the location and time. The information about the SDT setting is not limited to one piece, but may be multiple pieces. It may also include information about the prediction accuracy of the information about the SDT setting predicted by the AI / ML model. The UE can use the SDT setting derived by the AI / ML model. This enables more effective SDT processing.
[0354] (3) is the priority of the information regarding the SDT setting indicated by the information disclosed in (2). Priority may be set for the SDT setting. The predicted SDT setting and its priority may be associated. Priority is assigned to each of the multiple output SDT settings. For example, if the SDT timer expires in the first SDT setting, SDT processing may be performed using the SDT setting with the next priority. This enables more flexible and effective SDT processing.
[0355] For (4) and (5), it is preferable to appropriately apply (4) and (5) of the example of output information disclosed in the first embodiment. It is preferable to replace the RA processing described in (4) and (5) of the example of output information of the AI / ML model for RA processing disclosed in the first embodiment with SDT processing.
[0356] (6) is information on the predicted communication time by SDT processing. The predicted communication time by SDT processing may be set for each SDT setting candidate, for example. For example, it may be set for each base station candidate that performs SDT processing. For example, the UE can recognize which base station it is more effective to perform SDT processing for.
[0357] (7) is information on the predicted resource status of the base station. The information on the resource status of the base station may be the information disclosed in the example of input information (14) for the AI / ML model for SDT processing. The resource status of the base station may be the predicted resource status of the candidate base station. For example, a neighboring base station can recognize the predicted resource status of the base station on which SDT processing has been performed.
[0358] (8) is information about predicted UE traffic. The information about UE traffic may be the information disclosed in the example of input information (15) for the AI / ML model for SDT processing. The information about predicted UE traffic may be predicted UE traffic at the candidate base station. For example, a neighboring base station can recognize the predicted UE traffic of the base station where the RA process was performed.
[0359] (9) may be, for example, information about predicted RRC_Idle UEs. The information about RRC_Idle UEs may be the information disclosed in the example of input information for the AI / ML model for SDT processing (16). For example, a neighboring base station can recognize the number of RRC_Idle UEs when SDT processing is performed.
[0360] (10) may be, for example, information about predicted RRC_Inactive UEs. The information about RRC_Inactive UEs may be the information disclosed in the example of input information for the AI / ML model for SDT processing (17). For example, a neighboring base station can recognize the number of RRC_Inactive UEs when SDT processing is performed.
[0361] (11) may be, for example, information about the predicted load of the base station. The information about the load of the base station may be the information disclosed in the example of input information for the AI / ML model for SDT processing (18). For example, the neighboring base station can recognize the load situation of the base station when SDT processing is performed.
[0362] (12) may be appropriately applied to (13) of the example of output information disclosed in embodiment 1. The RA processing described in (13) of the example of output information of the AI / ML model for RA processing disclosed in embodiment 1 may be replaced with SDT processing.
[0363] An offset may be provided to the output information of the AI / ML model for SDT processing. The method of providing an offset to the output information of the AI / ML model for RA processing disclosed in the first embodiment may be applied as appropriate.
[0364] For example, an offset for each UE is set for the UL data volume threshold. For example, an offset for each UE is set for the received power threshold. Not only positive values but also 0 or negative values may be set as the offset value. In this way, it becomes possible to add an offset value for each UE to the SDT setting set for each cell, making it possible to set the SDT for each UE. By setting the offset, the SDT setting for each cell can be used as before.
[0365] The offset value may be used as input information for the AI / ML model for SDT processing, making it possible to obtain an AI / ML model and inference results that take the offset value into account.
[0366] As another method, AI / ML model output information for SDT processing may be provided for each UE. Part or all of the output information disclosed above may be output information for each UE. As disclosed above, output information may be per UE group, rather than per UE. For example, output information for each UE may be set by RRC, and the output information notified by SIB may be updated. Flexible output information setting is possible.
[0367] Information regarding the relationship between the state of the UE and the output information may be provided. The method of providing the information and the method of notifying the information may be appropriately applied to the methods disclosed in the first embodiment.
[0368] The output information of the AI / ML model for SDT processing disclosed above is appropriately notified to the RAN node or UE performing the SDT processing. A node having the AI / ML model may notify the UE of the information via the RAN node to which the UE is most recently connected. A node having the AI / ML model may broadcast the information to UEs under its control. In this way, it is possible to perform more effective SDT processing using the AI / ML model for SDT processing. The notification may include information indicating what the information is. For example, it may include information indicating that it is output information. This makes it easier to distinguish it from input information and feedback information.
[0369] The method of transmitting AI / ML model output information for SDT processing between base stations, between CU and DU, and between UE and base station may be appropriately applied by applying the output information transmission method disclosed in embodiment 1.
[0370] Feedback information from SDT processing using an AI / ML model for SDT processing (hereinafter, this feedback information will be referred to as "SDT processing feedback information using an AI / ML model for SDT processing"). Information on the performance of each node or communication through SDT processing may also be used as feedback information. Fourteen examples of SDT processing feedback information using an AI / ML model for SDT processing are disclosed below.
[0371] (1) Information about UE performance. (2) Information about UE QoE. (3) Information about UE L2 measurements. (4) Information about L2 measurements from the base station. (5) Information about UE SDT processing. (6) Information about UE SDT timer expiration. (7) Information about base station resources. (8) Information about UE traffic. (9) Information about RRC_Idle UEs. (10) Information about RRC_Inactive UEs. (11) Information about base station load. (12) Information about base station performance. (13) Information about propagation delay between the UE and the cell. (14) A combination of (1) to (13).
[0372] (1) may be, for example, (1) in the above example of input information for the AI / ML model for SDT processing.
[0373] (2) may be, for example, (22) in the above example of input information for the AI / ML model for SDT processing.
[0374] (3) may be, for example, (19) in the above example of input information for the AI / ML model for SDT processing.
[0375] (4) may be, for example, (20) in the above example of input information for the AI / ML model for SDT processing.
[0376] (5) may be, for example, (7) or (8) in the above example of input information for the AI / ML model for SDT processing.
[0377] (6) may be, for example, (9) or (10) in the above example of input information for the AI / ML model for SDT processing.
[0378] (7) may be, for example, (14) in the above example of input information for the AI / ML model for SDT processing.
[0379] (8) may be, for example, (15) in the above example of input information for the AI / ML model for SDT processing.
[0380] (9) may be, for example, (16) in the above example of input information for the AI / ML model for SDT processing.
[0381] (10) may be, for example, (17) in the above example of input information for the AI / ML model for SDT processing.
[0382] (11) may be, for example, (18) in the above example of input information for the AI / ML model for SDT processing.
[0383] (12) may be, for example, (23) in the above example of input information for the AI / ML model for SDT processing.
[0384] (13) may be, for example, (24) in the above example of input information for the AI / ML model for SDT processing.
[0385] The feedback information disclosed above may be an increase or decrease value, for example, an increase or decrease value of the traffic count of a UE, or an increase or decrease value of the number of RRC_Inactive UEs.
[0386] The SDT processing node appropriately notifies the node having the SDT processing AI / ML model of the SDT processing feedback information using the SDT processing AI / ML model disclosed above. This allows for evaluation of the SDT processing AI / ML model's reward calculations and allows for updates to the model. This allows for more effective SDT processing.
[0387] The method for transmitting SDT processing feedback information using the AI / ML model for SDT processing between base stations, between CU-DU, and between UE and base station may be appropriately applied to the method for transmitting RA processing feedback information using the AI / ML model for RA processing disclosed in embodiment 1.
[0388] FIG. 17 is a diagram showing an example of a sequence of SDT processing using AI / ML for SDT processing according to the second embodiment. This example shows an example in which a RAN node has an AI / ML model for SDT processing and performs training and inference. The RAN node may be, for example, a base station. Steps common to those in FIG. 15 are assigned the same step numbers, and common explanations will be omitted.
[0389] In steps ST1701 and ST1702, RAN nodes #1 and #2 are provided with an AI / ML model for SDT processing. An AI / ML model for SDT processing may be provided in all RAN nodes capable of executing AI / ML for SDT processing. In step ST1503, the UE is in an RRC-connected state with RAN node #1. RAN node #1 having the AI / ML model for SDT processing transmits an input information request for model training to the UE in step ST1711 and to a neighboring RAN node (RAN node #2 in this example) in step ST1712. Similarly, RAN node #2 having the AI / ML model for SDT processing transmits an input information request for model training to a neighboring RAN node (RAN node #1 in this example) and to a UE (not shown in the figure) connected to RAN node #2 in step ST1713.
[0390] The request may be triggered by receiving a request for input information for the AI / ML model for SDT processing from a neighboring RAN node. As another method, RAN node #1 may send a model training implementation request to a neighboring RAN node. For example, when the RAN node itself decides to implement model training, it may send a model training implementation request to a neighboring RAN node. In this way, model training using the AI / ML model for SDT processing can be implemented not only by RAN node #1 but also by neighboring RAN nodes.
[0391] The request may include information indicating the required information. The request may include information indicating the purpose of the input information for the AI / ML model for SDT processing. For example, the request may include information indicating that it is for SDT processing. For example, the request may include information indicating that it is for model training. This is effective, for example, when the input information for the AI / ML model for SDT processing includes information for model training. For example, when the information held by the UE includes information for model training, information for inference, and information for feedback, the UE can quickly grasp the information that should be notified to the RAN node, and as a result, quick notification from the UE to the RAN node is possible.
[0392] The request may be transmitted between the RAN node and the UE using, for example, RRC signaling. The request for input information for the AI / ML model for the RAN node may be included in a measurement configuration transmitted from the RAN node to the UE, thereby reducing the amount of signaling.
[0393] For example, Xn signaling may be used to transmit the request between RAN nodes.
[0394] In steps ST1714 and ST1716, the UE and RAN node #2 transmit model training input information to RAN node #1. RAN node #2 may transmit model training input information acquired from a connected UE to RAN node #1. RAN node #1 is then able to input information about UEs connected to RAN node #2 into the AI / ML model. In step ST1715, RAN node #1 transmits model training input information to RAN node #2. Also, although not shown in the figure, a UE connected to RAN node #2 transmits model training input information to RAN node #2. RAN node #1 may transmit model training input information acquired from the UE to RAN node #2. RAN node #2 is then able to input information about UEs connected to RAN node #1 into the AI / ML model.
[0395] Measurement results by the UE may be transmitted to the RAN node in a measurement report. The measurement results may be radio measurements, L2 measurements, or MDT measurements. Logged MDT measurements may be measurements made by the UE in RRC_Inactive or RRC_Idle state. These measurements may be transmitted via the same signaling or via different signaling. The UE measurements may be used as input for model training.
[0396] For example, RRC signaling may be used to transmit input information for model training between the RAN node and the UE. For example, a UEInformationResponse message may be used. For example, a MeasurementReport message may be used to transmit radio measurement results and L2 measurement results. Using conventional messages can avoid complex processing.
[0397] For example, Xn signaling may be used to transmit the input information between RAN nodes.
[0398] In steps ST1717 and ST1718, RAN node #1 and RAN node #2 train the SDT processing AI / ML model using input information for the SDT processing AI / ML model held by the own node and input information for the SDT processing AI / ML model acquired from the UE and adjacent RAN nodes. The SDT processing AI / ML model is updated through training. These processes may be performed periodically or each time SDT processing is performed in the own node or in a surrounding RAN node.
[0399] In step ST1521, RAN node #1 determines to perform RRC release for the UE. In steps ST1731 and ST1732, RAN node #1 transmits an inference input information request to the UE and neighboring RAN nodes. RAN node #2, which has an AI / ML model for SDT processing, similarly transmits an inference input information request to neighboring RAN nodes and to a UE (not shown in the figure) connected to RAN node #2 in step ST1733. This request may be triggered by receiving a request for input information for the AI / ML model for SDT processing from a neighboring RAN node. As an alternative method, RAN node #1 may transmit an inference execution request to a neighboring RAN node. For example, RAN node #1, which has determined to perform RRC release for the UE, may transmit an inference execution request to a neighboring RAN node. In this way, inference using the AI / ML model for SDT processing can be performed not only by RAN node #1 but also by neighboring RAN nodes.
[0400] The request may include information indicating the required information. The request may include information indicating the use of the input information for the SDT processing AI / ML model. For example, the request may include information indicating that the input information is for inference. This is effective, for example, when the input information for the SDT processing AI / ML model includes information for inference.
[0401] In steps ST1734 and ST1736, the UE and RAN node #2 transmit inference input information to RAN node #1. RAN node #2 may transmit inference input information acquired from a connected UE to RAN node #1. RAN node #1 becomes able to input information of a UE connected to RAN node #2 into the AI / ML model. In step ST1735, RAN node #1 transmits inference input information to RAN node #2. Also, although not shown in the figure, a UE connected to RAN node #2 transmits inference input information to RAN node #2. RAN node #1 may transmit inference input information acquired from the UE to RAN node #2. RAN node #2 becomes able to input information of a UE connected to RAN node #1 into the AI / ML model.
[0402] The method of transmitting an input information request for an AI / ML model for SDT processing (corresponding to the input information request for model training and the input information request for inference shown in Figure 17) and input information for an AI / ML model for SDT processing (corresponding to the input information for model training and the input information for inference shown in Figure 17) may be any of the methods disclosed above, as appropriate.
[0403] In steps ST1737 and ST1738, RAN nodes #1 and #2 perform inference using the SDT processing AI / ML model, using the SDT processing AI / ML model input information held by the own node and the SDT processing AI / ML model input information acquired from the UE and adjacent base stations. The RAN nodes derive SDT processing AI / ML model output information through this inference.
[0404] In step ST1741, RAN node #1 transmits AI / ML model output information for SDT processing to the UE. For example, UE-specific output information may be transmitted. For example, UE-specific SDT settings may be transmitted. For example, UE-specific offsets to be added to the SDT settings of the cell may be transmitted. Before the UE transitions to the RRC_Idle or RRC_Inactive state, for example, before performing RRC release processing, or during RRC release processing, it is advisable to perform transmission and reception processing of AI / ML model output information for SDT processing. In this way, the UE can acquire UE-specific output information.
[0405] The AI / ML model output information for SDT processing may be transmitted between RAN nodes. RAN node #1, which performed the inference in step ST1737, may transmit output information (inference result) to RAN node #2. RAN node #2, which performed the inference in step ST1738, may transmit output information (inference result) to RAN node #1. The AI / ML model output information for SDT processing can be shared between RAN nodes. RAN node #1 may transmit the output information received from RAN node #2 to the UE. For example, this may be transmitted in step ST1741. The UE can acquire the AI / ML model output information for SDT processing of each RAN node. This is also effective, for example, when the AI / ML model output information for SDT processing of RN node #1 differs from the AI / ML model output information for SDT processing of RAN node #2.
[0406] In Step ST1551, the RAN node #1 performs an RRC release procedure with the UE. The UE transitions to the RRC_Idle or RRC_Inactive state.
[0407] In steps ST1761 and ST1762, RAN nodes #1 and #2 transmit output information derived using the AI / ML model for SDT processing. For example, they transmit SDT configuration. The output information may be broadcast, for example, by being included in an SIB. These processes may be performed before the process of step ST1551. They may be performed after steps ST1737 and ST1738. In this way, RAN node #1 and surrounding RAN nodes can transmit the AI / ML model output information for SDT processing, and the UE can perform SDT processing using the AI / ML model output information for SDT processing.
[0408] The UE determines an RRC connection with a RAN node. The UE may be moving. In step ST1771, the UE evaluates whether or not SDT is possible using the SDT processing AI / ML model output information received from RAN node #2 and / or the SDT processing AI / ML model output information received from RAN node #1. The UE may also determine the RAN node that will perform the SDT processing. The figure illustrates an example in which the UE selects RAN node #2. In this way, SDT processing becomes possible between the UE and RAN node #2 using the output information derived by the SDT processing AI / ML model.
[0409] In step ST1772, the UE performs SDT processing with the RAN node #2. This allows the UE to communicate with the RAN node #2 without transitioning to an RRC connected state. For example, data communication between the UE and the NW becomes possible.
[0410] The UE transmits SDT processing feedback information using the AI / ML model for SDT processing to RAN node #2. The transmission of the feedback information may be performed during SDT processing. The feedback information may include information about the RAN node to which the UE had its last RRC connection. For example, it may include information identifying the RAN node. The RAN node that receives the information can recognize the RAN node to which the UE had its last RRC connection. The feedback information can then be transmitted to the RAN node.
[0411] As another method, the UE may transmit information requesting the transmission of feedback information to RAN node #2, either included in the feedback information or separately. The transmission of the information may be performed in the SDT process. Upon receiving the information, RAN node #2 may transmit feedback information to neighboring RAN nodes. RAN node #2 may transmit feedback information to RAN node #1 using information received from the UE regarding the RAN node to which it was last RRC connected. The UE can request that feedback information be transmitted from RAN node #2 to RAN node #1 as appropriate. Flexible control is possible.
[0412] The feedback information may be transmitted when the UE performs the SDT process, or may be transmitted to a RAN node that performs the SDT process and has an RRC connection. In this way, the RAN node can obtain the feedback information every time the UE performs the SDT process.
[0413] As another method, the RAN node #2 to which the UE is connected using the SDT process may transmit a request for feedback information to the UE. The transmission of the request may be performed in the SDT process. The UE that receives the request may transmit feedback information to the RAN node #2. In this way, the feedback information can be transmitted from the UE to the RAN node #2 as needed. This reduces the amount of signaling.
[0414] RAN node #2, which has received feedback information from the UE in the SDT process in step ST1772, transmits the UE feedback information to RAN node #1. In step ST1782, RAN node #2 transmits the feedback information that its own RAN node has to RAN node #1. In step ST1783, RAN node #1 may transmit the feedback information that its own RAN node has to RAN node #2.
[0415] The method for transmitting the RA processing feedback information request using the AI / ML model for SDT processing (not shown in Figure 17) and the SDT processing feedback information used for the AI / ML model for SDT processing (corresponding to the feedback information shown in Figure 17) may be appropriately applied using the method for input information disclosed above.
[0416] In this way, the RAN node can acquire information on the performance of RA processing using the AI / ML model for SDT processing. The RAN node may evaluate the effectiveness of SDT processing when using the AI / ML model for SDT processing, using the SDT processing feedback information using the AI / ML model for SDT processing that the RAN node holds and the SDT processing feedback information using the AI / ML model for SDT processing that the RAN node acquires from the UE and surrounding RAN nodes.
[0417] There may be no need to request input information or feedback information from the AI / ML model for SDT processing. Each node may transmit input information or feedback information from the AI / ML model for RA processing to a node having an AI / ML model for SDT processing as appropriate. For example, the transmission may be performed periodically. For example, the input information or feedback information may be transmitted when new input information or feedback information is obtained. This reduces the amount of signaling. When a node having an AI / ML model for SDT processing receives the information from another node or from itself, it may transmit an input information request to other nodes excluding the node. This reduces unnecessary signaling, such as transmitting the same input information.
[0418] Alternatively, one of the RAN nodes may perform AI / ML model training. The method disclosed in the first embodiment may be applied as appropriate. This simplifies the processing using the AI / ML model as a system.
[0419] In this way, AI / ML is introduced into the SDT processing, enabling more effective SDT processing.
[0420] By using the method disclosed in this embodiment, it is possible to perform training and inference in SDT processing using an AI / ML model. By deriving output information of the AI / ML model using various types of input information, it is possible to optimize SDT processing. Furthermore, by providing UE-specific SDT settings, it is possible to optimize SDT settings appropriate for the UE's state, such as the UL data volume threshold, the received power threshold, and the SDT timer. This enables more appropriate SDT settings. This allows, for example, a UE in RRC_INACTIVE to use SDT as much as possible, allowing the UE to transmit and receive data quickly. It is possible to optimize various KPIs in SDT processing, such as reducing delay time, improving reliability, and improving user experience.
[0421] As described above, it has been disclosed that the AI / ML model for SDT processing may be provided in a RAN node. As an alternative, the AI / ML model for SDT processing may be provided in a UE. A learning device and an inference device may also be provided in the UE. The UE can derive the configuration information required for SDT processing using the AI / ML model.
[0422] The method of providing the UE with an AI / ML model for SDT processing may be suitably applied to the method of providing the UE with an AI / ML model for RA processing disclosed in the first modification of the first embodiment.
[0423] By doing so, the same effect as described above can be obtained. In addition, since it is possible for the UE to perform inference when starting the SDT process, it is possible to execute a more appropriate SDT process according to the UE's situation. In addition, since the UE performs inference, it is possible to reduce the amount of signaling for inference and simplify the inference process.
[0424] Regarding the node where the AI / ML model for SDT processing is provided, the method of providing it in the RAN node and the method of providing it in the UE may be combined, thereby enabling more effective SDT processing.
[0425] Embodiment 3. A UE in RRC_Idle or RRC_Inactive is not connected to a base station. In communication processes performed by the UE in the RRC_Idle or RRC_Inactive state, such as cell selection and cell reselection, the UE uses information set for each cell. For this reason, the UE cannot perform communication processes suited to the individual state of the UE, and better communication performance cannot be obtained.
[0426] In this embodiment, as a method for solving such a problem, a method for improving communication performance in communication processing when a UE is in RRC_Idle or RRC_Inactive will be disclosed.
[0427] In this embodiment, information provided to or acquired by a UE for communication processing of a UE in RRC_Idle or RRC_Inactive is associated with information regarding the state of the UE. The associated information may be referred to as state association information. The information provided to or acquired by the UE is not limited to one piece of information, but may be multiple pieces of information. The information regarding the state of the UE is not limited to one piece of information, but may be multiple pieces of information. The state association information may be a list.
[0428] The information on the state of the UE may be appropriately applied to the example of input information (1) for the AI / ML model for RA processing disclosed in the first embodiment, such as the position, time, moving speed, and moving direction of the UE.
[0429] Communication processes performed by the UE in RRC_Idle and RRC_Inactive include, for example, cell selection processing, cell reselection processing, DRX processing, RA processing, SDT processing, EDT (Early Data Transmission) processing, Control Plane CIoT 5GS Optimization processing, and User Plane CIoT 5GS Optimization processing.
[0430] The base station transmits state association information to the UE. The UE may be a UE under the base station or may be an individual UE. The UE uses the state association information for communication processing in RRC_Idle and RRC_Inactive. The UE may use the state association information to derive information to be used for communication processing in RRC_Idle and RRC_Inactive. The UE performs communication processing in RRC_Idle and RRC_Inactive using the derived information. In this way, the UE can perform communication processing using information that takes into account the state of its own UE when starting communication processing in RRC_Idle and RRC_Inactive.
[0431] AI / ML may be introduced into the communication processing of a UE in RRC_Idle or RRC_Inactive. Configuration information for the communication processing is derived using AI / ML. The AI / ML model used for the communication processing is referred to as an AI / ML model for communication processing. State association information may be derived using AI / ML. The derived information may be used as AI / ML model output information.
[0432] In this embodiment, an AI / ML model for communication processing is provided in a RAN node. The RAN node may be, for example, a base station, a CU, a DU, an IAB node, an IAB donor, or the like. By providing the AI / ML model in the RAN, for example, communication processing setting information configured by the RAN can be transmitted directly to the UE. This avoids complication of processing. As a method for providing an AI / ML model for communication processing of a UE in RRC_Idle or RRC_Inactive in a RAN node, the method of providing an AI / ML model for RA processing disclosed in the first embodiment in a RAN node or the method of providing an AI / ML model for SDT processing disclosed in the second embodiment in a RAN node may be applied as appropriate. Information specific to each process, such as information regarding each process of input information for the AI / ML model, may be information regarding the individual communication processes described above (for example, the cell selection process, cell reselection process, etc. disclosed above). As another method, the communication processing AI / ML model may be provided in a CN node. The CN node may be, for example, an AMF, an NWDAF, or the like. By providing the AI / ML model in a CN node, for example, it becomes easier to set communication processing configuration information for a UE that moves across multiple RAN nodes. As another method, the communication processing AI / ML model may be provided in a management node. The management node may be, for example, an MnS node or an OAM. By providing the AI / ML model in a management node, for example, it becomes possible to manage and transmit communication processing configuration information in conjunction with management of base stations and UEs. Control combined with other management becomes possible. An example of control combined with other management is mobility robustness optimization. Similarly, the learning device and the inference device may be provided in a RAN node, a CN node, or a management node. Similar effects can be obtained. The above methods may also be combined. Flexible AI / ML operation becomes possible.
[0433] A validity period may be set for the configuration information for communication processing of a UE in RRC_Idle or RRC_Inactive. The validity period may be a deadline. The validity period may be a timer. The validity period is not limited to hours, minutes, and seconds, but may also be a year, month, date, day of the week, etc. Furthermore, it may be information related to time used in the network, such as HFN or SFN. For example, when a UE receives the configuration information, the validity period is activated, and the configuration information is not used when the validity period expires. When new configuration information is notified, the validity period is activated again, and communication processing is performed using the new configuration information. The validity period may be notified to the UE by a base station. For example, the base station may include the configuration information in an SIB and transmit it to a UE being served by it, or may transmit it individually to each UE by RRC signaling. The configuration information for communication processing of a UE in RRC_Idle or RRC_Inactive may take into account the radio wave propagation environment that changes over time, the load on the base station, etc.
[0434] The validity period of the setting information for communication processing of the UE in RRC_Idle or RRC_Inactive may be used as input information, output information, or feedback information of the AI / ML model for communication processing of the UE in RRC_Idle or RRC_Inactive. An AI / ML model can be used that takes into account the radio wave propagation environment that changes over time, the load of the base station, and the like.
[0435] By using the method disclosed in this embodiment, the UE can take into account the state of the UE in communication processing in RRC_Idle or RRC_Inactive. By being able to take into account the state of each UE, communication processing suitable for each UE can be performed. Furthermore, by introducing AI / ML into communication processing in RRC_Idle or RRC_Inactive by the UE, it becomes possible to use the AI / ML model to perform training and inference in communication processing in RRC_Idle or RRC_Inactive, and more effective communication processing can be performed.
[0436] As described above, it has been disclosed that an AI / ML model for communication processing of a UE in RRC_Idle or RRC_Inactive may be provided in a RAN node. As another method, an AI / ML model for communication processing of a UE in RRC_Idle or RRC_Inactive may be provided in the UE. A learning device or an inference device may be provided in the UE. The UE can derive setting information required for communication processing of a UE in RRC_Idle or RRC_Inactive using the AI / ML model.
[0437] The method of providing an AI / ML model for communication processing in a UE in RRC_Idle or RRC_Inactive may be appropriately applied to the method of providing an AI / ML model for RA processing in a UE disclosed in the first modification of the first embodiment.
[0438] By doing so, the same effect as described above can be obtained. Furthermore, since it is possible for the UE to perform inference when starting communication processing of the UE in RRC_Idle or RRC_Inactive, it is possible to execute more appropriate communication processing of the UE in RRC_Idle or RRC_Inactive according to the status of the UE. Furthermore, since the UE performs inference, it is possible to reduce the amount of signaling for inference and simplify the inference processing.
[0439] Regarding the node in which the AI / ML model for communication processing of UEs in RRC_Idle and RRC_Inactive is provided, the method of providing the AI / ML model in the RAN node and the method of providing the AI / ML model in the UE may be combined, thereby enabling more effective communication processing of UEs in RRC_Idle and RRC_Inactive.
[0440] In the above, it has been disclosed that an AI / ML model for communication processing of a UE in RRC_Idle or RRC_Inactive is provided, but an AI / ML model for communication processing may also be provided for each individual communication processing (e.g., the cell selection processing, cell reselection processing, etc. disclosed above) of a UE in RRC_Idle or RRC_Inactive.
[0441] Fourth Embodiment: IAB is supported in 3GPP (see Non-Patent Documents 2, 20, and 29). An IAB node consists of an IAB-MT (Mobile Termination) with terminal functionality and an IAB-DU with DU functionality. The parent IAB-DU of an IAB-MT is selected by the IAB donor CU based on measurements by the IAB-MT. This method only takes into account the radio wave propagation environment between the IAB-MT and the parent IAB-DU. Therefore, depending on the status of IAB nodes on the backhaul (BH) route (path) and the radio wave propagation environment between IAB nodes, the selected parent IAB-DU may not be optimal. For example, even if the radio wave propagation environment between the IAB-MT and the selected parent IAB-DU is the best, if the radio wave propagation environment between the IAB node containing the parent IAB-DU and its parent IAB node is poor, or if the load on one of the IAB nodes on the BH route is high, communication performance may deteriorate, such as a decrease in throughput, when using the BH route.
[0442] In this embodiment, a method for solving such a problem and enabling more effective IAB processing according to the communication situation is disclosed.
[0443] AI / ML is introduced into IAB processing. IAB processing includes, for example, BAP configuration processing and IAB node resource configuration processing. Other examples include IAB node integration processing, IAB-MT setup processing, IAB-DU setup processing, BH RLC channel establishment processing, and routing update processing. AI / ML is used to optimize the information required for IAB processing.
[0444] The methods disclosed in the first embodiment may be applied as appropriate to the AI / ML model, learning device, and inference device. Hereinafter, the AI / ML model used for IAB processing may be referred to as an AI / ML model for IAB processing.
[0445] In this embodiment, an AI / ML model for IAB processing, which is an AI / ML model used for IAB processing, is provided in a base station. A learning device and an inference device may also be provided in a base station. The AI / ML model for IAB processing may also be provided in a RAN node. For example, it may be provided in a CU or a DU. The AI / ML model for IAB processing may also be provided in an IAB donor. For example, it may be provided in an IAB donor CU or an IAB donor DU. Configuration information required for IAB processing can be derived by the base station and each node using the AI / ML model, making configuration easier.
[0446] Regarding the information on whether or not an AI / ML model for IAB processing is available, the method disclosed in the first embodiment may be applied as appropriate.
[0447] Input information for the IAB processing AI / ML model is provided. Twenty-two examples of input information for the IAB processing AI / ML model are disclosed below.
[0448] (1) Information about the status of the IAB node. (2) Information about the trajectory of the IAB node. (3) Information about the IAB node to which the IAB node connects. (4) Information about the IAB donor to which the IAB node connects. (5) Information about radio measurements of the IAB node. (6) Information about the MDT of the IAB node. (7) Information about the BAP settings of the IAB node. (8) Information about the BH link. (9) Information about resource settings of the IAB node. (10) Information about resource usage of the IAB node. (11) Information about UE traffic of the IAB node. (12) Information about the load of the IAB node. (13) Information about UE-NW communication using the IAB node. (14) Information about resource collisions at the IAB node. (15) Information about communication quality of the IAB node. (16) Information about L2 (Layer 2) measurements of the IAB node. (17) Information about L2 (Layer 2) measurements of the IAB donor. (18) Information about L2 (Layer 2) measurements of the UE passing through the IAB node. (19) Information about the performance of the UE passing through the IAB node. (20) Information about the QoE of the UE passing through the IAB node. (21) Information about the propagation delay between IAB nodes. (22) A combination of (1) to (21).
[0449] (1) may be, for example, the location, movement speed, and movement direction of the IAB node. The location of the IAB node may be, for example, coordinates. It may be, for example, an area. It may be, for example, information identifying the TA, RNA, base station, cell, CU, DU, and TRP in which the UE is located. It may be, for example, information identifying the TA, RNA, base station, cell, CU, DU, and TRP with the best reception quality (which may be reception power). It may also be information regarding time. These pieces of information may be associated with the time. It may also be the remaining battery level of the IAB node. It may also be information regarding the power consumption of the IAB node. These pieces of information may be combined. The state of the IAB node can be used as input information for training and inference.
[0450] In (2), the information regarding the trajectory of the IAB node may be, for example, the information regarding the state of the IAB node disclosed in (1), as appropriate. For example, it may be information regarding the connecting IAB node disclosed in (3). For example, it may be information regarding the connecting IAB donor disclosed in (4). It may be information regarding the past and current trajectories of the IAB node. The information regarding the past and current trajectories of the IAB node may be historical information. The information regarding the trajectory may be information regarding the future trajectory of the IAB node. The information regarding the future trajectory may be predicted information. These pieces of information may be presented as a list. Information regarding time may also be included. Information regarding the trajectory of the IAB node may be associated with information regarding time. Information regarding the trajectory of the IAB node can be used as input information for training or inference.
[0451] (3) may be, for example, information about IAB nodes to which an IAB node has previously connected. The IAB node may be a parent IAB node or a child IAB node. The IAB node information may be, for example, information that identifies the IAB node. For example, it may be configuration information about IAB nodes to which the IAB node has previously connected. For example, it may be history information about IAB nodes to which the IAB node has previously connected. For example, it may be a list of information about the IAB nodes. It may also include information about the location of the IAB node. It may also include information about time. Information about previously connected IAB nodes may be associated with information about location and time. The IAB nodes to which an IAB node has previously connected may be limited to information about IAB nodes connected through IAB processing.
[0452] (4) may be, for example, information about an IAB donor to which an IAB node is connected. The connection is not limited to a direct connection, but may also be an indirect connection. For example, it may be information about an IAB donor to which an IAB node is connected via another IAB node. The IAB donor may be an IAB donor CU or an IAB donor DU. The IAB donor information may be, for example, information identifying the IAB donor. For example, it may be configuration information about an IAB donor to which the IAB node has previously connected. For example, it may be history information about IAB donors to which the IAB node has previously connected. For example, it may be a list of information about the IAB donors. It may also include information about the location of the IAB node. It may also include information about time. The information about IAB donors to which an IAB node has previously connected may be associated with information about the location and time of the IAB node. The IAB donors to which an IAB node has previously connected may be limited to information about IAB donors to which the IAB node has previously connected through IAB processing.
[0453] (5) is, for example, a radio measurement result by an IAB node. The IAB node may be an IAB-MT. The radio measurement result may be, for example, RSRP, RSRQ, or SINR. It may include information about the location of the IAB node. It may also include information about time. The radio measurement result may be associated with information about the location or time. The reception quality depending on the location or state of the IAB node can be used as input information for training or inference.
[0454] (6) is, for example, the result of measurement by the IAB node using MDT. The IAB node may be an IAB-MT. The IAB node may perform measurements using MDT. For example, measurements may be performed using immediate MDT or logged MDT. Measurements may be performed using management-based MDT. Measurements may be performed using signaling-based MDT. The IAB node may support a trace function. Measurements may be performed using the trace function. The IAB node may associate the MDT measurement results with information about the IAB node's location and time. The information about the IAB node's location and time may be information measured by MDT. By configuring MDT for the IAB node, the MDT measurement results can be obtained as input information for the AI / ML model for IAB processing. More states and more information about the IAB node can be used as input information for training and inference.
[0455] (7) is, for example, BAP setting information set in the IAB node. For example, it is a BAP mapping setting, a BH RLC channel setting, etc. The BAP mapping setting is, for example, information about BH routing, information about traffic mapping, etc. The information about BH routing is, for example, BAP routing information, such as a BAP address and a path identifier. The information about traffic mapping is, for example, information about mapping between the BAP routing information and the BH RLC channel. Information about the number of IAB nodes on the BH route may be provided. This information may be for each BH route. This information may be included in the information about the BAP setting. This information may be included in the information about the BH routing. The information about the BAP setting may be set by the IAB donor CU. The BAP setting information set as an IAB node can be used as input information for training and inference.
[0456] (8) is, for example, the number of BH links set in the IAB node. For example, it may be information identifying the BH links. For example, it may be information indicating the number of BH links in use, the number of BH links not in use, which BH links are in use, which BH links are not in use, etc. Information regarding the BH links in the IAB node can be used as input information for training and inference.
[0457] (9) is resource configuration information configured in the IAB node. For example, it is information about cell configuration, information about activated cells, information about deactivated cells, information about child IAB nodes, information about cells of adjacent nodes, information about serving cells, etc. The resource configuration information may be configured from the IAB donor CU. The resource configuration information configured in the IAB node can be used as input information for training and inference.
[0458] (10) is, for example, the resource usage of an IAB node. It may be the current resource usage or the predicted resource usage. For example, it may be the resource usage per IAB node, per cell group, per cell, per DU, or per TRP. The resource may be a frequency resource or a time resource. The frequency resource usage may be the number of carriers, the number of RBs, the number of subcarriers, etc. For example, it may be the BWP usage. It may be per carrier, per PRB, or per subcarrier. The time resource usage may be the number of subframes, the number of slots, the number of symbols, etc. It may be per subframe, per slot, or per symbol. The amount of base station resources being used can be used as input information for training or inference.
[0459] (11) is, for example, UE traffic of an IAB node. It may be current traffic or predicted traffic. UE traffic may be, for example, the number of UEs connecting to an IAB node, or the number of UEs using an IAB node. For example, it may be the number of UEs connected or using each IAB node, each cell group, each cell, each DU, or each TRP. The number of UEs connected to an IAB node can be used as input information for training or inference.
[0460] (12) is load information of the IAB node. It may be current load information or predicted load information. The load information may be, for example, the number of DRBs configured in the IAB node or buffer usage. For example, it may be the number of child IAB nodes or the number of parent IAB nodes. For example, it may be the number of BH routes including the own IAB node. Load information other than the information on resource usage and information on UE traffic disclosed above may also be used. Information such as the degree of load on the IAB node can be used as input information for training or inference.
[0461] (13) is, for example, information regarding UE-NW communication using an IAB node. For example, it may be the number of UE-NW communications. For example, it may be setting information for UE-NW communication. For example, it may be information regarding PDU sessions, information regarding bearers, information regarding slices, etc. For example, it may be the number of PDU sessions, PDU session setting information, information identifying PDU sessions, the number of bearers, bearer setting information, information identifying bearers, the number of slices, slice setting information, information identifying slices, etc. The communication settings when a UE communicates with a NW via an IAB node can be used as input information for training or inference.
[0462] (14) is, for example, information regarding collisions between resources for a child IAB node and resources for a parent IAB node. It is collision information between resources used for communication with a child IAB node in an IAB-DU and resources used for communication between an IAB-MT and a parent IAB node (which may be an IAB donor). The resource collision information may be, for example, information regarding interference. The resource may be the information disclosed in (9) or (10). For example, information regarding collisions between resources scheduled by or scheduled by an IAB node can be used as input information for training or inference.
[0463] (15) is, for example, information regarding the communication quality between an IAB node and a parent IAB node. For example, it may be information regarding the communication quality between an IAB node and a child IAB node. It may also be information regarding the communication quality between an IA node and an IAB donor. For example, it may be information regarding the communication quality between an IAB-MT and an IAB donor CU. For example, it may be information regarding the communication quality between an IAB-MT and an IAB-DU of a parent IAB node. For example, it may be information regarding the communication quality between an IAB-DU and an IAB donor CU. For example, it may be information regarding the communication quality between an IAB-DU and an IAB donor CU. For example, it may be information regarding the communication quality between an IAB-DU and an IAB-MT of a child IAB node. The communication quality may be, for example, a packet loss rate or a delay time. For example, it may be information regarding data communication quality. For example, it may be information regarding the reception quality of an SSB transmitted from an IAB-DU by an IAB-MT. For example, it may be information regarding the reception quality of a PDCCH and a PDSCH transmitted from an IAB-DU by an IAB-MT. The reception quality may be RSRP, RSRQ, SINR, or BER. It may also include information about the location of the IAB node. It may also include information about time. Information about the communication quality of the IAB node may be associated with information about location and time. Information about the communication quality between the IAB node and other IAB nodes or IAB donors can be used as input information for training or inference.
[0464] (16) may be, for example, a measurement result by an IAB node of L2 measurement information. It may be a current L2 measurement result or a predicted L2 measurement result. The L2 measurement information may be information measured by a UE as disclosed in the examples of input information (25) and (26) for the AI / ML model for RA processing in the first embodiment, or information measured by a base station. The L2 measurement result by an IAB node may be an L2 measurement result when connected to an IAB donor, parent IAB node, or child IAB node through IAB processing. It may also include information about the location of the IAB node. It may also include information about time. Information about the L2 measurement by the IAB node may be associated with information about the location or time. Information about communication performance at an IAB node can be used as input information for training or inference.
[0465] (17) may be, for example, a measurement result by the IAB donor of the L2 measurement information. It may be a current L2 measurement result or a predicted L2 measurement result. The L2 measurement information may be information measured by the base station as disclosed in the example of input information (26) for the AI / ML model for RA processing in embodiment 1. It may also include information about the location of the IAB donor. It may also include information about time. Information about the L2 measurement by the IAB donor may be associated with information about the location or time. Information about communication performance at the IAB donor can be used as input information for training and inference.
[0466] (18) may be, for example, a measurement result of L2 measurement information measured by a UE via an IAB node. It may be a current L2 measurement result or a predicted L2 measurement result. The L2 measurement information may be, for example, information measured by a UE as disclosed in the example of input information (25) for the AI / ML model for RA processing in embodiment 1. For example, it may be a packet delay amount. The L2 measurement result by the UE may be an L2 measurement result by a UE connected to a NW via an IAB node by IAB processing. It may also include information about the location of the UE. It may also include information about time. Information about the L2 measurement by the UE may be associated with information about the location or time. Information about communication performance when a UE communicates with a NW via an IAB node can be used as input information for training or inference.
[0467] (19) may be, for example, the QoS of the UE. It may be a packet loss rate or a delay time. It may be, for example, data communication quality. It may be, for example, the reception quality of SSB by the UE. It may be, for example, the reception quality of PDCCH and PDSCH by the UE. The reception quality may be RSRP, RSRQ, SINR, or BER. The information about the UE's performance may be information after connecting to the NW via an IAB node by IAB processing. It may also be information about the performance required of the UE. It may also include information about the UE's location. It may also include information about time. The information about the UE's performance may be associated with information about the location and time. The information about the UE's performance can be used as input information for training or inference.
[0468] (20) may be, for example, a QoE measurement result. It may be information after connection to the NW via an IAB node by IAB processing. It may be information about the QoE required by the UE. It may also include information about the location of the UE. It may also include information about time. Information about QoE may be associated with information about the location and time of the UE. Measurement information at the UE application layer can be used as input information for training and inference.
[0469] (21) may be, for example, propagation delay information between an IAB node and a parent IAB node or a child IAB node. It is not limited to information between IAB nodes, but may also be information about propagation delay between an IAB node and an IAB donor DU, or between an IAB node and a UE. The information about propagation delay may be information about propagation delay disclosed in the example of input information for the AI / ML model for RA processing (29). For example, it may be information measured by the IAB-MT or by the IAB-DU. For example, it may be information measured by the UE or by the IAB donor DU. The information about propagation delay between IAB nodes may be information derived by the IAB donor CU. This makes it possible to obtain more accurate distances between IAB nodes and positions of IAB nodes.
[0470] The IAB node disclosed above may be an IAB node targeted for IAB processing. Information about the IAB node can be used as input information for IAB processing using AI / ML. The IAB node may be some or all of the IAB nodes in a BH route. Information about the IAB nodes constituting the BH route can be used as input information for training and inference. The IAB node may be an IAB node adjacent to the IAB node targeted for IAB processing. Information about not only the target IAB node but also adjacent IAB nodes can be used as input information for training and inference for IAB processing. The IAB node may be some or all of the IAB nodes in a PLMN. Information about IAB nodes in a PLMN that supports IAB can be used as input information for training and inference.
[0471] The input information disclosed above can be current information or predicted information, which can be used as input to the AI / ML model to derive more effective output information in the future.
[0472] The input information disclosed above may be associated with information about the state of the IAB node. The information about the state of the IAB node may be appropriately applied to the information disclosed above. By associating the input information with the state of the IAB node, it is possible to take into account the state of the IAB node in which the input information is received.
[0473] The input information disclosed above may be history information. IAB node history information may be provided. IAB-MT history information, IAB-DU history information, IAB donor history information, IAB donor DU history information, and IAB donor CU history information may be provided. Related input information may be included in this history information. This history information may be included in UE history information. It is preferable to include information on whether or not the node is an IAB node and information for identifying the IAB node. This makes it possible to identify the IAB node and the UE. In this way, information related to IAB nodes can be managed together, thereby simplifying processing. Furthermore, the method of including information in UE history information allows the use of conventional processing related to UE history, thereby avoiding complication of processing.
[0474] Input information for the AI / ML model for IAB processing may be transmitted prior to model training and inference.
[0475] An IAB node transmits input information for the AI / ML model for IAB processing to another IAB node. An IAB node may transmit input information for the AI / ML model for IAB processing to an IAB node having an AI / ML model for IAB processing. An IAB node having an AI / ML model for IAB processing may transmit a request for input information for the AI / ML model for IAB processing to another IAB node. The request may include information about the IAB node that is the target of IAB processing. The information about the IAB node may be an identifier of the IAB node. An IAB node that receives the request transmits input information for the AI / ML model for IAB processing to the IAB node that originated the request. RRC signaling may be used to transmit input information for the AI / ML model for IAB processing or a request for such information between IAB nodes. For example, an RRC message may be provided for this transmission. For example, the current information and the predicted information may be sent in separate messages.
[0476] An IAB donor may transmit input information for the AI / ML model for IAB processing to another IAB donor. An IAB donor may be an IAB donor CU. An IAB donor having an AI / ML model for IAB processing may transmit input information for the AI / ML model for IAB processing to an IAB donor. An IAB donor having an AI / ML model for IAB processing may transmit a request for input information for the AI / ML model for IAB processing to another IAB donor. The request may include information about the IAB donor to be processed. The information about the IAB donor may be an IAB donor identifier. An IAB donor receiving the request transmits input information for the AI / ML model for IAB processing to the requesting IAB donor. Inter-base station signaling may be used to transmit input information for the AI / ML model for IAB processing or a request for such information between IAB donors. For example, Xn signaling may be used, for example, an Xn message may be provided for the transmission, or for example, the current information and the predicted information may be transmitted in separate messages.
[0477] The IAB-DU (which may be an IAB donor DU) may transmit input information for the AI / ML model for IAB processing to the IAB donor CU. The IAB donor CU having the AI / ML model for IAB processing may transmit input information for the AI / ML model for IAB processing. The IAB donor CU having the AI / ML model for IAB processing may transmit a request for input information for the AI / ML model for IAB processing to the IAB-DU. The request may include information about the IAB-DU that is the target of IAB processing. The information about the IAB-DU may be an identifier of the IAB node or an identifier of the IAB-DU. The IAB-DU that receives the request transmits input information for the AI / ML model for IAB processing to the requesting IAB donor CU. DU-to-CU signaling may be used to transmit input information for the AI / ML model for IAB processing and requests for such information between the IAB-DU and the IAB donor CU. For example, F1 signaling may be used. For example, F1 messages may be provided for such transmission. For example, current and predicted information may be transmitted in separate messages.
[0478] The IAB-MT may transmit input information for the AI / ML model for IAB processing to the IAB donor CU. The input information for the AI / ML model for IAB processing may be transmitted to an IAB donor CU having an AI / ML model for IAB processing. An IAB donor CU having an AI / ML model for IAB processing may transmit a request for input information for the AI / ML model for IAB processing to the IAB-MT. The request may include information about the IAB-MT that is the target of IAB processing. The information about the IAB-MT may be an identifier of the IAB node or an identifier of the IAB-MT. Upon receiving the request, the IAB-MT transmits the input information for the AI / ML model for IAB processing to the IAB donor CU that made the request. RRC signaling may be used to transmit input information for the AI / ML model for IAB processing and requests for such information between the IAB-MT and the IAB donor CU. For example, RRC messages may be provided for such transmission.
[0479] Output information of the AI / ML model for IAB processing is disclosed. Information used in IAB processing may be used as output information. Thirteen examples of output information of the AI / ML model for IAB processing are disclosed below.
[0480] (1) Information about the trajectory of IAB nodes. (2) Information about BAP settings of IAB nodes. (3) Information about BH links. (4) Information about resource settings of IAB nodes. (5) Setting priority. (6) Information about candidate IAB nodes to connect to. (7) Probability of reaching candidate IAB nodes to connect to. (8) Information about resource usage of IAB nodes. (9) Information about UE traffic of IAB nodes. (10) Information about load of IAB nodes. (11) Information about resource collisions at IAB nodes. (12) Information about propagation delays between IAB nodes. (13) A combination of (1) to (12).
[0481] (1) may be, for example, predicted future IAB node trajectory information. The information on the IAB node trajectory may be the information disclosed in example (2) of input information for an AI / ML model for IAB processing. Information on past and current trajectories may be included. Along with the predicted future IAB node trajectory information, history information on past and current IAB node trajectories may be included. This information may be presented as a list. Information on time may also be included. Information on the IAB node trajectory and information on time may be associated with each other.
[0482] (2) is information about the BAP setting predicted by the AI / ML model. The information about the BAP setting may be the information disclosed in the example of input information (7) for the AI / ML model for IAB processing. It may also include information about the predicted location of the IAB node. It may also include information about the predicted time. The information about the BAP setting may be associated with information about the location and time. The information about the BAP setting is not limited to one, but may be multiple. It may also include information about the prediction accuracy of the information about the BAP setting predicted by the AI / ML model. The IAB node can use the BAP setting derived by the AI / ML model. This enables more effective IAB processing.
[0483] (3) is information about the BH link predicted by the AI / ML model. The information about the BH link may be the information disclosed in example (8) of input information for the AI / ML model for IAB processing. It may also include information about the position of the predicted IAB node. It may also include information about the predicted time. The information about the BH link may be associated with information about the position or time. The information about the BH link is not limited to one, and may be multiple. It may also include information about the prediction accuracy of the information about the BH link predicted by the AI / ML model. The IAB node can use the BH link derived by the AI / ML model. This enables more effective IAB processing.
[0484] (4) is information about resource configuration predicted by the AI / ML model. The information about resource configuration may be the information disclosed in example (9) of input information for the AI / ML model for IAB processing. It may also include information about the predicted location of the IAB node. It may also include information about the predicted time. The information about resource configuration may be associated with information about location and time. The information about resource configuration is not limited to one piece, but may be multiple pieces. It may also include information about the prediction accuracy of the information about resource configuration predicted by the AI / ML model. The IAB node can use the resource configuration derived by the AI / ML model. This enables more effective IAB processing.
[0485] (5) is the priority of the information related to the settings indicated by the information disclosed in (2) to (4). Priorities may be set for some or all of this information. A predicted setting and its priority may be associated. A priority is assigned to each of the multiple output settings. For example, if a communication interruption occurs with the BH in the first setting, IAB processing may be performed using the setting with the next priority. This enables more flexible and effective IAB processing.
[0486] (6) is information about predicted IAB node candidates to connect to. IAB node candidates to connect to may be provided. The number of IAB node candidates is not limited to one, and may be multiple. Information about the predicted accuracy of the predicted IAB node candidates to connect to may also be included. For example, an IAB node can recognize which IAB node it is more effective to connect to.
[0487] (7) is information about the predicted probability of reaching a candidate IAB node to connect to. It may also include information about the confidence interval of the probability of reaching a candidate IAB node predicted by the AI / ML model. For example, an IAB node can recognize which IAB node it is more effective to connect to.
[0488] (8) is information on the predicted resource usage of the connecting IAB node. The information on the resource usage of the IAB node may be the information disclosed in the example of input information for the AI / ML model for IAB processing (10). The resource usage of the IAB node may be the predicted resource usage of the candidate IAB node. For example, a neighboring IAB node can recognize the predicted resource usage of the IAB node on which IAB processing has been performed.
[0489] (9) is information about predicted UE traffic. The information about UE traffic may be the information disclosed in the example of input information for the AI / ML model for IAB processing (11). The information about predicted UE traffic may be predicted UE traffic at the IAB node candidate. For example, a neighboring IAB node can recognize the predicted UE traffic of the IAB node where the IAB processing was performed.
[0490] (10) may be, for example, information about the predicted load of the IAB node to be connected. The information about the load of the IAB node may be the information disclosed in the example of input information for the AI / ML model for IAB processing (12). For example, the surrounding IAB nodes can recognize the load status of the IAB node when IAB processing is performed.
[0491] (11) is information about the predicted resource collision of the connecting IAB node. The information about the resource collision of the IAB node may be the information disclosed in the example of input information for the AI / ML model for IAB processing (14). The information about the resource collision of the IAB node may be information about the predicted resource collision of the candidate IAB node. For example, a neighboring IAB node can recognize information about the predicted resource collision of the IAB node for which IAB processing has been performed.
[0492] (12) may be, for example, information about predicted future propagation delays between IAB nodes. The information about propagation delays between IAB nodes may be the information disclosed in the example of input information (21) for the AI / ML model for IAB processing. Information about past and present propagation delays between IAB nodes may be included. Along with information about predicted future propagation delays between IAB nodes, historical information about propagation delays between past and present IAB nodes may be included. This information may be organized as a list. Information about time may also be included. Information about propagation delays between IAB nodes may be associated with information about time. For example, an IAB node may recognize information about predicted propagation delays of IAB nodes for which IAB processing has been performed.
[0493] The output information of the AI / ML model for IAB processing disclosed above is appropriately notified to the IAB node and IAB donor DU that perform the IAB processing. In this way, it becomes possible to perform more effective IAB processing using the AI / ML model for IAB processing.
[0494] The method for transmitting the AI / ML model output information for IAB processing may be the same as the method for transmitting the setting information by IAB processing.
[0495] An IAB donor CU may transmit output information of an AI / ML model for IAB processing to an IAB node (which may be an IAB donor DU). An IAB donor CU having an AI / ML model for IAB processing may transmit AI / ML model output information for IAB processing to an IAB donor DU or IAB node involved in the IAB processing. An IAB donor DU or IAB node involved in the IAB processing may transmit a request for AI / ML model output information for IAB processing to an IAB donor CU having an AI / ML model for IAB processing. The IAB donor CU receiving the request transmits AI / ML model output information for IAB processing to the IAB donor DU or IAB node that issued the request. The request or the output information may include information about the IAB donor DU or IAB node that is the target of the IAB processing. This information may be an identifier. DU-CU signaling may be used to transmit output information of the AI / ML model for IAB processing or a request for that information between the IAB donor CU and the IAB donor DU, or between the IAB-DU in the IAB node. For example, F1 signaling may be used. For example, F1 messages may be provided for this transmission. RRC signaling may be used to transmit output information of the AI / ML model for IAB processing or a request for that information between the IAB donor CU and the IAB-MT in the IAB node. For example, RRC messages may be provided for this transmission.
[0496] Feedback information based on IAB processing using the AI / ML model for IAB processing (hereinafter, this feedback information is referred to as "IAB processing feedback information using the AI / ML model for IAB processing") is disclosed. The feedback information includes information on the performance of UE-NW communication, information on the performance of IAB nodes, and information on the performance of IAB donors. The IAB node performance and IAB donor performance may be the performance of the IAB node or IAB donor used in UE-NW communication. Fifteen examples of IAB processing feedback information using the AI / ML model for IAB processing are disclosed below.
[0497] (1) Information regarding L2 (Layer 2) measurements of UEs passing through an IAB node. (2) Information regarding the performance of UEs passing through an IAB node. (3) Information regarding the QoE of UEs passing through an IAB node. (4) Information regarding resource usage of an IAB node. (5) Information regarding UE traffic at an IAB node. (6) Information regarding the load of an IAB node. (7) Information regarding UE-NW communications using an IAB node. (8) Information regarding resource collisions at an IAB node. (9) Information regarding communication quality of an IAB node. (10) Information regarding L2 (Layer 2) measurements of an IAB node. (11) Information regarding L2 (Layer 2) measurements of an IAB donor. (12) Information regarding IAB nodes to which an IAB node connects. (13) Information regarding IAB donors to which an IAB node connects. (14) Information regarding propagation delays between IAB nodes. (15) A combination of (1) to (14).
[0498] (1) to (3) may be, for example, (18) to (20) in the above example of input information for the AI / ML model for IAB processing.
[0499] (4) to (11) may be, for example, (10) to (17) in the above example of input information for the AI / ML model for IAB processing.
[0500] (12) and (13) may be, for example, (3) and (4) in the above example of input information for the AI / ML model for IAB processing.
[0501] (14) may be, for example, (21) in the above example of input information for the AI / ML model for IAB processing.
[0502] The IAB processing node and UE appropriately notify the node having the IAB processing AI / ML model of the IAB processing feedback information using the IAB processing AI / ML model disclosed above. This enables evaluation of the IAB processing AI / ML model, such as reward calculation, and enables updating of the model. This enables more effective IAB processing.
[0503] The method for transmitting IAB processing feedback information using the AI / ML model for IAB processing between base stations, between CU-DU, and between UE and base station may be appropriately applied using the input information transmission method disclosed above.
[0504] 18 is a diagram showing an example of a sequence of IAB processing using an AI / ML for IAB processing according to the fourth embodiment. This figure discloses an example in which an IAB donor CU has an AI / ML model for IAB processing and performs training and inference.
[0505] In step ST1801, an AI / ML model for IAB processing is provided in the IAB donor CU. In steps ST1811 to ST1813, the IAB donor CU having the AI / ML model for IAB processing transmits input information requests for model training to the IAB donor DU, IAB node #2, and IAB node #1. The IAB donor DU, IAB node #1, and IAB node #2 may be limited to those under the umbrella of the IAB donor CU.
[0506] The request may be triggered by an IAB process. For example, the request may be triggered by an IAB node integration process from an IAB node. For example, the request may be triggered by a BAP configuration update process, an IAB node resource configuration update process, or a BH routing update process for an IAB node. Model training can be performed using input information when a new IAB node is added or when the configuration is updated.
[0507] The request may include information indicating the required information. The request may include information indicating the purpose of the input information for the AI / ML model for IAB processing. For example, the request may include information indicating that it is for IAB processing. For example, the request may include information indicating that it is for model training. This is effective, for example, when the input information for the AI / ML model for IAB processing includes information for model training. For example, when the information held by the IAB node includes information for model training, information for inference, and information for feedback, the IAB node can quickly grasp the information that should be notified to the IAB donor CU, thereby enabling the IAB node to quickly notify the IAB donor CU.
[0508] The request may be transmitted between the IAB donor CU and the IAB donor DU or the IAB node, for example, using F1 signaling. The request may be transmitted between the IAB donor CU and the IAB node, for example, using RRC signaling. For example, the request for input information may be included in the measurement configuration transmitted from the IAB donor CU to the IAB node. This reduces the amount of signaling.
[0509] In steps ST1814 to ST1816, IAB node #1, IAB node #2, and IAB donor DU transmit model training input information to the IAB donor CU. The IAB nodes and IAB donor DU may transmit model training input information acquired from the connected UEs to the IAB donor CU. The IAB donor CU can input information about the UEs connected to the IAB nodes and IAB donor DUs into the AI / ML model.
[0510] For example, F1 signaling may be used to transmit input information between the IAB donor DU or IAB node and the IAB donor CU. For example, RRC signaling may be used to transmit input information between the IAB node and the IAB donor CU.
[0511] Measurements by IAB nodes may be sent to the IAB donor CU in measurement reports. These measurements may be radio measurements or L2 measurements. These measurements may be sent using the same signaling or different signaling. Measurements by IAB nodes may be used as input for model training.
[0512] In step ST1817, the IAB donor CU trains the IAB processing AI / ML model using the input information for the IAB processing AI / ML model held by its own node and the input information for the IAB processing AI / ML model acquired from the IAB node and the IAB donor DU. The IAB processing AI / ML model is updated through training. These processes may be performed as appropriate. They may be performed periodically or each time IAB processing is performed. By performing these processes periodically, for example, it becomes possible to appropriately input time-varying load fluctuations at the IAB node as input information. By performing these processes each time IAB processing is performed, it becomes possible to appropriately input input information, for example, when a new IAB node is added or when settings are updated. Model training of a more optimal IAB processing AI / ML model can be performed.
[0513] In steps ST1821 to ST1823, the IAB donor CU transmits a request for input information for inference to the IAB donor DU, IAB node #2, and IAB node #1. The IAB donor DU, IAB node #1, and IAB node #2 may be limited to those under the umbrella of the IAB donor CU. The request may be triggered by IAB processing. The request for input information for inference allows the input information for inference to be acquired and input to the AI / ML model for IAB processing, and output information for IAB processing can be derived.
[0514] The request may include information indicating the required information. The request may include information indicating the use of the input information for the IAB processing AI / ML model. For example, the request may include information indicating that the input information is for inference. This is effective, for example, when the input information for the IAB processing AI / ML model includes information for inference.
[0515] In steps ST1824 to ST1826, IAB node #1, IAB node #2, and IAB donor DU transmit inference input information to the IAB donor CU. The IAB nodes and IAB donor DU may transmit inference input information acquired from the UEs connected to them to the IAB donor CU. The IAB donor CU can input information about the UEs connected to the IAB nodes and IAB donor DUs into the AI / ML model.
[0516] The method for requesting input information for inference and transmitting input information for inference may be the same as the method for requesting input information for model training and transmitting input information for model training disclosed above, as appropriate.
[0517] In step ST1827, the IAB donor CU performs inference using the IAB processing AI / ML model, using the IAB processing AI / ML model input information held by its own node and the IAB processing AI / ML model input information acquired from the IAB node and the IAB donor DU. The IAB donor CU derives IAB processing AI / ML model output information through this inference.
[0518] In step ST1831, IAB processing is performed among the IAB donor CU, IAB donor DU, IAB node #2, and IAB node #1. For example, IAB node integration processing is performed for IAB node #1. In the IAB processing, the IAB donor CU transmits AI / ML model output information for IAB processing to the IAB donor DU, IAB node #2, and IAB node #1. For example, it transmits BAP setting information and IAB node resource setting information to IAB node #1. It may also transmit output information to IAB node #2 and the IAB donor DU to update IAB node #2 and the IAB donor DU. In this way, the IAB nodes and IAB donor DU used in the IAB processing can obtain the AI / ML model output information for IAB processing. Using this output information, IAB processing can be performed between the IAB donor CU, IAB donor DU, IAB node #2, and IAB node #1.
[0519] The IAB donor CU may transmit the AI / ML model output information for IAB processing to the IAB donor DU, IAB node #2, and IAB node #1 separately from the IAB processing. For example, the transmission may be performed before the IAB processing. For example, the transmission may be performed after the output information is derived using the AI / ML model for IAB processing. The IAB processing may be performed after the transmission of the AI / ML model output information for IAB processing.
[0520] The method for transmitting output information may be appropriately applied to the method for transmitting setting information by IAB processing or the method for transmitting an input information request for an AI / ML model for IAB processing disclosed above.
[0521] In steps ST1851 to ST1853, IAB node #1, IAB node #2, and IAB donor DU transmit IAB processing feedback information using the AI / ML model for IAB processing to the IAB donor CU. The feedback information may include information identifying the IAB node or IAB donor DU itself. The IAB donor CU that receives this information can recognize the node that transmitted the feedback information.
[0522] In this way, the IAB donor CU can obtain information regarding the performance of IAB processing using the AI / ML model for IAB processing. The IAB donor CU can evaluate the effectiveness of IAB processing when using the AI / ML model for IAB processing by using the IAB processing feedback information using the AI / ML model for IAB processing that its own node holds and the IAB processing feedback information using the AI / ML model for IAB processing that it obtains from IAB node #1, IAB node #2, and the IAB donor DU.
[0523] The method for transmitting feedback information may be the same as the method for transmitting input information for model training disclosed above.
[0524] Communication between the UE and the NW may be performed using the IAB node set in Step ST1831. In Step ST1861, the UE performs an RRC connection process with the IAB donor CU via IAB node #1, IAB node #2, and the IAB donor DU. Communication between the UE and the NW is performed using this RRC connection.
[0525] The feedback information may be information related to the performance of communication between the UE and the NW. In steps ST1871 to ST1874, the UE, IAB node #1, IAB node #2, and IAB donor DU transmit IAB processing feedback information using the AI / ML model for IAB processing to the IAB donor CU. The feedback information may include information identifying the UE, its own IAB node, and the IAB donor DU. The IAB donor CU that receives this information can recognize the node that transmitted the feedback information.
[0526] In this way, the IAB donor CU can acquire information about the communication performance between the UE and the NW via the IAB node or IAB donor that uses the output information derived by the AI / ML model for IAB processing. The IAB donor CU can evaluate the effectiveness of IAB processing when the AI / ML model for IAB processing is used, using the IAB processing feedback information using the AI / ML model for IAB processing that is held by the own node and the IAB processing feedback information using the AI / ML model for IAB processing that is acquired from the UE, IAB node #1, IAB node #2, and IAB donor DU.
[0527] The feedback information may be transmitted by the method of transmitting input information for the AI / ML model for IAB processing disclosed above, as appropriate. The feedback information may be transmitted from the UE to the IAB donor CU by, for example, RRC signaling.
[0528] Although it has been disclosed that the feedback information on the performance of the IAB process and the feedback information on the performance of communication between the UE and the NW via the IAB node are transmitted separately, this is not limiting and the transmission of these pieces of feedback information may be combined. For example, the transmission of these pieces of feedback information may be performed in the processing of steps ST1871 to ST1874 after the RRC connection processing of step ST1861. As another method, part of the feedback information on the performance of the IAB process may be transmitted in steps ST1851 to ST1853, and other feedback information may be transmitted in combination with the feedback information on the performance of communication between the UE and the NW in steps ST1871 to ST1874. This makes it possible to flexibly transmit various types of feedback information in accordance with the communication processing. The IAB donor CU can perform more effective IAB processing by using this feedback information to evaluate the effectiveness of the IAB processing.
[0529] A request for input information or feedback information from the AI / ML model for IAB processing may not be necessary. Each node may transmit input information or feedback information from the AI / ML model for IAB processing to a node having the AI / ML model for IAB processing as appropriate. For example, the transmission may be performed periodically. For example, the input information or feedback information may be transmitted when new input information or feedback information is obtained. This reduces the amount of signaling. When a node having an AI / ML model for IAB processing receives the information from another node or from itself, it may transmit an input information request to other nodes excluding the node. This reduces unnecessary signaling, such as transmitting the same input information.
[0530] In this way, AI / ML is introduced into the IAB processing, enabling more effective IAB processing.
[0531] Other IAB donor CUs may have an AI / ML model for IAB processing. For example, an IAB donor CU in a PLMN that supports IAB may have an AI / ML model for IAB processing. Input information, requests for input information, output information, and feedback information may be exchanged with other IAB donor CUs, or some or all of the information may be exchanged using Xn signaling.
[0532] By doing this, even when IAB processing is performed due to a change in the IAB donor CU to which the IAB node is connected due to movement of the IAB node, input information and feedback information from other IAB donor CUs, IAB nodes and IAB donor DUs affiliated with the IAB donor CU, etc. can be input to the IAB processing AI / ML model. By acquiring information from even more nodes and inputting it into the IAB processing AI / ML model, output information suitable for various states of the IAB node and the UEs communicating using it can be derived. Performing IAB processing using this output information enables more effective IAB processing.
[0533] By using the method disclosed in this embodiment, it is possible to perform training and inference in IAB processing using an AI / ML model. By deriving output information of the AI / ML model using various types of input information, it is possible to optimize IAB processing. Furthermore, it is possible to take into account information other than measurement results by IAB-MT when configuring the BAP of an IAB node or the BH route. For example, it is possible to optimize IAB processing suited to the state of the IAB node, such as determining which IAB node to include in the BH route, and how to configure the IAB node and resources. Since more effective IAB processing is possible, it is possible to improve the communication performance between the UE and the NW using IAB.
[0534] 18 discloses an example in which an AI / ML model for IAB processing is provided in an IAB donor CU. Alternatively, the AI / ML model for IAB processing may be provided in an IAB node. A learning device and an inference device may be provided in the IAB node. The IAB node can derive the setting information required for IAB processing using the AI / ML model.
[0535] An IAB node having an AI / ML model for IAB processing may transmit requests for input information and feedback information for the AI / ML model for IAB processing to other IAB nodes, IAB donor DUs, and IAB donor CUs. Other IAB nodes, IAB donor DUs, and IAB donor CUs may transmit AI / ML model input information and feedback information for IAB processing to an IAB node having an AI / ML model for IAB processing.
[0536] An IAB node that performs inference using an AI / ML model for IAB processing and derives configuration information necessary for IAB processing may transmit output information including the configuration information to an IAB donor CU. For example, the output information may be transmitted during the IAB processing. For example, the output information may be transmitted before the IAB processing. In this manner, the IAB donor CU can use the output information to configure the IAB node and the IAB donor DU for the IAB processing.
[0537] Alternatively, an IAB node may transmit the AI / ML model output information for IAB processing to other IAB nodes, IAB donor DUs, and IAB donor CUs. For example, the output information may be transmitted during the IAB processing. For example, the output information may be transmitted before the IAB processing. In this manner, each node can perform the IAB processing using the output information. For example, the transmission of setting information for the IAB processing from the IAB donor CU to each node can be reduced.
[0538] An IAB node having an AI / ML model for IAB processing may perform inference using the model when it determines IAB processing. When it determines IAB processing, the IAB node may transmit a request for inference input information to each node. Each node may transmit inference input information to the IAB node in response to the request. The IAB node can derive AI / ML model output information using the inference input information. In this way, for example, when the IAB node determines IAB processing for IAB integration, it can derive setting information for IAB processing using the AI / ML model. In this way, input information closer to the IAB processing timing can be used. More effective IAB processing can be implemented.
[0539] Regarding the node where the AI / ML model for IAB processing is provided, the method of providing it in the IAB donor CU and the method of providing it in the IAB node may be combined, thereby enabling more effective IAB processing.
[0540] Variant 1 of Embodiment 4. In an IAB, an RLF may occur in a BH. When a BH RLF occurs, an IAB node performs an RRC re-establishment process using a new parent IAB node. The IAB node performs a BH route setup and a BAP setup between the IAB donor CU and the new parent IAB node. However, if the IAB node moves or the surrounding environment changes over time, the radio wave propagation environment fluctuates from moment to moment between the IAB node and the parent IAB node, and in the BH route using the new parent IAB node. This may result in the RRC re-establishment process failing. Furthermore, even if the RRC re-establishment process is successful, if the communication status of the BH link quickly deteriorates, a BH RLF may occur again, resulting in a problem of communication failure between the UE and the NW.
[0541] In this modification, a method for solving such a problem will be disclosed.
[0542] AI / ML is introduced into BH RLF processing. BH RLF processing includes, for example, parent IAB node selection when a BH becomes RLF, BAP setting processing, IAB node resource setting processing, etc. By using AI / ML in BH RLF processing, for example, the information used for parent IAB node selection, etc. is optimized.
[0543] The methods disclosed in the first embodiment may be applied as appropriate to the AI / ML model, learning device, and inference device.
[0544] In this modification, a BH RLF processing AI / ML model, which is an AI / ML model used for BH RLF processing, is provided in a base station. A learning device and an inference device may also be provided in a base station. The BH RLF processing AI / ML model may also be provided in a RAN node. For example, it may be provided in a CU or a DU. The BH RLF processing AI / ML model may also be provided in an IAB donor. For example, it may be provided in an IAB donor CU or an IAB donor DU. Configuration information required for BH RLF processing can be derived by a base station or each node using the AI / ML model, making configuration easier.
[0545] Regarding the information on whether or not an AI / ML model for BH RLF processing is available, the method disclosed in the first embodiment may be applied as appropriate.
[0546] Input information for the AI / ML model for BH RLF processing is provided. For example, it may be input information for the AI / ML model for IAB processing or RLF-related information. Seven examples of input information for the AI / ML model for BH RLF processing are disclosed below.
[0547] (1) Input information for the AI / ML model for IAB processing. (2) Information about RLF detection. (3) Information about RLF detection indication. (4) Information about the BH link at the IAB node. (5) Information about RLF recovery. (6) Information about RLF recovery indication. (7) A combination of (1) to (6).
[0548] (1) may be, for example, an example of input information for the AI / ML model for IAB processing disclosed in the fourth embodiment.
[0549] (2) may be, for example, information indicating which BH link the RLF is in. For example, it may include information identifying the BH link. For example, it may be information regarding the IAB node that caused the BH RLF. For example, it may include information identifying the IAB node. The IAB node may be a parent IAB node, a child IAB node, an IAB donor DU, etc. For example, it may be information indicating which parent IAB node the BH RLF occurred between. For example, it may be information indicating which path the RLF occurred on. For example, it may include information identifying the path. For example, it may be a BAP path identifier. For example, it may be a BAP routing identifier. Information on where the RLF was detected can be used as input information for training or inference. It may also include information regarding time. Information regarding RLF detection and information regarding time may be associated with each other.
[0550] (3) may be, for example, transmission information of the BH RLF detection indication. It may also be information indicating to which child IAB node the RLF detection indication was transmitted. For example, it may be reception information of the RLF detection indication. It may also be information indicating from which parent IAB node the RLF detection indication was received. It may also include information identifying the IAB node. Information on which the RLF detection indication was transmitted or received can be used as input information for training or inference. It may also include information related to time. Information related to the RLF detection indication and information related to time may be associated with each other.
[0551] (4) may be, for example, information about available BH links at the IAB node. For example, it may be information about BH links other than the BH link that caused the BH RLF. For example, it may be information about unavailable BH links at the IAB node. For example, it may be information about the BH link that received the BH RLF detection. The status of the BH links at the IAB node can be used as input information for training or inference. It may also include information about time. Information about the BH links at the IAB node may be associated with information about time.
[0552] (5) may be, for example, information about BH RLF recovery. For example, it may be BH RLF recovery success information. For example, it may be information about IAB nodes for which BH RLF recovery was successful. It may include information for identifying IAB nodes. For example, it may be information indicating which parent IAB node the BH RLF recovery was successful between, or which child IAB node the BH RLF recovery was successful between. The status of BH RLF recovery can be used as input information for training or inference. It may also include information about time. The information about BH RLF recovery and the information about time may be associated with each other.
[0553] (6) may be, for example, transmission information of a BH RLF recovery indication. It may also be information indicating to which child IAB node the BH RLF recovery indication was transmitted. For example, it may be reception information of a BH RLF recovery indication. It may also be information indicating from which parent IAB node the BH RLF recovery indication was received. It may also include information identifying the IAB node. Information on which the BH RLF recovery indication was transmitted or received can be used as input information for training or inference. It may also include information related to time. Information related to the RLF recovery indication and information related to time may be associated with each other.
[0554] The input information disclosed above can be current information or predicted information, which can be used as input to the AI / ML model to derive more effective output information in the future.
[0555] The input information disclosed above may be associated with information about the state of the IAB node. The information about the state of the IAB node may be appropriately applied to the information disclosed above. By associating the input information with the state of the IAB node, it is possible to take into account the state of the IAB node in which the input information is received.
[0556] The input information disclosed above may be history information. BH RLF history information may be provided. This may be included in IAB node history information, IAB-MT history information, IAB-DU history information, IAB donor history information, IAB donor DU history information, IAB donor CU history information, UE history information, etc. In this way, information about the BH RLF can be managed in association with the IAB node, thereby simplifying processing. Furthermore, the method of including the information in the UE history information allows the use of conventional processing related to UE history, thereby avoiding complication of processing.
[0557] Input information for the AI / ML model for BH RLF processing may be transmitted prior to model training and inference.
[0558] The input information transmission method for the BH RLF processing AI / ML model may be the input information transmission method disclosed in the fourth embodiment, as appropriate.
[0559] The output information of the AI / ML model for BH RLF processing is disclosed. Information used in BH RIF processing may also be used as the output information. Eight examples of output information of the AI / ML model for BH RLF processing are disclosed below.
[0560] (1) Output information of the AI / ML model for IAB processing. (2) Information on cell selection and cell reselection. (3) Information on cells available for cell selection and cell reselection. (4) Information on frequencies available for cell selection and cell reselection. (5) Information on BH routes. (6) Information on BH links. (7) Information on BH RLF settings. (8) A combination of (1) to (7).
[0561] (1) may be, for example, an example of output information of the AI / ML model for IAB processing disclosed in the fourth embodiment.
[0562] (2) may be, for example, setting information used in the cell selection process. For example, it may be setting information used in the cell reselection process. For example, it may be a threshold value of received power used in the cell selection process. For example, it may be a threshold value of received power used in the cell reselection process. For example, an offset may be provided in the setting information for the cell selection process. For example, an offset may be provided in the setting information for the cell reselection process. An offset for the IAB node may be provided. For example, an offset related to the state of the IAB node may be provided. It becomes possible to perform cell selection process and cell reselection process suitable for the IAB node.
[0563] The offset may be broadcast in an SIB or may be notified by RRC signaling. It may be transmitted by being included in cell selection configuration information or cell reselection configuration information. It may also include information indicating that the offset is for an IAB node. In the cell selection process and cell reselection process, the operation of a normal UE and an IAB node can be made different. It becomes possible to perform cell selection process and cell reselection process suitable for an IAB node.
[0564] (3) may be, for example, a cell of an IAB node capable of cell selection or cell reselection. For example, it may be a candidate cell of an IAB node capable of cell selection or cell reselection. The number of cells is not limited to one, but may be multiple. The number of IAB nodes is not limited to one, but may be multiple. Information identifying the IAB node may be included. Information identifying the cell may be included. Priority may be assigned to IAB nodes. For example, it may be information indicating the priority of an IAB node for cell selection or cell reselection, or information indicating the priority of a cell. For example, it may be information such as giving priority to the IAB node that was last connected. Since it becomes possible to specify an IAB node, cell, or candidate thereof in the cell selection or cell reselection process by BH RLF, more effective processing can be performed.
[0565] (4) may be, for example, a carrier frequency available for cell selection and cell reselection. For example, it may be a band available for cell selection and cell reselection. It may also be a candidate carrier frequency or band available for cell selection and cell reselection. The number of carrier frequencies or bands is not limited to one, and may be multiple. It may include information identifying the carrier frequency or band. For example, it may be information indicating the priority of the carrier frequency or band for cell selection and cell reselection. For example, it may be information such as prioritizing the last connected carrier frequency. Since it becomes possible to specify a carrier frequency or band, or a candidate for them, in the cell selection or cell reselection process by BH RLF, more effective processing can be performed.
[0566] (5) may be, for example, a BH route to be set in an IAB node in BH RLF recovery. For example, it may be a BH route candidate. The BH route may be, for example, BAP path information. It may include information to identify the BAP path. For example, it may be BAP routing information. It may include information to identify the BAP routing. The BH route is not limited to one, and may be multiple. For example, it may be information indicating the priority of the BH route. For example, it may be information such as giving priority to the BH route that was last connected. Since it becomes possible to set a BH route or a candidate thereof in BH RLF recovery, more effective processing can be performed.
[0567] (6) may be, for example, a BH link selected by an IAB node in BH RLF recovery. For example, it may be a candidate BH link. It may include information for identifying the BH link. The number of BH links is not limited to one, and may be multiple. For example, it may be information indicating the priority of the BH link. For example, it may be information such as giving priority to the last connected BH link. Since it is possible to set a BH link or its candidate in BH RLF recovery, more effective processing can be performed.
[0568] (7) may be, for example, a timer used for BH RLF processing. For example, it may be a timer from detecting out-of-sync to starting RRC re-establishment when synchronization is not possible. For example, it may be a timer from starting RRC re-establishment to transitioning to RRC_Idle state when cell selection is not possible. For example, it may be a counter for detecting physical problems. For example, it may be a counter for out-of-sync or a counter for in-sync.
[0569] Information regarding the relationship between the state of the IAB node and the output information may be provided. The information may be a list. The information regarding the state of the IAB node may be appropriately applied from the information regarding the state of the IAB node disclosed in the input information described above. For example, the information may be a list indicating the relationship between the position of the IAB node and information regarding cell selection. The information regarding the state of the IAB node is not limited to one piece of information, and may be a combination of multiple pieces of information disclosed above. In this way, it is possible to indicate the relationship between the state of the IAB node and the output information.
[0570] The method of transmitting the output information of the AI / ML model for BH RLF processing may be the same as or similar to the method of transmitting the output information disclosed in the fourth embodiment.
[0571] The output information of the AI / ML model for BH RLF processing disclosed above is appropriately notified to the IAB node and IAB donor DU that perform the BH RLF processing, thereby enabling more effective BH RLF processing to be performed using the AI / ML model for BH RLF processing.
[0572] Feedback information by BH RLF processing using an AI / ML model for BH RLF processing is disclosed. This information may include information on the performance of UE-NW communication, information on the performance of the IAB node, and information on the performance of the IAB donor. The IAB node performance and IAB donor performance may be the performance of the IAB node or IAB donor used in communication between the UE and the NW. Information on performance after a BH RLF may also be used. The input information for the AI / ML model for BH RLF processing disclosed above may be applied as appropriate. The IAB processing feedback information using the AI / ML model for IAB processing disclosed in the fourth embodiment may be applied as appropriate.
[0573] The BH RLF processing node and UE appropriately notify the node having the BH RLF processing AI / ML model of the BH RLF processing feedback information using the BH RLF processing AI / ML model disclosed above. This enables evaluation of the BH RLF processing AI / ML model's reward calculation and updating of the model. This enables more effective BH RLF processing.
[0574] The method for transmitting BH RLF processing feedback information using the AI / ML model for BH RLF processing between base stations, between CU and DU, and between UE and base station may be suitably applied to the input information transmission method disclosed above.
[0575] 19 and 20 are diagrams showing an example sequence of BH RLF processing using AI / ML for BH RLF processing according to Modification 1 of Embodiment 4. Fig. 19 shows the first half of the example sequence, and Fig. 20 shows the second half of the example sequence. Figs. 19 and 20 disclose an example in which an IAB donor CU has an AI / ML model for BH RLF processing and performs training and inference.
[0576] In step ST1901, the IAB donor CU is provided with an AI / ML model for BH RLF processing. In step ST1905, data communication is being performed among the UE, the recovery IAB node, the parent IAB node of the initial path, the intermediate IAB node, the IAB donor DU, and the IAB donor CU.
[0577] In step ST1911, the IAB donor CU having the AI / ML model for BH RLF processing transmits an input information request for model training to each IAB node and each IAB donor DU. The IAB nodes and IAB donor DUs do not have to be limited to the initial path and recovery path shown in the example of the figure. The IAB nodes and IAB donor DUs may be limited to those under the umbrella of the IAB donor CU.
[0578] The request may be triggered by a BH RLF process. For example, the request may be triggered by a BH RLF process due to BH RLF detection at any IAB node under an IAB donor CU. BH RLF information can be used as input information for model training. Also, for example, as disclosed in the fourth embodiment, the request may be triggered by an IAB process from an IAB node. The request may be triggered by a BAP setting update process, an IAB node resource setting update process, or a BH routing update process for an IAB node. Model training becomes possible using input information when a new IAB node is added or when settings are updated.
[0579] The request may include information indicating the required information. The request may include information indicating the purpose of the input information for the AI / ML model for BH RLF processing. For example, the request may include information indicating that it is for BH RLF processing. For example, the request may include information indicating that it is for model training. This is effective, for example, when the input information for the AI / ML model for BH RLF processing includes information for model training. For example, when the information held by the IAB node includes information for model training, information for inference, and information for feedback, the IAB node can quickly grasp the information that should be notified to the IAB donor CU, thereby enabling the IAB node to quickly notify the IAB donor CU.
[0580] The request may be transmitted between the IAB donor CU and the IAB donor DU or the IAB node, for example, using F1 signaling. The request may be transmitted between the IAB donor CU and the IAB node, for example, using RRC signaling. For example, the request for input information may be included in the measurement configuration transmitted from the IAB donor CU to the IAB node. This reduces the amount of signaling.
[0581] In step ST1912, each IAB node and each IAB donor DU transmits model training input information to the IAB donor CU. The IAB node and IAB donor DU may transmit model training input information acquired from the connected UE to the IAB donor CU. The IAB donor CU can input information about the UE connected to the IAB node or IAB donor DU into the AI / ML model.
[0582] For example, F1 signaling may be used to transmit input information between the IAB donor DU or IAB node and the IAB donor CU. For example, RRC signaling may be used to transmit input information between the IAB node and the IAB donor CU.
[0583] Measurements by IAB nodes may be sent to the IAB donor CU in measurement reports. These measurements may be radio measurements or L2 measurements. These measurements may be sent using the same signaling or different signaling. Measurements by IAB nodes may be used as input for model training.
[0584] In step ST1913, the IAB donor CU trains the AI / ML model for BH RLF processing using the input information for the AI / ML model for BH RLF processing held by the own node and the input information for the AI / ML model for BH RLF processing acquired from the IAB node and the IAB donor DU. The AI / ML model for BH RLF processing is updated through training. These processes may be performed as appropriate. They may be performed periodically, each time BH RLF processing is performed, or each time IAB processing is performed. By performing these processes periodically or each time IAB processing is performed, the same effect as disclosed in the fourth embodiment can be obtained. Furthermore, by performing these processes each time BH RLF processing is performed, it is possible to appropriately input input information, for example, when a new path is set or when the setting of the IAB node of the path is updated. This allows for more optimal model training of the AI / ML model for BH RLF processing.
[0585] In step ST1921, the IAB donor CU transmits a request for inference input information to each IAB node and each IAB donor DU. The IAB nodes and IAB donor DUs do not have to be limited to the initial path and recovery path shown in the example of the figure. The IAB donor DUs and IAB nodes may be limited to those under the IAB donor CU. The request may be triggered by BH RLF processing. The request may be triggered by IAB processing. The request for inference input information allows the inference input information to be acquired and input to the AI / ML model for BH RLF processing, and output information for BH RLF processing can be derived.
[0586] The request may include information indicating the required information. The request may include information indicating the use of the input information for the BH RLF processing AI / ML model. For example, the request may include information indicating that the input information is for inference. This is effective, for example, when the input information for the BH RLF processing AI / ML model includes information for inference.
[0587] In step ST1922, each IAB node and each IAB donor DU transmits inference input information to the IAB donor CU. The IAB node and IAB donor DU may transmit inference input information acquired from the UE connected to them to the IAB donor CU. The IAB donor CU can input information about the UE connected to the IAB node and IAB donor DU into the AI / ML model.
[0588] The transmission of the inference input information request and the transmission of the inference input information may be performed periodically. This allows the output information to be updated periodically. The transmission of the inference input information request and the transmission of the inference input information may be performed when a physics problem is detected. If the physics problem is not resolved, the system transitions to BH RLF, so communication may still be possible during the physics problem detection stage. As an alternative method, the transmission of the inference input information request and the transmission of the inference input information may be performed when loss of synchronization is detected a predetermined number of times. The predetermined number of times may be set to a value smaller than the predetermined number of times required to detect a physics problem. The predetermined number of times may be set in advance. For example, it may be statically set in a standard or transmitted from the IAB donor CU. By obtaining the inference input information at these stages, input information that better reflects the BH RLF situation can be obtained. More optimal output information can be obtained.
[0589] The method for requesting input information for inference and transmitting input information for inference may be the same as the method for requesting input information for model training and transmitting input information for model training disclosed above, as appropriate.
[0590] In step ST1923, the IAB donor CU performs inference using the BH RLF processing AI / ML model, using the input information for the BH RLF processing AI / ML model held by its own node and the input information for the BH RLF processing AI / ML model acquired from the IAB node and the IAB donor DU. The IAB donor CU derives BH RLF processing AI / ML model output information through this inference.
[0591] In step ST1931, the IAB donor CU transmits AI / ML model output information for BH RLF processing to each IAB donor DU and each IAB node. For example, it transmits configuration information for cell selection processing and cell reselection processing to each IAB donor DU and each IAB node. For example, it transmits BAP configuration information and IAB node resource configuration information to each IAB node. The transmitted output information may be configuration information for recovery due to BH RLF. In this way, the IAB node and IAB donor DU used for BH RLF processing can acquire AI / ML model output information for BH RLF processing. When an IAB node detects a BH RLF, it can use the output information to perform BH RLF processing between the IAB donor CU, each IAB donor DU, and each IAB node.
[0592] The method of transmitting output information may be appropriately applied to the method of transmitting setting information for BH RLF processing or the method of transmitting an input information request for the AI / ML model for BH RLF processing disclosed above.
[0593] In step ST1941, the recovery IAB node detects a BH RLF.
[0594] In step ST1951, the recovery IAB node performs a cell selection process to select a parent IAB node of the recovery path. The recovery IAB node performs an RRC re-establishment process with the IAB donor CU via the parent IAB node of the selected recovery path, intermediate IAB nodes, and the IAB donor DU. These processes may use the AI / ML model output information for BH RLF processing. This allows for the selection of a parent node that will be a more effective recovery path.
[0595] In step ST1952, IAB processing is performed between the IAB donor CU and the parent IAB node of the recovery path, the intermediate IAB node, the IAB donor DU, and the recovery IAB node. For example, this includes BAP configuration for each IAB node and resource configuration for each IAB node. In addition, a BH route from the recovery IAB node to the IAB donor CU is set. It is recommended that the AI / ML model output information for BH RLF processing be used in these processes. This allows for more effective recovery path and IAB node configuration.
[0596] In step ST1953, a BAP setting release process is performed between the IAB donor CU and the parent IAB node, intermediate IAB node, and IAB donor DU of the initial path. A release process for the IAB node resource setting may be performed. The path setting from the recovery IAB node may be released. The release process releases the settings of the IAB nodes and IAB donor DU of the initial path, thereby reducing the load.
[0597] The release process in step ST1953 may be limited to the path used by the recovery IAB node. The release process may not be performed for other IAB nodes. Alternatively, the update process may be performed for other IAB nodes. For example, when initial paths are set for multiple IAB nodes, this can avoid complicated control when a BH RLF is detected between one IAB node and the other IAB nodes.
[0598] In this way, the recovery IAB node is connected to the IAB donor CU via the recovery path. In step ST1961, data communication is performed between the UE and the IAB donor CU via the recovery IAB node, the parent IAB node of the recovery path, the intermediate IAB node, and the IAB donor DU.
[0599] In steps ST1971 to ST1975, the recovery IAB node, the parent IAB node on the recovery path, the intermediate IAB nodes, the IAB donor DU, and the UE transmit BH RLF processing feedback information using the AI / ML model for BH RLF processing to the IAB donor CU. The feedback information may include information identifying the own IAB node, the IAB donor DU, or the UE. The IAB donor CU that receives this information can recognize the node that transmitted the feedback information.
[0600] In this way, the IAB donor CU can acquire information regarding the performance of the BH RLF processing using the AI / ML model for BH RLF processing. The IAB donor CU can evaluate the effectiveness of the BH RLF processing when using the AI / ML model for BH RLF processing, using the BH RLF processing feedback information using the AI / ML model for BH RLF processing that is held by the own node and the BH RLF processing feedback information using the AI / ML model for BH RLF processing that is acquired from the recovery IAB node, the parent IAB node of the recovery path, the intermediate IAB node, the IAB donor DU, and the UE.
[0601] The feedback information may be transmitted by the method of transmitting input information for the AI / ML model for IAB processing disclosed above, as appropriate. The feedback information may be transmitted from the UE to the IAB donor CU b...
Claims
1. A communications system comprising: a base station compatible with a fifth-generation wireless access system; and a communications terminal connected to said base station, wherein the base station determines settings to be used in communications processing performed by the communications terminal using a model that has been trained using information obtained from the communications terminal already connected to said base station and other adjacent base stations, and notifies the communications terminal connected to said base station of the determined settings, and the communications terminal, upon receiving notification of the settings from a first base station that is the base station that determined the settings, uses the settings to perform communications processing for the first base station or communications processing for a second base station that is different from the first base station, after terminating the connection with the first base station.
2. The communication system according to claim 1, characterized in that the first base station uses the model to determine settings to be used in a random access process performed when the communication terminal connects to the first base station or the second base station, and notifies the communication terminal connected to the base station of the determined settings, and the communication terminal, having received notification of the settings from the first base station, uses the settings to perform a random access process for reconnecting to the first base station after terminating the connection with the first base station, or a random access process for connecting to a second base station that is a different base station from the first base station.
3. The communication system according to claim 1, characterized in that: the first base station uses the model to determine settings to be used in a data transmission process performed when the communication terminal is not connected to a base station, and notifies the communication terminal connected to the base station of the determined settings; and the communication terminal, having received notification of the settings from the first base station, after terminating the connection with the first base station, performs a data transmission process to the first base station or the second base station using the settings when it is not connected to either the first base station or the second base station.
4. A base station operating as an access / backhaul integration donor in a communication system compatible with a fifth-generation wireless access system, comprising: a base station that uses a model that has been trained using information obtained from each node in the access / backhaul integration to determine settings for processing to be executed when configuring the access / backhaul integration; and notifies each node in the access / backhaul integration of the determined settings.
5. The base station according to claim 4, comprising a central unit and one or more distributed units, wherein the central unit determines settings for processing to be performed when configuring the access / backhaul integration using a trained model using information obtained from the distributed units and information obtained from each node of the access / backhaul integration, and notifies the distributed units and each node of the access / backhaul integration of the determined settings.
6. A base station operating as an integrated access / backhaul donor in a communication system compatible with a fifth-generation wireless access system, comprising: a base station that determines settings to be used in a reconnection process between the integrated access / backhaul nodes using a model that has been trained using information obtained from each integrated access / backhaul node; and notifies each integrated access / backhaul node of the determined settings.
7. The base station according to claim 6, characterized in that it is composed of a central unit and one or more distributed units, and the central unit uses a trained model using information obtained from the distributed units and information obtained from each of the integrated access / backhaul nodes to determine settings to be used in re-connection processes between the distributed units and the integrated access / backhaul nodes and between the integrated access / backhaul nodes, and notifies the distributed units and each of the integrated access / backhaul nodes of the determined settings.
8. A communications terminal that connects to a base station compatible with a fifth-generation wireless access system, the communications terminal determining settings to be used in a random access process that the communications terminal performs to connect to a base station using a model that has been trained using information acquired from a first base station, which is a connected base station, and a second base station different from the first base station, and using the determined settings to perform a random access process to reconnect to the first base station after terminating the connection with the first base station, or a random access process to connect to a second base station different from the first base station.
Citation Information
Patent Citations
Network node and method in a wireless communication network
JP2020530221A
Cited By
Communication method, base station, terminal and storage medium
CN116419253A
A communication method, a base station, a terminal, and a storage medium
CN116419253B
Network node, base station, and communication method
WO2026154639A1
