Ai / ML-based signaling overhead reduction for UE measurement reporting
AI/ML models on UE devices optimize channel measurement reporting, addressing inefficiencies and reducing signaling overhead, thereby improving network resource management and battery life.
Patent Information
- Application Number
- PCT/IB2026/054457
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-05-08
- Filing Date
- 2026-05-05
- Publication Date
- 2026-11-12
AI Technical Summary
Existing UE channel measurement reporting methods in wireless communication systems are inefficient and result in unnecessary signaling overhead, which can burden network resources and battery life.
Implementing AI/ML models on UE devices to process reference signal measurements and determine whether to transmit channel measurement reports, optimizing reporting based on configured modes and previous indicators, thereby reducing unnecessary transmissions.
Reduces signaling overhead and conserves network resources while improving battery life by intelligently managing UE reporting, enhancing network efficiency and reducing redundant measurements.
Smart Images

Figure IB2026054457_12112026_PF_FP_ABST
Abstract
Description
AI / ML-BASED SIGNALING OVERHEAD REDUCTION FOR UE MEASUREMENT REPORTING FIELD OF TECHNOLOGY
[0001] The present disclosure relates generally to wireless communications and more particularly to User Equipment (UE) reporting in wireless networks and even more particularly to the use of Artificial Intelligence (AI)ZMachine Learning (ML) in UE reporting.BACKGROUND
[0002] User Equipment (UE) may be configured in accordance with a Network (NW) configuration for providing channel measurement reports which inform further decision making in wireless communication systems. For example, the NW may configure the UE with one or more rules for generating channel measurement reports used in the determining whether to undertake a particular mobile operation, such as a cell switch or handoff.
[0003] Developments and improvement in UE channel measurement reporting are desirable.SUMMARY
[0004] According to a first aspect of the present disclosure, there is provided an apparatus, comprising at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to: receive a configuration from a network for reporting measurements with a machine learning model running on the apparatus, the machine learning model taking as input a reporting mode and reference signal measurements; determine a reference signal measurement of each of a plurality of reference signals; input, to the machine learning model, the reference signal measurement of each of the plurality of reference signals; infer, by the machine learning model, a reporting indicator for each respective one of the plurality of reference signals, the reporting indicator indicating whether a channel measurement report is to include information of the respective one of the plurality of reference signals; and transmit, to a network node, the channel measurement report for one or more of the plurality of reference signals according to a respective reporting indicator.
[0005] In some or all examples of the first aspect, the channel measurement report includes up to a maximum number of the plurality of reference signals according to the respective reporting indicator.
[0006] In some or all examples of the first aspect, the machine learning model further takes as input one or more previous reporting indicators previously inferred by the machine learning model.
[0007] In some or all examples of the first aspect, the one or more previous reporting indicators were inferred by the machine learning model within a preconfigured time window of the inputting the reference signal measurement ofeach of the plurality of reference signals, the preconfigured time window being within 10 measurement periods of the inputting.
[0008] In some or all examples of the first aspect, the apparatus receives an indication of the preconfigured time window from the network node.
[0009] In some or all examples of the first aspect, the machine learning model comprises a plurality of machine learning models.
[0010] In some or all examples of the first aspect, each of the plurality of machine learning models is trained on a corresponding reporting mode of a plurality of reporting modes, and wherein the plurality of reporting modes includes at least one of: a reporting mode associated with the apparatus switching cells; a reporting mode associated with load balancing; a reporting mode associated with adding a cell to a prepared cell list; or a reporting mode associated with removing a cell from the prepared cell list.
[0011] In some or all examples of the first aspect, the instructions, when executed by the at least one processor, further cause the apparatus to select a subset of the plurality of machine learning models for inferring the reporting indicator.
[0012] In some or all examples of the first aspect, each of the subset of the plurality of machine learning models provides a numerical output.
[0013] In some or all examples of the first aspect, the numerical output is normalized to a value in a range of 0 and 1.
[0014] In some or all examples of the first aspect, the numerical output comprises a binary value based on a threshold value.
[0015] In some or all examples of the first aspect, the binary value comprises: a 1 when the numerical output is greater than the threshold value, or a 0 when the numerical output is less than the threshold value.
[0016] In some or all examples of the first aspect, the instructions, when executed by the at least one processor, further cause the apparatus to consolidate the numerical outputs of the subset of the plurality of machine learning models based on averaging corresponding numerical outputs of the subset of the plurality of machine learning models for inferring a corresponding reporting indicator for each of the plurality of reference signals.
[0017] In some or all examples of the first aspect, the instructions, when executed by the at least one processor, further cause the apparatus to evaluate each of the averaged outputs against an output threshold for inferring the corresponding reporting indicator for each of the plurality of reference signals.
[0018] In some or all examples of the first aspect, the instructions, when executed by the at least one processor,further cause the apparatus to consolidate the numerical outputs of the subset of the plurality of machine learning models based on applying a logic function to corresponding numerical outputs of the subset of the plurality of machine learning models for inferring a corresponding reporting indicator for each of the plurality of reference signals.
[0019] In some or all examples of the first aspect, the logic function comprises an AND operation or an OR operation.
[0020] In some or all examples of the first aspect, the reporting mode comprises a plurality of reporting modes.
[0021] In some or all examples of the first aspect, the reporting mode comprises at least one of a cell switch mode, a load balance mode, an add cell preparation mode, or a remove cell preparation mode.
[0022] In some or all examples of the first aspect, the reporting mode comprises a scrambled reporting mode.
[0023] In some or all examples of the first aspect, the received configuration identifies one or more of a first reporting mode and a second reporting mode different from the first, wherein the instructions, when executed by the at least one processor, further cause the apparatus to perform the inferring with one or more machine learning models associated with the identified one or more of the first reporting mode and the second reporting mode.
[0024] According to a second aspect of the present disclosure, there is provided an apparatus, comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to: receive a configuration from a network for estimating resources based on a reporting configuration having one or more reporting modes and a machine learning model running on the apparatus, the machine learning model taking as input the reporting configuration and reference signal measurements; determine a reference signal measurement at a first time for each of a plurality of reference signals; input, to the machine learning model, the reporting configuration and the reference signal measurement of each of the plurality of reference signals; infer, by the machine learning model, a predicted number of channel measurement reports to be transmitted at a second time subsequent to the first time according to the one or more reporting modes, and request, from a network node, reservation of resources for the second time based on the predicted number.
[0025] In some or all examples of the second aspect the instructions, when executed by the at least one processor, further cause the apparatus to, at the second time: measure the reference signal measurement of each of the plurality of reference signals; input, to the machine learning model, the reporting configuration and the reference signal measurement of each of the plurality of reference signals; infer, by the machine learning model, a reporting indicator for each of the plurality of reference signals, the reporting indicator indicating whether to transmit a channel measurement report for the plurality of reference signals according to the one or more reporting modes, and transmit, to the network node, the channel measurement reports according to the one or more reporting modes for one or more of the plurality of reference signals according to a respective reporting indicator.
[0026] In some or all examples of the second aspect, the first time comprises a window of time.
[0027] In some or all examples of the second aspect, the reporting configuration comprises an L1 measurement reporting event.
[0028] In some or all examples of the second aspect, the resources comprise uplink resources.
[0029] According to a third aspect of the present disclosure, there is provided an apparatus, comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to: receive a configuration from a network for adjusting a reporting measurement threshold based on a reporting configuration and a machine learning model running on the apparatus, the machine learning model taking as input the reporting configuration and reference signal measurements, the reporting measurement threshold defining whether channel measurements of a reference signal are to be reported to the network; determine a reference signal measurement at a first time for each of a plurality of reference signals; input, to the machine learning model, the reporting configuration, the reference signal measurement of each of the plurality of reference signals, and an indication of one or more of the plurality of reference signals reported to the network at a second time previous to the first time; infer, by the machine learning model, a change in a reporting measurement threshold for the reporting configuration, and adjust the reporting measurement threshold based on the inferred change.
[0030] In some or all examples of the third aspect, the instructions, when executed by the at least one processor, further cause the apparatus to transmit, to a network node, the channel measurement report for the plurality of reference signals for which the reference signal measurement meets the adjusted reporting measurement threshold.
[0031] In some or all examples of the third aspect, the instructions, when executed by the at least one processor, further cause the apparatus to transmit to the network, the adjusted reporting measurement threshold for the reporting configuration.
[0032] In some or all examples of the third aspect, the instructions, when executed by the at least one processor, further cause the apparatus to transmit, to the network, the change in the reporting measurement threshold.
[0033] In some or all examples of the third aspect, the reporting configuration comprises an L1 measurement reporting event.
[0034] In some or all examples of the third aspect, the instructions, when executed by the at least one processor, further cause the apparatus to receive, from the network, a permissible range of adjustments for adjusting the reporting measurement threshold of the reporting configuration.
[0035] In some or all examples of the first, second and third, aspect, the instructions, when executed by the at least one processor, further cause the apparatus to: detect an apparatus battery level, and transmit a capability of theapparatus; wherein the capability comprises a request for Signal Overhead Reduction (SOR) when the apparatus battery level is below a threshold, and wherein the capability comprises an indication of using the machine learning model when the apparatus battery is above the threshold.
[0036] In some or all examples of the first, second and third, aspect, the reference signal measurement comprises an L1 measurement, an L3 measurement, a reference signal received power (RSRP), a reference signal received quality (RSRQ), or a signal-to-interference-plus-noise ratio (SI NR).
[0037] In some or all examples of the first, second and third, aspect, the apparatus comprises a user equipment (UE).
[0038] According to a fourth aspect of the present disclosure, there is provided a method for causing performance of an apparatus, the method including storing instructions in memory that, when executed by at least one processor of the apparatus, cause the apparatus to perform according to some or all of the examples of the first, second, and third aspects.
[0039] According to a first aspect of the present disclosure, there is provided a non-transitory computer-readable medium comprising instructions stored thereon for execution by at least one processor of an apparatus to cause the apparatus to perform according to some or all of the examples of the first, second, and third aspects.
[0040] The above summary provides a basic understanding of some aspects of the present disclosure. This summary is not an extensive overview of the present disclosure. Nor is it intended to be used to limit the scope of the present disclosure. Other aspects and features of the present disclosure will become apparent to those of ordinary skill in the art upon review of the following description of example implementations in conjunction with the accompanying figures.BRIEF DESCRIPTION OF THE DRAWINGS
[0041] Embodiments of the present disclosure will now be described, by way of example only, with reference to the attached figures.
[0042] FIG. 1 shows an example of a communication network to which examples disclosed herein may be applied.
[0043] FIG. 2 shows an example method of exchanging messages in a communication network in consideration of performing a cell switch.
[0044] FIG. 3A shows an example of an implementation for training an AI / ML model to infer a set of channel reporting indicators.
[0045] FIG. 3B shows an example of an implementation of a trained AI / ML model operating in an inference phasefor inferring a set of channel reporting indicators.
[0046] FIG. 3C shows an example method of exchanging messages in a communication network in consideration of performing a cell switch based on channel measurement reports.
[0047] FIG. 4A shows an example of an implementation for an AI / ML model in a training phase and an inference phase for predicting a number of channel measurement reports to report at a future time epoch.
[0048] FIG. 4B shows an example method of exchanging messages in a communication network in consideration of performing a cell switch and further reserving UL resources for a predicted number of channel measurement reports.
[0049] FIG. 5A shows an example of an implementation for training an AI / ML model to infer a change in a reporting configuration.
[0050] FIG. 5B shows an example of an implementation of a trained AI / ML model operating in an inference phase for inferring a change in a reporting configuration.
[0051] FIG. 5C shows an example method of exchanging messages in a communication network in consideration of performing a cell switch and further modifying a reporting configuration.
[0052] FIG. 6 shows an example method of exchanging messages in a communication network in consideration of a UE signaling its capability based on a battery state.
[0053] FIG. 7 shows an example method of measurement reporting with a user equipment.
[0054] FIG. 8 shows an example method of reserving resources with a user equipment.
[0055] FIG. 9 shows an example method of adjusting reporting criteria with a user equipment.DETAILED DESCRIPTION
[0056] For simplicity and clarity of illustration, reference numerals may be repeated among the figures to indicate corresponding or analogous elements. Numerous details are set forth to provide an understanding of the examples described herein. The examples may be practiced without these details. In other instances, well-known methods, procedures, and components are not described in detail to avoid obscuring the examples described. The description is not to be considered as limited to the scope of the examples described herein.
[0057] The following embodiments are exemplary. Although the specification may refer to “an”, “one”, or “some” embodiment(s) in several locations of the text, this does not necessarily mean that each reference is made to the same embodiment(s), or that a particular feature only applies to a single embodiment. Single features of different embodiments may also be combined to provide other embodiments. Further, when a particular feature, structure, or characteristic is described in connection of an embodiment, it is within the knowledge of one skilled in the art to apply such feature,structure, or characteristic in connection with other embodiments whether or not explicitly described. It shall be understood that although the terms “first,” “second” and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another.
[0058] For the purposes of the present disclosure, the phrases “at least one of A or B”, “at least one of A and B”, and “A and / or B” means (A), (B), or (A and B). For the purposes of the present disclosure, the phrase “A, B, and / or C” means (A), (B), (C), (A and B), (A and C), (B and C), or (A, B, and C).
[0059] Embodiments described may be implemented in a communication network, such as any of the following radio access technologies (RATs): Worldwide Interoperability for Micro-wave Access (WiMAX), Global System for Mobile communications (GSM, 2G), GSM EDGE radio access Network (GERAN), General Packet Radio Service (GRPS), Universal Mobile Telecommunication System (UMTS, 3G) based on basic wideband-code division multiple access (W-CDMA), high-speed packet access (HSPA), Long Term Evolution (LTE), LTE-Advanced, and enhanced LTE (eLTE), 5G (also called NR), or any future RAT such as 6G. Moreover, communication within the communication network may utilize any proper wireless communication technology, comprising but not limited to: Code Division Multiple Access (CDMA), Frequency Division Multiple Access (FDMA), Time Division Multiple Access (TDMA), Frequency Division Duplex (FDD), Time Division Duplex (TDD), Multiple-Input Multiple-Output (MIMO), Orthogonal Frequency Division Multiple (OFDM), and / or Discrete Fourier Transform spread OFDM (DFT-s-OFDM).
[0060] As used herein, the term “network device” or “network node” refers to a node in a communication network via which user equipment may access the network and / or which is capable of controlling radio communication and managing radio resources within a cell. The network node or network device may be referred to as a base station (BS), an access point (AP) or an access node. The network device may be, depending on the applied technology, for example, a node B (NodeB or NB), an evolved NodeB (eNodeB or eNB), an NR NB (also referred to as a gNB), a Remote Radio Unit (RRU), a radio head (RH), a remote radio head (RRH), a relay, an Integrated Access and Backhaul (IAB) node, a low power node, a non-terrestrial network (NTN) or non-ground network device such as a satellite network device, a low earth orbit (LEO) satellite and a geosynchronous earth orbit (GEO) satellite, or an aircraft network device.
[0061] Moreover, in connection of split radio access network (RAN), the network device may refer to a centralised unit (CU) of a base station and / or a distributed unit (DU) of a base station. An interface between CU and DU may be referred to as an F1 interface in NR. In the split RAN architecture, node operations may be carried out, at least partly, in the central / centralized unit, CU, (e.g. server, host or node) operationally coupled to the DU, (e.g. a radio head / node). One CU may control one or more DUs, acting at least as transmit / receive (Tx / Rx) nodes. In some embodiments, the DUs may comprise e.g. a radio link control (RLC), medium access control (MAC) layer and a physical (PHY) layer, whereas the CU may comprise the layers above RLC layer, such as a packet data convergence protocol (PDCP) layer, a radio resource control (RRC) and an internet protocol (IP) layers. Other functional splits are possible too. In practice, anyprocessing task may be performed in either the CU or the DU and the boundary where the responsibility is shifted between the CU and the DU may depend on the applied implementation.
[0062] The term “terminal device” refers to any end device that may be capable of wireless communication. By way of example, a terminal device may be referred to as a communication device, user equipment (UE), a Subscriber Station (SS), or a Mobile Station (MS). The terminal device may include a mobile phone, a cellular phone, a smart phone, voice over IP (VoIP) phones, wireless local loop phones a tablet, awearable terminal device, a personal digital assistant (PDA), portable computers, desktop computer, image capture terminal devices such as digital cameras, gaming terminal devices, music storage and playback appliances, vehicle-mounted wireless terminal devices, USB dongles, an Internet of Things (loT) device, a watch or other wearable, a head-mounted display (HMD), a vehicle, a drone, a medical device and applications (e.g., remote surgery), an industrial device and applications (e.g., a robot and / or other wireless devices operating in an industrial and / or an automated processing chain contexts), a consumer electronics device, a device operating on commercial and / or industrial wireless networks, and the like.
[0063] A term “resource”, as used herein, may refer to radio resources in time domain, in frequency domain, in space domain, and / or in code domain. Some examples of resources include e.g. a physical resource block (PRB), a radio frame, a subframe, a time slot, a subband, a frequency region, a sub-carrier, a beam, etc. The term “transmission” and / or “reception” may refer to wirelessly transmitting and / or receiving via a wireless propagation channel on radio resources.
[0064] Fig. 1 illustrates an example of a communication network to which examples disclosed herein may be applied. The communication network or a cellular communication network may comprise a network node 110 providing one or more cells, such as cell 100, and a network node 112 providing one or more other cells, such as cell 102. Each cell may be, e.g., a macro cell, a micro cell, femto, or a pico cell, for example. The cell may define a coverage area or a service area of the corresponding access node.
[0065] The network node 110 my provide a user equipment (UE) 120 (one or more UEs) with wireless access to the communication network. The wireless access may comprise downlink (DL) communication from the network node to the UE 120 and uplink (UL) communication from the UE 120 to the network node. Examples of uplink channels comprise physical uplink control channel (PUCCH) for transmitting control information and physical uplink shared channel (PUSCH) for transmitting data towards the network. Examples of downlink channels comprise physical downlink control channel (PDCCH) for transmitting control information and physical downlink shared channel (PDSCH) fortrans-mitting data towards the user equipment.
[0066] There may be a plurality of UEs 120, 122 in the system. Each of them may be served by the same or by different network nodes 110, 112. UE may be configured with dual connectivity (DC), wherein the UE, e.g. UE 120, may be connected to multiple network nodes 110, 112. The UEs 120, 122 may communicate with each other, in case device-to-device (D2D) communication interface is established between them via a so-called sidelink (SL). Such D2D communications may be referred to as machine-to-machine, peer-to-peer (P2P) communications, or vehicle-to-vehicle (V2V), for example.
[0067] In the case of multiple network nodes in the communication network, the network nodes may be connected to each other via an interface. LTE specifications call such an interface as X2 interface. An interface between an LTE node and a 5G node, or between two 5G nodes may be called Xn interface.
[0068] The network nodes 110 and 112 may be further connected via an-other interface to a core network 116 of the communication network. The LTE specifications specify the core network as an evolved packet core (EPC), and the core network may comprise e.g. a mobility management entity (MME) and a gateway node. The MME may handle mobility of terminal devices in a tracking area encompassing a plurality of cells and handle signalling connections between the terminal devices and the core network. The gateway node may handle data routing in the core network and to / from the terminal devices. The 5G specifications specify the core network as a 5G core (5GC). The 5G core may comprise e.g. an access and mobility management function (AMF) and a user plane function / gateway (UPF) and other functions. The AMF may handle termination of non-access stratum (NAS) signalling, NAS ciphering & integrity protection, registration management, connection management, mobility management, access authentication and authorization, security context management. The UPF node may support packet routing and forwarding, packet inspection and quality of service (QoS) handling, for example.
[0069] FIG. 2 illustrates an example method 200 of exchanging messages in support of a mobile operation. In the illustrative example of FIG. 2, the mobile operation comprises a cell switching operation, also known as a handoff, involving switching a UE from a first node to a second node, for example from a first network node or source node, to a second network node or target node. Mobile operations in accordance with the present disclosure may comprise a plurality of phases. For example, the method 200 illustrates a preparation phase, an execution phase, and a completion phase. With further reference to the method 200, steps 201 to 212 can constitute a preparation phase, steps 213 to 220 may constitute an execution phase, and steps 221 to 223 may constitute a completion phase.
[0070] At 201, the UE, such as a UE 120, provides a channel measurement report (or measurement report for short) for a plurality of channel reference signals (or reference signals for short) to a source DU, such as to a network node 110. The UE may generate and transmit a channel measurement report in accordance with a configuration provided to the UE, which may among other functions, configure the UE for providing channel measurement reports according to one or more reporting modes. For example, the UE may be configured to transmit channel measurement reports for a particular mobile operation, such as a handoff, on a periodic or aperiodic basis, or based on other criteria, such as a measurement threshold. The channel measurement report may include a plurality of reference signal measurements for a corresponding plurality of reference signals. For example, the channel measurement report mayinclude reference signal measurements provided by the UE corresponding to the reference signals of the source DU (e.g. the network node 110), a target DU (e.g. the network node 112), and other nodes such as candidate nodes. The reference signal measurements may comprise, for example, a reference signal received power (RSRP), a reference signal received quality (RSRQ), or a signal-to-interference-plus-noise ratio (SI NR). Other examples of reference signal measurements include, but are not limited to, L1 and L3 measurements as may be used for example in L1 / L2 inter-cell mobility or lower layer triggered mobility (LTM). In an embodiment, the UE provides a channel measurement report based on a first type of reference signal measurement. In an embodiment, the first type of reference signal measurement comprises L3 measurements.
[0071] At 202, the source node forwards the channel measurement report received from the UE to a further network node, such as a CU. The channel measurement report may be provided via an UL channel from the source DU to the CU in accordance with an RRC message transfer protocol.
[0072] At 203, the CU makes a determination based on the channel measurement report. For example, the CU may make a determination on further steps for a mobile operation associated with the channel measurement report. In the illustrative example of FIG. 2, the CU determines, based on the channel measurement report, whether a handoff may be necessary, and accordingly, makes a decision about cell preparation.
[0073] At 204, the CU, in response to 203, transmits to the target node, a UE context setup request in preparation for supporting a handoff of the UE from the source cell to the target cell.
[0074] At 205, the target node, in response to 204, transmits to the CU, a response to the UE context setup request.
[0075] At 206, the CU, in response to 205, transmits to the source node, a UE context modification request which may for example include, provisions of the target cell information, such as the target cell, CSI resource configuration, TCI states, and so forth.
[0076] At 207, the source node, in response to 206, transmits to the CU, a response to the UE context modification request.
[0077] At 208, the CU, in response to 207 creates an RRC Reconfiguration. The RRC Reconfiguration may comprise one or more configurations for configuring operation of the UE and / or one or more prepared cells. For example, the RRC Reconfiguration may configure the UE to provide channel measurement reports for a mobile operation, such as a handoff, based on a second type of reference signal measurement different from the first type of reference signal measurement provided at 201. In an embodiment, the second type of reference signal measurement comprises an L1 measurement.
[0078] At 209, the CU transmits, to the source DU, the RRC Reconfiguration. The RRC Reconfiguration may beprovided via a DL channel, from the CU to the source DU, in accordance with an RRC message transfer protocol.
[0079] At 210, the source DU, in response to 209, transmits the RRC Reconfiguration to the UE.
[0080] At 211, the UE, in response to step 210, transmits an RRC Reconfiguration Complete to the source DU.
[0081] At 212, the source DU, in response to 211, transmit the RRC Reconfiguration Complete to the CU.
[0082] At 213, the UE provides a channel measurement report to the source DU. The UE is configured to provide the channel measurement report in accordance with the RRC Reconfiguration received from the source DU at 210. For example, the RRC Reconfiguration may configure the UE to report based on a reporting mode, to report based on periodic or aperiodic timing, or to report based on other criteria of one or more mobile operations. As an illustrative example, the RRC Reconfiguration can configure the UE to report a channel measurement report with a plurality of L1 reference signal measurements for a corresponding plurality of reference signals, for use in determining a handoff operation.
[0083] At 214, the source DU makes a determination based on the channel measurement report of 213. For example, the source DU may make a determination to initiate a timing advance (TA) acquisition in response to the channel measurement report for a handoff operation.
[0084] At 215, in response to 214, a TA acquisition procedure may be initiated, for example, by the CU. The TA acquisition procedure may trigger the UE to acquire TA for a set of candidate cells. In an embodiment, the set of candidate cells comprises a set of cells corresponding to the target node.
[0085] At 216, the UE continues to provide a channel measurement report to the source DU, mirroring 213.
[0086] At 217, the source DU, based on the channel measurement reports, decides to handover the UE from the source DU to a cell of the target DU, triggering further handoff procedures.
[0087] At 218, the source DU, in response to 217, transmits a cell switch command to the UE. In an embodiment, the cell switch command comprises a MAC CE trigger cell change command. In an embodiment, the cell switch command comprises the RRC Reconfiguration for a target cell of the target node. In response to the cell switch command, the UE switches to the target DU. The UE may be configured to perform Random Access (RA) to a target cell of the target node in the absence of TA availability. In an embodiment, the cell switch command comprises a TA of the target cell of the target node.
[0088] At 219, the UE initiates communication with the target DU. In an embodiment initiating communication with the target DU comprises transmitting an RRC Reconfiguration Complete message. Communication with the target DU may utilize UL resources of the target cell configured in accordance with the cell switch command of 218.
[0089] At 220, the target DU transmits an UL RRC message transfer to the CU.
[0090] At 221, the CU transmits a UE context release request to the source DU.
[0091] At 222, the source DU, in response to 221, transmits a UE context release complete to the CU.
[0092] At 223, the CU, in response to 222, performs a path switch from the source DU to the target DU.
[0093] The above steps illustrate a non-limiting example of a mobile operation, particularly an LTM handoff procedure for an intra-CU scenario reflective of Release 18 Mobility Work Item Description (WID). In this illustrative example, the handoff from a source node to a target cell of a target node is based on L1 measurements made in the MAC layer of a DU. However, aspects in accordance with the present disclosure are not so limited to the particular mobile operation, measurements, reporting requirements, and procedures of this illustrative example and may be applicable to other mobile operations, measurements, reporting requirements, and procedures. For example, R18 prescribes a periodic reporting scheme wherein the UE provides channel measurement reports of neighboring cells on a periodic basis. However, reporting may be based on other standards, such as R19 which prescribes an LTM eventbased reporting scheme for reporting particular reference signals when L1 measurements trigger off a particular condition. In an embodiment the LTM events are based on L1 measurements. In an embodiment, the LTM event comprises event LTM2, wherein reporting triggers when a beam of a source cell falls below an absolute threshold. In an embodiment, the LTM event comprises event LTM3, wherein reporting triggers when a beam of a candidate cell exceeds a beam of a serving cell by more than an offset. In an embodiment, the LTM event comprises event LTM4, wherein reporting triggers when a beam of a candidate cell exceeds an absolute threshold. In an embodiment, the LTM event comprises event LTM5, wherein reporting triggers when a beam of a serving cell falls below a first threshold and when a beam of a candidate cell exceeds a second threshold. In an embodiment, the LTM event comprises a plurality of LTM events. In an embodiment, the LTM event comprises a plurality of conditions.
[0094] UE reporting configurations may be controlled and provided to the UE by the network. With reference to the illustrative example of FIG. 2, a reporting configuration for a cell switch or handoff may configure the UE to report reference signal measurements, for example the top or highest reference signal measurements, of neighboring cells and / or candidate reference signals when reference signal measurements of the serving cell are less strong (e.g. an LTM3 event). As a more specific example, the network may configure the UE with a reporting mode based on the following criteria:target_RSRP- serving_RSRP> first offset
[0095] In other words, the network configures the UE to report when the RSRP of a target cell exceeds the RSRP of the serving cell by more than a first offset, such as a first offset comprising 1 dB. When the target RSRP exceeds the serving RSRP by more than a second offset, such as a second offset comprising 3dB, the cell switch itself may be further triggered. Examples of RSRP measurements include L1 and L3 measurements.
[0096] As further illustrative examples, other mobile operations or reporting mode configurations include, but are not limited to, load balancing, add cell preparation, and remove cell preparation, and so forth. In a load balancing operation, the network may configure the UE such that the reference signals of the serving cell and reference signals of the candidate cells are not compared, but rather, each are configured with a different threshold (e.g. an LTM5 event). In an add cell preparation operation, the network may configure the UE to add a cell to a prepared cell list, each having a beam which meets a threshold condition (e.g. an LTM4 event which exceeds a threshold). In a remove cell preparation operation, the network may configure the UE to remove a cell from the prepared cell list when the cell has a beam which falls below a threshold condition (e.g. an LTM4 event which falls below a threshold). Other mobile operations, reporting triggers, LTM events, and configurations are possible. In an embodiment, the reporting mode may be conditional on two or more mobile operations, for example, a reporting mode triggered by meeting the conditions of two or more mobile operations, such as meeting the conditions of both a cell switch operation and a load balancing operation.
[0097] Aspects in accordance with the present disclosure leverage Artificial Intelligence (AI)ZMachine Learning (ML) to provide improvements in UE channel measurement reporting. Artificial intelligence (Al) can be broadly defined as getting computers to perform tasks mimicking a human brain. Machine learning (ML) is one category of Al techniques: computer algorithms able to automatically improve their performance without explicit programming. Al algorithms were first conceived in the 1950’s but only in recent years AI / ML has become useful in vast area of real-world applications, partly due to advancements in computational power and in providing storage capacity for data.
[0098] AI / ML can help adjust and optimize radio access network (RAN) parameters and settings using (real-time) monitoring and prediction of network performance, quality, and demand. Additionally, AI / ML can identify and diagnose degradation in network performance, as well as provide protection from cyberattacks. AI / ML is usable in energy saving, load balancing, mobility optimization, link adaptation and security just to mention but a few.
[0099] It is envisioned that AI / ML will enable real-time analysis as well as automated operation and control in 5G and beyond RAN. This requires the availability of data streamed from wireless devices in a timely manner, especially in extremely time-critical applications such as real-time video monitoring and extended reality (XR). This may be reflected in network architecture, such as by placing and moving ML agents to the required locations in the network, for example for data collection. User devices (mobile devices) may assist network in decision-making in resource management, thus a user device may act as an infrastructure resource.
[0100] As the network evolves to programmable and flexible cloud native implementation, AI / ML-based network automation will be used to simplify network management and optimization. It is expected that parts of the air interface, in particular signal processing algorithms, are supported and eventually even replaced with machine learning models. Thus, a 6G wireless communication standard will natively support an Al-based air interface.
[0101] Machine learning algorithms are usually classified into four different types: supervised learning,unsupervised learning, semi-supervised learning and reinforcement learning.
[0102] In supervised learning the algorithm learns from labelled data. Fortraining, the algorithm receives input data and corresponding correct output labels. The algorithm is trained to predict accurate labels for new data.
[0103] In unsupervised learning the algorithm analyses unlabelled data. The aim is to discover patterns, relationships, or structures within the data, for example, un-supervised learning algorithms make groups of similar data points.
[0104] Semi-supervised learning is a hybrid machine learning approach that combines labelled and unlabelled data for training. A limited amount of labelled data and a larger set of unlabelled data is used to improve training. This approach is useful when acquiring labelled data is expensive or time-consuming as is the case in many real-world applications. Semi-supervised learning techniques can be applied to various tasks, such as classification, regression, and anomaly detection, allowing models to make more accurate predictions and generalize better in real-world scenarios.
[0105] Reinforcement learning is a machine learning algorithm which learns from trial and error. An ML agent interacts with environment and learns from experience aiming to maximize cumulative rewards. The ML agent receives feedback through rewards or penalties based on its actions. The agent learns to take actions that lead to the most favourable outcomes overtime. The algorithm adapts to changing environments, and achieve long-term goals through a sequence of actions.
[0106] An example of an ML algorithm found applicable to adjust and optimize the radio access network (RAN) parameters and settings is deep learning. Deep learning is a subset of machine learning algorithms using a neural network. Neural networks are also known as artificial neural networks (ANNs) or simulated neural networks (SNNs). Deep learning can be based on supervised, semi-supervised or unsupervised learning.
[0107] Artificial neural networks (ANNs) are comprised of an input layer, one or more hidden layers, and an output layer. Each node of a layer, or an artificial neuron, connects to another one and has an associated weight as well as a threshold value. If the output of an individual node is above the threshold value specified to this node, the node is activated and sending or passing data to the next layer of the neural network.
[0108] In the case the supervised learning is applied in the training of a neural network, the training is carried out by using examples, each of which contains a known "input" and "result", forming probability-weighted associations between them. The training comprises determining the difference between the output of the neural network (a prediction) for an input and a target output for the same input. The difference is called an error value. The neural network then adjusts its weighted associations according to a learning rule and using this error value. Successive adjustments makes the neural network produce output that is approaching the target output. After a sufficient number of these adjustments,the training can be terminated based on a certain criteria.
[0109] FIG. 3A illustrates an example implementation 300 of training an AI / ML model 310 for inferring a set of reporting indicators for UE channel measurement reporting. The AI / ML model 310 trains on a first input 312 (e.g. reference signal measurements), a second input 314 (e.g. a reporting configuration), and provides an output 320 (e.g. a set of reporting indicators). The AI / ML model 310 may also receive, as a further input, the output 320 from a previous time epoch. AI / ML models trained in this regard may learn to infer when to report reference signal measurements on a probabilistic or statistical basis, rather than on a deterministic basis, reducing transmission of unnecessary reporting, reducing UL burden on the NW, among other improvements in resource reduction.
[0110] The first input 312 comprises reference signal measurements provided by a UE at a time Tfor a plurality of reference signals and cells. Examples of reference signal measurements include, but are not limited to, RSRP, RSRQ, SI NR, L1 , and L3 measurements. The first input 312 may comprise an array or other suitable data structure for inputting the reference signal measurements to the AI / ML model 310. The array may be dimensioned wherein there are B rows corresponding to the number of reference signals and C columns corresponding to the number of cells to form a B x C array of reference signal measurements.
[0111] The second input 314 comprises a mode, such as a reporting mode, as used by a network to configure a UE for reporting reference signal measurements. Examples of modes for reporting include time-based criteria (e.g. periodic, semi-persistent, and aperiodic), reporting modes which include but are not limited to cell switching, load balancing, add cell preparation, remove cell preparation, and so forth, and LTM events. A reporting mode may further include a combination of reporting mode configurations. For example, a reporting mode may trigger reporting when concurrently meeting two or more different reporting mode criteria, such as meeting the reporting mode criteria for a first reporting mode, such as a cell switching reporting mode, concurrently with meeting the criteria for a second reporting mode, such as a load balancing reporting mode.
[0112] The network may further encode the reporting mode, for example, the network may encode a cell switch as 01, load balancing as 02, and so forth. In an embodiment, the network scrambles the reporting mode. In other words, the network provides a reporting mode that does not reveal or otherwise explicitly identify the reporting mode to the UE, obfuscating the purpose for which the UE provides channel measurement reports. As such, while the UE may be configured to report reference signal measurements in accordance with a particular reporting mode, the UE may lack explicit knowledge of what the reporting mode is. In this regard, the network reduces its exposure to the UE.
[0113] The output 320 comprises a set of reporting indicators. The set of reporting indicators can include an indication of which reference signal measurements of the first input 312 for a given time T should be reported when the UE is configured according to the reporting mode of the second input 314. The output 320 may comprise indexes of the reference signal measurements to be reported. The output 320 may comprise an array dimensioned to match the Bx Cindexing of the first input 312, wherein each entry in the output 320 comprises a reporting indicator for each respective reference signal measurement. The reporting indicator may comprise a positive indicator indicating that the corresponding reference signal measurement ought to be reported or may comprise a negative / null indicator indicating that the corresponding reference signal measurement should not be reported. For example, the reporting indicator may comprise a binary value wherein a 1 corresponds to a reporting indicator indicative of the respective reference signal measurement meeting a reporting mode criteria, and a 0 corresponds to a reporting indicator indicative of the respective reference signal measurement not meeting a reporting mode criteria, or vice versa. In an embodiment, the reporting indicator comprises a numerical output such as a score or a probability. The score or probability may be provided on a scale, such as a scale of 0 to 100, or 0 to 1 , or otherwise normalized to a value, such as a value from 0 to 1. The normalized value may be further assigned a binary value based on a comparison to a threshold. For example, a score or probability equal-to or above a threshold may be assigned a binary value of 1 , and a score or probability less than the threshold may be assigned a binary value of 0, or vice versa. In an embodiment, the threshold is 0.5.
[0114] The output 320 for a previous time epoch T - 1 may be provided as a further input to the AI / ML model 310 at a given time T. In an embodiment, the previous epoch of time may fall within a window of time W (or preconfigured time window) relative to a given time T. In an embodiment, the previous epoch of time may fall within a particular number of measurement periods or time epochs of the current time T. For example, a previous output within 10 measurement periods of the current time T may be provided as an input to the AI / ML model 310. Training the AI / ML model 310 with the previous reporting indicators may assist the AI / ML model 310 in learning to identify redundant reports or otherwise predict future reporting. For example, if the AI / ML model 310 determines that a reference signal measurement at time T should be reported, then it’s likely that the same reference signal will have a reference signal measurement at time T + 1 that also requires reporting (or vice-versa), making further reporting of the reference signal measurement at time T + 1 redundant. Accordingly training the AI / ML model 310 in accordance with the present disclosure provides, among other advantages, the ability to withhold reporting redundant reference signal measurements, reducing UL resource burden on the network. Furthermore, predicting the number of reference signal measurements at a given future time T+ k may be leveraged to reserve resources in anticipation of triggering a mobile operation associated with a reporting mode.Accordingly, the disclosed methodology for AI / ML-based reporting predictors optimizes measurement reporting by filtering out redundant reports not only across the time dimension but also along the beam dimension.
[0115] In an embodiment, the UE and the NW may create a common dataset for training the AI / ML model 310 with supervised learning, wherein the first input 312 comprises UE reference signal measurements, the second input 314 comprises a NW configured reporting mode, and a labelled output 320 comprising a set of reporting indicators, such as the indexes of the reference signals that are expected to be reported. Such training data may include instances of an output 320 wherein the set of reporting indicators comprises an empty set or null value.
[0116] In an embodiment, a UE may generate a dataset for training the AI / ML model 310. The UE, during a data collection phase, records reference signal measurements for use as the first input 312 and reports according to a time based reporting mode (e.g. periodic, semi-persistent, or aperiodic) or according to an event triggered reporting mode. The selected reporting criteria functions as the second input 314. Where the reporting mode is made explicit to the UE, the UE can infer which reference signal measurements the NW used and label an output 320 accordingly.
[0117] In an embodiment, the NW may generate a dataset for training the AI / ML model 310. The NW records the reported reference signal measurements of the UE for use as the first input 312. The NW further reports measurements according to a time based reporting mode (e.g. periodic, semi-persistent, or aperiodic) or according to an event triggered reporting mode. The selected reporting criteria functions as the second input 314. The NW further labels an output 320 based on which of the reference signal measurements of the first input 312 are expected to be reported according to the reporting criteria of the second input 314.
[0118] The AI / ML model 310 may be trained at the UE, NW, or other hardware resources within or external to the NW. Training teaches the AI / ML 310 to, among other functions, learn which reference signal measurements are reported based on past experience. In other words, the AI / ML model 310 learns which information is statistically relevant to the NW. Reporting measurements thus becomes a function of probability and statistics rather than a function of deterministic reporting criteria (e.g. periodic reporting, semi-persistent reporting, aperiodic reporting, event-triggered reporting, etc.). Once trained, the AI / ML model 310 may reside on the UE or the NW. Where the AI / ML model 310 resides on the NW, the NW may transfer the AI / ML model 310 to the UE, for example, as a precursor to or in advance of an RRC Reconfiguration, or, during the RRC Reconfiguration. Where the AI / ML model 310 resides on the UE, the NW may further configure use of the AI / ML model 310, reconfigure the AI / ML model 310, or otherwise provide a newAI / ML model 310.
[0119] Validating the AI / ML model 310 may include comparing a desired output (e.g. a labelled output) against an actual output of the AI / ML model 310 based on a metric. Different metrics may promote different trade-offs between resource overhead and probability of reporting desired reference signal measurements. Example metrics for validating the AI / ML model 310 include set distance metrics.
[0120] An example metric for comparing outputs includes an overlap coefficient between two outputs A and B:|A n B|Orertop('4-g) =min(MI,|B|)
[0121] An overlap coefficient metric promotes higher probability of including desired reference signal measurements in the channel measurement reporting at the expense of larger channel measurement reports as this metric does not penalize inclusion of reference signal measurements that are not in the desired output, leading to higher resource consumption, for example, higher UL resource use.
[0122] An example metric for comparing outputs includes a symmetric difference metric between two outputs A and 8:d(A, B) = |( \B) U (B\ )|
[0123] A symmetric difference metric promotes compact resource sets, as this metric penalizes inclusion of reference signal measurements in the channel measurement reporting that are not in the desired output at the expense of having a higher probability of a desired reference signal measurement not being included in the channel measurement reporting.
[0124] FIG. 3B illustrates an example implementation 330 of a plurality of AI / ML models 310a to 310M residing on a UE. Each of the AI / ML model is configured for operating in an inference phase having been trained for inferring a set of reporting indicators for channel measurement reporting. For example, each AI / ML model is configured according to training on deterministic reporting criteria and reference signal measurements, for inferring a set of reporting indicators indicating which reference signal measurements to report, as described for example with respect to the non-limiting examples and embodiments of FIG. 3A.
[0125] In the illustrative example of FIG. 3B, the implementation 330 includes a plurality M of AI / ML models, in particular a first AI / ML model 310a, a second AI / ML model 310b, and up to and including an / VIthAI / ML model 310M. In an embodiment, the UE may implement only a single AI / ML model. A NW may provide a configuration to a UE for configuring each instance of an AI / ML model running on the UE, for inferring a set of reporting indicators according to the reporting mode associated with the configuration. For example, the NW may provide a configuration for the first AI / ML model 310a to infer a set of reporting indicators based on the training of a first reporting mode and may provide a configuration for the second AI / ML model 310b to infer a set of reporting indicators based on the training of a second reporting mode. In other words, the NW may provide configurations for an AI / ML to infer a set of reporting indicators according to the reporting modes used to train the AI / ML models, including reporting modes such as periodic, semi-persistent, or aperiodic reporting, or reporting modes such as cell switching, load balancing, add cell preparation, and remove cell preparation, or reporting modes corresponding to LTM events, and / or combinations of two or more reporting modes. The NW may provide the configuration as an encoding corresponding to a reporting mode used in the training of an AI / ML model. For example, the NW may encode a configuration as 001 explicitly indicating a particular reporting mode associated with the code 001 , such as cell switching, upon which the AI / ML model will be configured to infer a set of reporting indicators. The NW may also provide a scrambled configuration which does not explicitly identify the reporting mode for the UE but otherwise configures the AI / ML model to infer a set of reporting indicators according to a reporting mode that does not become exposed to the UE. In an embodiment, the trained AI / ML models reside with the NW and the NW transfers one or more trained AI / ML models to the UE for operating in an inference phase.
[0126] In the inference phase, the trained AI / ML models receive as inputs, UE reference signal measurements overtime. Examples of reference signal measurements include, but are not limited to, RSRP, RSRQ, SI NR, L1, and L3 measurements. The UE may provide the reference signal measurements to each of the AI / ML models 310a to 310M as an input 312, for use in inferring a set of reporting indicators for a given time T. The reference signal measurements may be taken at the same epoch of time T for which the inference is being made, or, may be taken from a previous epoch of time T-1 or T-W for which an inference has already been made, wherein HZ corresponds to a window of time. In an embodiment, the window of time HZ is 100 ms or less. In an embodiment, the window of time HZ is 10 measurement periods or less. The use of measurements from a previous epoch of time for a current epoch of time may reduce channel measurement reporting burden as the same input would generate the same (i.e. redundant) channel measurement report of a previous time epoch. The input 312 may comprise an array or other suitable data structure for inputting the reference signal measurements to the AI / ML models. The array may be dimensioned wherein there are B rows corresponding to the number of reference signals being measured and C columns corresponding to the number of cells from which measurements are being taken, thereby forming a Bx C array.
[0127] Each of the AI / ML models 310a to 310M infers a corresponding set of reporting indicators 320a to 320M based on the input 312 and the configuration of the AI / ML model itself. For example, the first AI / ML model 310a infers a first set of reporting indicators 320a according to its trained configuration, the second AI / ML model 310b infers a second set of reporting indicators 320b according to its trained configuration, and the M-nAI / ML model 310M infers an M-nset of reporting indicators 320M according to its trained configuration. Each set of reporting indicators provides an indication of which reference signal measurements of the input 312 for a given time T should be reported in accordance with the configuration of the corresponding AI / ML model. The outputs 320a to 320M may comprise indexes of the reference signal measurements to be reported. The outputs 320a to 320M may comprise an array dimensioned to match the B x C indexing of the input 312, wherein each entry of the outputs 320a to 320M comprises a reporting indicator for each corresponding reference signal measurement. Examples of a reporting indicator include, but are not limited to, a numerical output such as a numerical score or value, a positive indicator, a negative / null indicator, a binary value, and a normalized value.
[0128] The plurality of sets of reporting indicators 320a to 320M may be consolidated to provide a consolidated set of reporting indicators 322 for use in preparing a channel measurement report for reporting to a node, such as a source node. The consolidated set of reporting indicators 322 may be derived based on both a model selection 340 and a consolidation function 342. For example, the NW may configure the UE in accordance with a model selection 340 to select only certain AI / ML models for use in generating the consolidated set of reporting indicators 322, such as selecting a subset of the plurality of AI / ML models 310a to 310M. As an illustrative example, the model selection 340 may comprise a subset of the AI / ML models comprising the first and second AI / ML models corresponding to the first AI / ML model 310a and the second AI / ML model 310b. Consequently, the consolidated channel measurement report 322 is derived only from the first set of reporting indicators 320a and the second set of reporting indicators 320b. Theimplementation 330 further includes applying a consolidation function 342 to the set(s) of reporting indicators of the selected models 340. The consolidation function 342 may comprise a logic function such as an OR operation or an AND operation. For example, the consolidation function 342 may be configured to apply an OR operation between the first set of reporting indicators 320a and the second set of reporting indicators 320b to generate a single consolidated set of reporting indicators 322 for use in preparing a channel measurement report for reporting to a source node. The NW may also configure the UE reporting with other functions or criteria. For example, the NW may set a maximum limit of reference signal measurements to report.
[0129] The consolidating function 342 may also apply other functions, such as non-logic functions, such as statistical functions including mean, medium, and average functions, or other mathematical functions. For example, the consolidating function may apply a statistical function, such as an average, to the corresponding numerical outputs from each set of reporting indicators 320a to 320M. For example, the consolidation function may average the rn entry of all sets of reporting indicators 320a to 320M, to determine an averaged corresponding reporting indicator for the rn entry; and, likewise, may determine an average corresponding numerical output for each respective entry of the plurality of sets of reporting indicators. The output of the consolidation, for example, the averaged numerical output for each respective entry of the plurality of sets of reporting indicators, may be further assessed against an output threshold to determine the corresponding reporting indicator for each of the plurality of reference signals.
[0130] FIG. 3C illustrates an example method 200a of exchanging messages in support of a mobile operation. In the illustrative example of FIG. 3C, the mobile operation comprises a cell switching operation, also known as a handoff, involving switching a UE from a source node to a target cell of a target node. Embodiments of a mobile operation in accordance with the present disclosure may comprise a plurality of phases, wherein the method 200a illustrates a preparation phase and an execution phase.
[0131] The method 200a may include one or more of the same or similar steps described and illustrated with respect to the method 200. For example, the method 200a may include a preparation phase having steps 201 through 212 which are the same or otherwise similar to steps 201 through 212 of the preparation phase of the method 200. For example, the method 200a may include the same steps 201 through 209, 211, and 212 as the method 200. In this regard, the method 200a may include a similar, but different step 210a. For example, at 210a, the method 200a may similarly comprise the source DU, in response to step 209, transmitting the RRC Reconfiguration to the UE. However, the RRC Reconfiguration may further comprise one or more instructions or configurations for configuring the UE to use trained AI / ML models in an inference phase, as may be configured by the CU at 208. For example, the CU may configure one or more aspects of the implementation 330. For example, the CU may provide, configure, or otherwise select, one or more AI / ML models 310a to 310M for inferring a set of reporting indicators, wherein each AI / ML model is associated with a reporting mode, wherein a reporting may comprise a plurality of reporting modes. The CU may further provide aconsolidation function 342 for generating a consolidated set of reporting indicators 322 based on consolidating a plurality of sets of reporting indicators output by a corresponding plurality of selected AI / ML models. The CU may further configure other functions and features relating to the AI / ML models, such as, a maximum number of reporting indicators to include in a set of reporting indicators. The CU may provide one or more aspects of the configuration in an explicit or scrambled fashion.
[0132] The method 200a may include an execution phase having one or more steps which are the same or otherwise similar to steps of the execution phase of the method 200.
[0133] The method 200a may include a step 213a, wherein the UE, in response to 210, operates one or more AI / ML models in an inference mode, selected in accordance with the RRC Reconfiguration, to generate a corresponding one or more channel measurement reports based on the corresponding set(s) of reporting indicators output by the one or more AI / ML models. In this illustrative example for a cell switch or handoff, at least one selected AI / ML model will be trained based on a reporting mode for a cell switch or handoff. The UE may further consolidate a plurality of sets of reporting indicators and further generate a single channel measurement report based on the consolidated set of reporting indicators.
[0134] The method 200a may include a step 214a, wherein the UE, in response to 213a, transmits one or more channel measurement reports to the source DU.
[0135] The method 200a may include a step 215a, wherein the source DU changes a configuration of the UE. For example, the source DU may provide a configuration which modifies one or more aspects of the implementation 330. For example, the source DU may provide a configuration which modifies one or more of a configuration of an AI / ML model, a selection of AI / ML models, a consolidation function, a maximum number of reporting indicators for an AI / ML output, a maximum number of reference signal measurements to report, and so forth.
[0136] The method 200a may include a step 215a, wherein the UE, in response to 214a, generates one or more channel measurement reports in accordance with the modified configuration. The UE may further consolidate a plurality of channel measurement reports to generate a single channel measurement report. For example, a plurality of AI / ML models may provide a corresponding plurality of sets of reporting indicators which may be further consolidated according to a consolidation function into a single set of reporting indicators for use in generating a single channel measurement report.
[0137] The method 200a may include a step 216a, wherein the UE, in response to step 215a, transmits one or more channel measurement reports to the source DU.
[0138] At 217, the source DU, based on the one or more channel measurement reports, can decide to handover the UE from the source DU to a cell of the target DU, triggering further handoff procedures.
[0139] The above steps illustrate a non-limiting example of aspects of a mobile operation, in particular a cell switch or handoff, wherein the UE is configured with an Al / ML trained to infer when to report reference signal measurements.
[0140] In this illustrative example, one or more steps may be omitted or excluded that may otherwise be required, or be optional, in a cell switch or handoff procedure. For example, the method 200a does not illustrate or describe any steps for a completion phase, such as steps 221 to 223 of the completion phase of the method 200. However, such steps may be included with the method 200a.
[0141] FIG. 4A illustrates an example implementation 400 of an AI / ML model 410 in either a training phase or an inference phase. The AI / ML model 410 is trained and used to predict a number N of channel measurement reports to be reported at a future time. The AI / ML model 410 trains on a first input 412 (e.g. reference signal measurements), a second input 414 (e.g. a set of reporting configurations), and provides an output 420 (e.g. a prediction) for use in estimating future resource requirements, such as UL resources, for the predicted number of channel measurement reports expected to be reported at a future point in time. The prediction may thus be used to preemptively reserve resources and further reduce or eliminate delay when subsequently proceeding to request the resources required to provide the report.
[0142] In both the training phase and the inference phase, the first input 412 comprises reference signal measurements provided by a UE for a plurality of reference signals and cells. Examples of reference signal measurements include, but are not limited to, RSRP, RSRQ, SINR, L1, and L3 measurements. The input 412 may comprise reference signal measurements for a current time T or for a previous epoch of time, such as within a previous window of time T- W. The previous window of time may comprise, for example, a period of time spanning a number of measurement periods relative to the current time T, such as within 10 measurement periods, or may comprise a fixed time such as 100 ms or less. The first input 412 may comprise an array or other suitable data structure for inputting the reference signal measurements to the AI / ML model 410. The array may be dimensioned wherein there are B rows corresponding to the number of reference signals and C columns corresponding to the number of cells to form a B x C array of reference signal measurements.
[0143] In both the training phase and the inference phase, the second input 414 comprises a rule-based configuration. The rule based configuration may comprise one or more event triggered reporting modes, as used by a NW to configure a UE for reporting reference signal measurements. Examples of event triggered reporting modes include, but are not limited to cell switching, load balancing, add cell preparation, remove cell preparation, and so forth, and LTM events. An event triggered reporting mode may further include a combination of singular event triggered reporting modes. For example, an event triggered reporting mode may trigger providing a channel measurement report when meeting the criteria of two singular event triggered reporting modes, such as concurrently meeting the reporting criteria of each of cell switching and load balancing. Accordingly, a UE may be configured to provide a channel measurement report for each of one or more event triggered reporting modes, each of which may comprise one or morereporting modes. For example, a UE may be configured to report based on three different event triggered reporting modes. The first reporting mode may comprise event triggered criteria for cell switching, the second reporting mode may comprise event triggered criteria for cell switching and load balancing, and a third reporting mode may comprise event triggered criteria for an add cell preparation. As such, the UE at any given time may provide 0, 1 , 2, or 3 channel measurement reports.
[0144] In both the training phase and the inference phase, the output 420 comprises a predicted number of channel measurement reports which are expected to be reported at a future time per the reporting configuration of the second input 414. In other words, given a set of reference signal measurements (i.e. the first input 412), the AI / ML model 410 predicts how many channel measurement reports are expected to be reported at a future time based on the reporting configuration (i.e. the second input 414). The future time may comprise a next epoch of time T + 1 relative to the current time T. Or, for example, the future time may comprise at a future epoch of time T + K relative to the current time T. The future epoch of time may correspond to a number of measurement periods K into to the future relative to the current time T. For example, the future epoch of time may correspond to up to 10 measurement periods into the future. Or for example, the future epoch of time may correspond up to a flat amount of time into the future, such as 100 ms or less. When operating in the inference phase, the output 420 of the AI / ML model 410 may be leveraged to preemptively reserve resources, such as an UL resource for each expected channel measurement report, at a future time epoch.
[0145] In an embodiment, the training phase may comprise recording reference signal measurements provided by the UE, for a particular reporting configuration, and recording the number of corresponding channel measurement reports triggered based on the reporting configuration. The recorded number of reports may function as a labelled output fortraining the AI / ML model in a supervised manner.
[0146] The AI / ML model 410 may be trained at the UE, NW, or other hardware resources within or external to the NW. Once trained, the AI / ML model 410 may reside on the UE or the NW. Where the AI / ML model 410 resides on the NW, the NW may transfer the AI / ML model 410 to the UE, for example, as a precursor to or in advance of an RRC Reconfiguration, or, during the RRC Reconfiguration. Where the AI / ML model 410 resides on the UE, the NW may further configure use of the AI / ML model 410, reconfigure the AI / ML model 410, or otherwise provide a newAI / ML model 410.
[0147] FIG. 4B illustrates an example method 200b of exchanging messages in support of a mobile operation. In the illustrative example of FIG. 4B, the mobile operation comprises a cell switching operation, also known as a handoff, involving switching a UE from a source node to a target cell of a target node. Embodiments of a mobile operation in accordance with the present disclosure may comprise a plurality of phases, wherein the method 200a illustrates a preparation phase and an execution phase.
[0148] The method 200a may include one or more of the same or similar steps described and illustrated withrespect to the method 200. For example, the method 200b may include a preparation phase having steps 201 through 212 which are the same or otherwise similar to steps 201 through 212 of the preparation phase of the method 200. For example, the method 200b may include the same steps 201 through 209, 211, and 212 as the method 200. In this regard, the method 200b may include a similar, but different step 210b. For example, step 210b may similarly comprise the source DU, in response to step 209, transmitting the RRC Reconfiguration to the UE. However, the RRC Reconfiguration may further comprise one or more instructions or configurations for configuring the UE to use a trained AI / ML model in an inference phase, as may be configured by the CU in step 208. For example, the CU may configure one or more aspects of the implementation 400. For example, the CU may provide, configure, or otherwise enable, a trained AI / ML model 410 to predict a number of channel measurement reports expected to be reported at a future time for a corresponding reporting configuration of the UE.
[0149] The method 200b may include an execution phase having one or more steps which are the same or otherwise similar to steps of the execution phase of the method 200.
[0150] The method 200b may include a step 213b wherein the UE, in response to 210b, operates an AI / ML model trained to predict a number N of channel measurement reports to be reported at a future time. The AI / ML model generates the prediction based on its reporting configuration and a set of reference signal measurements provided by the UE.
[0151] The method 200b may include a step 214b, wherein the UE, in response to 213b, transmits to the source DU, a predicted number N of channel measurement reports to report at a future time T + K. In this regard, the UE may request to reserve N number of resources, for the future time period T+ K. For example, the UE may request to reserve N UL resources, in other words, 1 UL for each expected channel measurement report, for the time T + K, io reduce or otherwise eliminate delay in requesting and accessing an UL at the time T+ K.
[0152] The method 200b may include a step 215b, wherein at the time T + K, the UE transmits N channel measurement reports to the source DU.
[0153] The method 200b may include a step 216b, wherein the source DU, based on the one or more channel measurement reports, decides to handover the UE from the source DU to a cell of the target DU, triggering further handoff procedures.
[0154] The above steps illustrate a non-limiting example of aspects of a mobile operation, in particular a cell switch or handoff, wherein the UE is configured with an AI / ML trained to predict a number of channel measurement reports to report at a future time. In this illustrative example, one or more steps may be omitted or excluded that may otherwise be required, or be optional, in a cell switch or handoff procedure. For example, the method 200b does not illustrate or describe any steps for a completion phase, such as steps 221 to 223 of the completion phase of the method 200.However, such steps may be included with the method 200b.
[0155] FIG. 5A illustrates an example implementation 500 of an AI / ML model 510 in a training phase. The AI / ML model 510 is trained to modify reporting configurations. The AI / ML model 510 trains on a first input 512 (e.g. reference signal measurements), a second input 514 (e.g. a set of reporting configurations), a third input 516 (e.g. a set of previous channel measurement reports) and provides an output 520 (e.g. a change in configuration) for use in modifying a reporting configuration, for example by adjusting a reporting measurement threshold that triggers a channel measurement report. The UE can thus autonomously and dynamically modify reporting configurations, avoiding the need for a NW to provide a reconfiguration.
[0156] In a training phase, the first input 512 comprises reference signal measurements provided by a UE for a plurality of reference signals and cells. Examples of reference signal measurements include, but are not limited to, RSRP, RSRQ, SI NR, L1, and L3 measurements. The input 512 may comprise reference signal measurements for a current time T or for a previous epoch of time, such as within a previous window of time T- W. The previous window of time may comprise, for example, a period of time spanning a number of measurement periods relative to the current time T, such as within 10 measurement periods, or may comprise a fixed time such as 100 ms or less. The first input 512 may comprise an array or other suitable data structure for inputting the reference signal measurements to the AI / ML model 510. The array may be dimensioned wherein there are B rows corresponding to the number of reference signals and C columns corresponding to the number of cells to form a B x C array of reference signal measurements.
[0157] In a training phase, the second input 514 comprises a rule-based configuration. The rule based configuration may comprise one or more event triggered reporting modes, as used by a NW to configure a UE for reporting reference signal measurements. Examples of event triggered reporting modes include, but are not limited to cell switching, load balancing, add cell preparation, remove cell preparation, and so forth, and LTM events. An event triggered reporting mode may further include a combination of singular event triggered reporting modes. For example, an event triggered reporting mode may trigger a channel measurement report when meeting the criteria of two singular event triggered reporting modes, such as concurrently meeting the reporting criteria of each of cell switching and load balancing. Accordingly, a UE may be configured to provide a channel measurement report for each of one or more event triggered reporting modes, each of which may comprise one or more reporting modes. For example, a UE may be configured to report based on three different event triggered reporting modes. The first reporting mode may comprise event triggered criteria for cell switching, the second reporting mode may comprise event triggered criteria for cell switching and load balancing, and the third reporting mode may comprise event triggered criteria for an add cell preparation.
[0158] In a training phase, the third input 516 comprises a channel measurement report from a previous time epoch, for example, the previous time epoch T- 1. The previous time epoch may also fall within a window of time Wrelative to a given time T, such as within a number of measurement periods relative to the current time T. For example, channel measurement reports from within the previous 10 measurement periods may be provided as the third input 516 when training the AI / ML model 510 relative to the current time T. As a further example, channel measurement reports within a previous amount of time, such as within the previous 100 ms, may be provided as the third input 516 when training the AI / ML model 510 relative to the current time T.
[0159] The channel measurement report can include an indication of which reference signal measurements fora previous time epoch should be reported when the UE is configured according to the reporting configuration of the second input 514. The channel measurement report may comprise indexes of the reference signal measurements to be reported. The channel measurement report may comprise an array dimensioned to match the B x C indexing of the first input 512, wherein each entry of the channel measurement report comprises a reporting indicator for each corresponding reference signal measurement. The reporting indicator may comprise a positive indicator indicating that the corresponding reference signal measurement ought to be reported or may comprise a negative / null indicator indicating that the corresponding reference signal measurement should not be reported. For example, the reporting indicator may comprise a binary value wherein a 1 corresponds to a positive reporting indicator and a 0 corresponds to a negative / null reporting indicator, or vice versa. In an embodiment, the reporting indicator comprises a numerical output such as a score or a probability. The score or probability may be provided on a scale of 0 to 1 or otherwise normalized to a value from 0 to 1. The normalized value may be further assigned a binary value based on a comparison to a threshold. For example, a score or probability equal-to or above a threshold may be assigned a binary value of 1, and a score or probability less than the threshold may be assigned a binary value of 0, or vice versa. In an embodiment, the threshold is 0.5.
[0160] In a training phase, the output 520 comprises a change in configuration for a reporting mode of the reporting configuration. The change in configuration may apply to a threshold criterion of the reporting mode. For example, the reporting configuration may include a reporting mode which triggers according to the following criteria:target_RSRP- serving_RSRP> first offset
[0161] In other words, when the RSRP of a target cell exceeds the RSRP of the serving cell by more than a first offset, such as a first offset comprising 1 dB, the reporting mode is triggered. The AI / ML model 510 however, may be configured to modify a reporting measurement threshold represented by the first offset in the expression above. For example, the NW may configure the first offset to have a value of 1 dB. The AI / ML model 510 may learn from training however, that the first offset may be modified in a manner that does not impair reporting. For example, the AI / ML model 510 may be trained to modify the first offset threshold to increase the first offset (i.e., the reporting measurement threshold) from 1 dB to 2dB. An increase in the reporting measurement threshold typically reduces report burden as fewer reference signals will meet the increased reporting measurement threshold. In other words, increasing thereporting measurement threshold eliminates reporting reference signals having the lowest reportable outputs. Such an increase to the threshold criterion may be made within a permissible range to ensure that the top reference signals continue to be reported, where the permissible range of adjustments may be preconfigured or communicated from the network to the UE. The output 520 of the AI / ML model 510 thus allows the UE to autonomously and dynamically modify a reporting criterion without the need of the NW having to reconfigure the UE.
[0162] In an embodiment, the UE and the NW may create a common dataset for training the AI / ML model 510 with supervised learning, wherein the first input 512 comprises UE reference signal measurements, the second input 514 comprises a NW reporting configuration, the third input 516 comprises channel measurement reports from a previous time epoch relative to the UE reference signal measurements, and a labelled output 520 comprising changes to a reporting configuration as provided by a NW.
[0163] In an embodiment, a UE may generate a dataset for training the AI / ML model 510. The UE, during a data collection phase, records reference signal measurements for use as the first input 512 and provides channel measurement reports, which may be used as the third input 516, according to according to a time based reporting mode (e.g. periodic, semi-persistent, or aperiodic) or according to an event triggered reporting mode. The selected reporting configurations functions as the second input 514. Where the reporting mode is made explicit to the UE, the UE can infer which channel measurement reports were used by the NW and infer the modified configuration (e.g. a delta or change in a threshold criterion) and label an output 520 accordingly.
[0164] In an embodiment, the NW may generate a dataset for training the AI / ML model 510. The NW records the reported reference signal measurements of the UE for use as the first input 512. The NW further reports measurements according to a time based reporting mode (e.g. periodic, semi-persistent, or aperiodic) or according to an event triggered reporting mode in accordance with the reporting configuration which functions as the second input 514. The NW further labels a desired output 520 (e.g. a delta or change to a reporting threshold criterion).
[0165] The AI / ML model 510 may be trained at the UE, NW, or other hardware resources within or external to the NW. Training teaches the AI / ML 510 to, among other functions, learn to modify reporting configurations. In other words, the AI / ML model 510 learns to autonomously and dynamically modify configurations, preempting the NW from having to reconfigure the UE. Once trained, the AI / ML model 510 may reside on the UE or the NW. Where the AI / ML model 510 resides on the NW, the NW may transfer the AI / ML model 510 to the UE, for example, as a precursor to or in advance of an RRC Reconfiguration, or, during the RRC Reconfiguration. Where the AI / ML model 510 resides on the UE, the NW may further trigger use of the AI / ML model 510.
[0166] FIG. 5B illustrates an example implementation 530 of a trained AI / ML model 510 configured for inferring a change in a reporting configuration. The trained AI / ML model 510 resides on a UE and may be enabled or triggered for operation in an inference based on a NW configuration provided to the UE.
[0167] In the illustrative example of FIG. 5B, the inference phase includes the same first input 512 and second input 514 as described with respect to the training phase of the implementation 500. The inference phase differs from the training phase, in part, in not requiring the third input 516 depicted in the training phase of the implementation 500. The AI / ML model 510 is otherwise trained to infer an output 520, when in in the inference phase, for changing a reporting configuration. The change in the reporting configuration can include a change to a threshold criterion of a reporting mode, as similarly described with respect to the output 520 of the training phase of the implementation 500. The output 520 of the AI / ML model 510 thus allows the UE to autonomously and dynamically modify a reporting criterion without the need of the NW having to reconfigure the UE. Modifying a reporting configuration in response to the output 520 may further trigger a UE to generate a new channel measurement report based on the modified reporting configuration.
[0168] FIG. 5C illustrates an example method 200c of exchanging messages in support of a mobile operation. In the illustrative example of FIG. 5C, the mobile operation comprises a cell switching operation, also known as a handoff, involving switching a UE from a source node to a target cell of a target node. Embodiments of a mobile operation in accordance with the present disclosure may comprise a plurality of phases, wherein the method 200c illustrates a preparation phase and an execution phase.
[0169] The method 200c may include one or more of the same or similar steps described and illustrated with respect to the method 200. For example, the method 200c may include a preparation phase having steps 201 through 212 which are the same or otherwise similar to steps 201 through 212 of the preparation phase of the method 200. For example, the method 200c may include the same steps 201 through 209, 211, and 212 as the method 200. In this regard, the method 200c may include a similar, but different step 210c. For example, step 210c of the method 200c may similarly comprise the source DU, in response to step 209, transmitting the RRC Reconfiguration to the UE. However, the RRC Reconfiguration may further comprise one or more instructions or configurations for enabling the UE to use a trained AI / ML model in an inference phase, as may be configured by the CU in step 208. For example, the CU may configure one or more aspects of the implementation 530. For example, the CU may provide a configuration for enabling use a trained AI / ML model 510 for inferring a change in a reporting condition, for further use in autonomously and dynamically modifying a reporting criterion without the need of the NW having to reconfigure the UE.
[0170] The method 200c may include an execution phase having one or more steps which are the same or otherwise similar to steps of the execution phase of the method 200.
[0171] The method 200c may include a step 213c, wherein the UE, in response to 210a, operates an AI / ML model 510 trained to infer a change in a reporting configuration. In this illustrative example, the trained AI / ML model 510 infers a change to a reporting threshold criterion, triggering the UE to automatically change a reporting configuration. For example, the change in the reporting configuration may comprise a delta for modifying a reporting threshold criterion for a reporting mode.
[0172] The method 200c may include a step 214c, wherein the UE, in response to 213c, transmits to the source DU, a modified reporting configuration. The UE may further concurrently transmit to the source DU, a channel measurement report, generated based on the modified reporting configuration.
[0173] The method 200c may include a step 215c, wherein the source DU, based on the new channel measurement report, decides to handover the UE from the source DU to a cell of the target DU, triggering further handoff procedures.
[0174] The above steps illustrate a non-limiting example of aspects of a mobile operation, in particular a cell switch or handoff, wherein the UE is configured with an AI / ML model trained to infer a change in a reporting configuration. In this illustrative example, one or more steps may be omitted or excluded that may otherwise be required, or be optional, in a cell switch or handoff procedure. For example, the method 200c does not illustrate or describe any steps for a completion phase, such as steps 221 to 223 of the completion phase of the method 200. However, such steps may be included with the method 200c.
[0175] FIG. 6 illustrates an example method 600 of exchanging messages in a NW between a UE and a gNB in support of determining a UE capability which may assist in reducing signaling overhead, providing benefits for both the UE and the NW in terms of power consumption. The UE capability may be predicated in part on a UE power or battery level.
[0176] At 601, the UE, such as a UE 120, applies a model for signal overhead reduction and determining a UE capability. The model may include first determining a current power or battery level of the UE. The UE battery level may be stated as a percentage, a binary value (e.g. 0 and 1 where 1 indicates a low power state, or vice versa), a numerical value, or a qualitative value such as critically low, low, medium, and high. The UE battery level may be further compared to one or more thresholds indicative of UE capability. For example, the model may define a first threshold wherein a UE battery level above the first threshold indicates a UE capability of being able to use trained AI / ML models without requiring Signaling Overhead Reduction (SOR). And, for example, the model may define a second threshold wherein a UE battery level below the second threshold indicates a UE capability requiring SOR.
[0177] At 602, the gNB may transmit to the UE, a UE capability enquiry.
[0178] At 603 and 604, the UE, in response to 602, may transmit to the gNB, including on a volunteer basis, a UE capability. The UE may transmit the UE capability to the gNB over UE assistance information (UAI). The UE capability may comprise, for example, whether UE-side SOR is supported and a current UE battery level.
[0179] At 605, the gNB makes a determination, based on the UE capability and available resources, whether to activate UE-side SOR. For example, if the UE battery level is below a threshold requiring SOR, the gNB may decide to activate UE-side SOR. Conversely, if the UE battery level is above a threshold, the gNB may decide not to activate UE-side SOR.
[0180] At 606, the gNB, in response to 605, may activate SOR for one or more reporting modes of the UE. When running the UE-side SOR, the UE may reduce power consumption and extend the remaining UE battery life.
[0181] At 607, the gNB may continue to monitor relevant key performance indicators (KPI) or otherwise request UE capability, in determining whether to deactivate SOR, and, where appropriate, may configure the UE to deactivate the UE-side SOR for one or more reporting modes.
[0182] The above steps illustrate a non-limiting example of aspects of configuring a UE to enable UE-side SOR. In this illustrative example, one or more steps may be omitted or excluded that may otherwise be required or be optional.
[0183] FIG. 7 illustrates an example method 700 of channel measurement reporting with a UE.
[0184] At 701, the UE receives a configuration from a NW for reporting measurements with a machine learning model running on the UE. The machine learning model may be trained and configured to infer in accordance with one or more embodiments for reporting a set of reporting indicators as disclosed herein. When operating in inference mode, the machine learning model may take as inputs, a reporting mode and reference signal measurements.
[0185] At 702, the UE determines a reference signal measurement of each of a plurality of reference signals.
[0186] At 703, the UE inputs, to the machine learning model, the reference signal measurement of each of the plurality of reference signals.
[0187] At 704, the machine learning model infers a reporting indicator for each respective one of the plurality of reference signals. Each reporting indicator includes an indication of whether a channel measurement report is to include information for the respective one of the plurality of reference signals.
[0188] At 705, the UE transmits to the NW node, the channel measurement report with one or more of the plurality of reference signals according to a respective reporting indicator.
[0189] FIG. 8 illustrates an example method 800 of reserving resources with a UE.
[0190] At 801 , the UE receives a configuration from a NW for estimating resources based on a reporting configuration having one or more reporting modes for a machine learning model running on the UE. The machine learning model may be trained and configured to infer in accordance with one or more embodiments for estimating resources, such as UL resources, as disclosed herein. When operating in inference mode, the machine learning model may take as inputs, the reporting configuration and reference signal measurements.
[0191] At 802, the UE determines, at a first time, a reference signal measurement of each of a plurality of reference signals.
[0192] At 803, the UE inputs, to the machine learning model, the reference signal measurement of each of the plurality of reference signals taken at the first time.
[0193] At 804, the machine learning model infers a prediction for a number of channel measurement reports to be transmitted at a second time, subsequent to the first time, according to the one or more reporting modes of the reporting configuration.
[0194] At 805, the UE provides, to a network node, a request to reserve a number of resources for the second time based on the predicted number of channel measurement reports. For example, the machine learning model may predict that, at the second time, the UE will transmit five channel measurement reports based on the one or more reporting modes of the reporting configuration. Consequently, the request to reserve resources may comprise a request to reserve five UL resources such as five UL communication channels, one for each channel measurement report.
[0195] FIG. 9 illustrates an example method 900 of adjusting reporting criteria.
[0196] At 901 , the UE receives a configuration from a NW for adjusting a reporting measurement threshold of a reporting configuration for a machine learning model running on the UE. The reporting measurement threshold for a given reporting mode of a reporting configuration defines a threshold for determining whether a channel measurement of a reference signal should be reported for the given reporting mode. The machine learning model may be trained and configured to infer in accordance with one or more embodiments for adjusting a reporting measurement threshold as disclosed herein. When operating in inference mode, the machine learning model may take as inputs, the reporting configuration, reference signal measurements for each of a plurality of reference signals at a first time, and an indication of one or more of the plurality of reference signals reported to the network at a second time previous to the first time.
[0197] At 902, the UE determines, at a first time, a reference signal measurement of each of a plurality of reference signals.
[0198] At 903, the UE inputs, to the machine learning model, the reference signal measurement of each of the plurality of reference signals taken at the first time, and, an indication of one or more of the plurality of reference signals reported to the network at a second time previous to the first time.
[0199] At 904, the machine learning model infers a change in a reporting measurement threshold for the reporting configuration, and at 905, the UE adjusts the reporting measurement threshold based on the inferred change. The UE may thus, with the machine learning model and without NW intervention, autonomously adjust a reporting configuration and provide channel measurement reports according to the adjusted reporting configuration.
[0200] Embodiments of the present disclosure including functions, processes, and operations, may be implemented in software, hardware, application logic or a combination of software, hardware and application logic. Thesoftware, application logic and / or hardware may reside on memory, or any computer media. In an example embodiment, the application logic, software or an instruction set is maintained on any one of various conventional computer-readable media. In the context of this application, a “memory” or “computer-readable medium” may be any non-transitory media or means that contains, stores, communicates, propagates or transports the instructions for use by or in connection with an instruction execution system, apparatus, or device, such as a computer.
[0201] Reference to, where relevant, “computer-readable medium”, “computer program product”, “tangibly embodied computer program” etc., or a “processor” or “processing circuitry” etc. should be understood to encompass not only computers having differing architectures such as single / multi-processor architectures and sequencers / parallel architectures, but also specialized circuits such as field programmable gate arrays FPGA, application specify circuits ASIC, signal processing devices / apparatus and other devices / apparatus. References to computer program, instructions, code etc. should be understood to express software for a programmable processor firmware such as the programmable content of a hardware device / apparatus as instructions for a processor or configured or configuration settings for a fixed function device / apparatus, gate array, programmable logic device / apparatus, etc.
[0202] As used in the present disclosure, the term “circuitry”, for example in the hardware processing circuitry that may be used to train and implement AI / ML models in accordance with aspects and embodiments of the present disclosure, may refer to one or more or all of the following: (a) hardware-only circuit implementations (such as implementations in only analog and / or digital circuitry) and (b) combinations of hardware circuits and software, such as (as applicable): (I) a combination of analog and / or digital hardware ci rcuit(s) with software / firmware and (ii) any portions of hardware processor(s) with software (including digital signal processor(s)), software, and memory(ies) that work together to cause an apparatus, such as a mobile phone or server, to perform various functions) and (ill) hardware ci rcuit(s) and or processor(s), such as a microprocessor(s) or a portion of a microprocessors), that requires software (e.g., firmware) for operation, but the software may not be present when it is not needed for operation. This definition of circuitry applies to all uses of this term in this application, including in any claims. As a further example, as used in the present disclosure, the term circuitry also covers an implementation of merely a hardware circuit or processor (or multiple processors) or portion of a hardware circuit or processor and its (or their) accompanying software and / or firmware. The term circuitry also covers, for example and if applicable to the particular claim element, a baseband integrated circuit or processor integrated circuit for a mobile device or a similar integrated circuit in server, a cellular network device, or other computing or network device.
[0203] The functions, processes, and operations described herein may be performed in a different order, or may be performed concurrently with each other, or a combination thereof. Furthermore, one or more of the functions, processes, and operations may be optional or may be combined. It will be appreciated that the method diagrams shown in FIGS. 2, 3C, 4B, 5C, and 6 and the various embodiments described with reference thereto, are examples only. Variousoperations and processes depicted therein may be omitted, may be reordered, may be combined, or a combination of reordered and combined.
[0204] The scope of the claims should not be limited by the preferred embodiments set forth in the examples but should be given the broadest interpretation consistent with the description as a whole.
Claims
Claims1. An apparatus, comprising:at least one processor; andat least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to:receive a configuration from a network for reporting measurements with a machine learning model running on the apparatus, the machine learning model taking as input a reporting mode and reference signal measurements;determine a reference signal measurement of each of a plurality of reference signals; input, to the machine learning model, the reference signal measurement of each of the plurality of reference signals;infer, by the machine learning model, a reporting indicator for each respective one of the plurality of reference signals, the reporting indicator indicating whether a channel measurement report is to include information of the respective one of the plurality of reference signals; andtransmit, to a network node, the channel measurement report for one or more of the plurality of reference signals according to a respective reporting indicator.
2. The apparatus of claim 1, wherein the channel measurement report includes up to a maximum number of the plurality of reference signals according to the respective reporting indicator.
3. The apparatus of claim 1 or claim 2, wherein the machine learning model further takes as input one or more previous reporting indicators previously inferred by the machine learning model.
4. The apparatus of claim 3, wherein the one or more previous reporting indicators were inferred by the machine learning model within a preconfigured time window of the inputting the reference signal measurement of each of the plurality of reference signals, the preconfigured time window being within 10 measurement periods of the inputting.
5. The apparatus of claim 4, wherein the apparatus receives an indication of the preconfigured time window from the network node.
6. The apparatus of any one of claims 1-5, wherein the machine learning model comprises a plurality of machine learning models.
7. The apparatus of claim 6, wherein each of the plurality of machine learning models is trained on a corresponding reporting mode of a plurality of reporting modes, and wherein the plurality of reporting modes includes at least one of:a reporting mode associated with the apparatus switching cells;a reporting mode associated with load balancing;a reporting mode associated with adding a cell to a prepared cell list; ora reporting mode associated with removing a cell from the prepared cell list.
8. The apparatus of claim 6 or claim 7, wherein the instructions, when executed by the at least one processor, further cause the apparatus to select a subset of the plurality of machine learning models for inferring the reporting indicator.
9. The apparatus of claim 8, wherein each of the subset of the plurality of machine learning models provides a numerical output.
10. The apparatus of claim 9, wherein the numerical output is normalized to a value in a range of 0 and 1.
11. The apparatus of claim 9 or 10, wherein the numerical output comprises a binary value based on a threshold value.
12. The apparatus of claim 11, wherein the binary value comprises:a 1 when the numerical output is greater than the threshold value, or a 0 when the numerical output is less than the threshold value.
13. The apparatus of any one of claims 9-12, wherein the instructions, when executed by the at least one processor, further cause the apparatus to consolidate the numerical outputs of the subset of the plurality of machine learning models based on averaging corresponding numerical outputs of the subset of the plurality of machine learning models for inferring a corresponding reporting indicator for each of the plurality of reference signals.
14. The apparatus of claim 13, wherein the instructions, when executed by the at least one processor, further cause the apparatus to evaluate each of the averaged outputs against an output threshold for inferring the corresponding reporting indicator for each of the plurality of reference signals.
15. The apparatus of claim 11 or claim 12, wherein the instructions, when executed by the at least one processor, further cause the apparatus to consolidate the numerical outputs of the subset of the plurality of machine learning models based on applying a logic function to corresponding numerical outputs of the subset of the plurality of machine learning models for inferring a corresponding reporting indicator for each of the plurality of reference signals.
16. The apparatus of claim 15, wherein the logic function comprises an AND operation or an OR operation.
17. The apparatus of any one of claims 1-16 wherein the reporting mode comprises a plurality of reporting modes.
18. The apparatus of any one of claims 1-16 wherein the reporting mode comprises at least one of a cell switch mode, a load balance mode, an add cell preparation mode, or a remove cell preparation mode.
19. The apparatus of any one of claims 1 -16 wherein the reporting mode comprises a scrambled reporting mode.
20. The apparatus of any one of claims 1 -16 wherein the received configuration identifies one or more of a first reporting mode and a second reporting mode different from the first, wherein the instructions, whenexecuted by the at least one processor, further cause the apparatus to perform the inferring with one or more machine learning models associated with the identified one or more of the first reporting mode and the second reporting mode.
21. The apparatus of any one of claims 1 -20, wherein the instructions, when executed by the at least one processor, further cause the apparatus to:detect an apparatus battery level, andtransmit a capability of the apparatus;wherein the capability comprises a request for Signal Overhead Reduction (SOR) when the apparatus battery level is below a threshold, andwherein the capability comprises an indication of using the machine learning model when the apparatus battery is above the threshold.
22. The apparatus of any one of claims 1-21, wherein the reference signal measurement comprises an L1 measurement, an L3 measurement, a reference signal received power (RSRP), a reference signal received quality (RSRQ), ora signal-to-interference-plus-noise ratio (SINR).
23. An apparatus, comprising:at least one processor; andat least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to:receive a configuration from a network for estimating resources based on a reporting configuration having one or more reporting modes and a machine learning model running on the apparatus, the machine learning model taking as input the reporting configuration and reference signal measurements;determine a reference signal measurement at a first time for each of a plurality of reference signals;input, to the machine learning model, the reporting configuration and the reference signal measurement of each of the plurality of reference signals;infer, by the machine learning model, a predicted number of channel measurement reports to be transmitted at a second time subsequent to the first time according to the one or more reporting modes, andrequest, from a network node, reservation of resources for the second time based on the predicted number.
24. The apparatus of claim 23, wherein the instructions, when executed by the at least one processor, further cause the apparatus to, at the second time:measure the reference signal measurement of each of the plurality of reference signals;input, to the machine learning model, the reporting configuration and the reference signal measurement of each of the plurality of reference signals;infer, by the machine learning model, a reporting indicator for each of the plurality of reference signals, the reporting indicator indicating whether to transmit a channel measurement report for the plurality of reference signals according to the one or more reporting modes, andtransmit, to the network node, the channel measurement reports according to the one or more reporting modes for one or more of the plurality of reference signals according to a respective reporting indicator.
25. The apparatus of claim 23 or claim 24, wherein the first time comprises a window of time.
26. The apparatus of any one of claims 23 to 25, wherein the reporting configuration comprises an L1 measurement reporting event.
27. The apparatus of any one of claims 23 to 26, wherein the resources comprise uplink resources.
28. The apparatus of any one of claims 23 to 27, wherein the instructions, when executed by the at least one processor, further cause the apparatus to:detect an apparatus battery level, andtransmit a capability of the apparatus;wherein the capability comprises a request for Signal Overhead Reduction (SOR) when the apparatus battery level is below a threshold, andwherein the capability comprises an indication of using the machine learning model when the apparatus battery level is above the threshold.
29. The apparatus of any one of claims 23 to 28, wherein the reference signal measurement comprises an L1 measurement, an L3 measurement, a reference signal received power (RSRP), a reference signal received quality (RSRQ), ora signal-to-interference-plus-noise ratio (SINR).
30. An apparatus comprising:at least one processor; andat least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to:receive a configuration from a network for adjusting a reporting measurement threshold based on a reporting configuration and a machine learning model running on the apparatus, the machine learning model taking as input the reporting configuration and reference signal measurements, the reporting measurement threshold defining whether channel measurements of a reference signal are to be reported to the network; determine a reference signal measurement at a first time for each of a plurality of reference signals; input, to the machine learning model, the reporting configuration, the reference signal measurement of each of the plurality of reference signals, and an indication of one or more of the plurality of reference signals reported to the network at a second time previous to the first time;infer, by the machine learning model, a change in a reporting measurement threshold for the reporting configuration, andadjust the reporting measurement threshold based on the inferred change.31.The apparatus of claim 30, wherein the instructions, when executed by the at least one processor, further cause the apparatus to transmit, to a network node, the channel measurement report for the plurality of reference signals for which the reference signal measurement meets the adjusted reporting measurement threshold.
32. The apparatus of claim 31, wherein the instructions, when executed by the at least one processor, further cause the apparatus to transmit to the network, the adjusted reporting measurement threshold for the reporting configuration.
33. The apparatus of claim 31, wherein the instructions, when executed by the at least one processor, further cause the apparatus to transmit, to the network, the change in the reporting measurement threshold.
34. The apparatus of any one of claims 30 to 33, wherein the reporting configuration comprises an L1 measurement reporting event.
35. The apparatus of any one of claims 30 to 34, wherein the instructions, when executed by the at least one processor, further cause the apparatus to receive, from the network, a permissible range of adjustments for adjusting the reporting measurement threshold of the reporting configuration.
36. The apparatus of any one of claims 30-35, wherein the instructions, when executed by the at least one processor, further cause the apparatus to:detect an apparatus battery level, andtransmit a capability of the apparatus;wherein the capability comprises a request for Signal Overhead Reduction (SOR) when the apparatus battery level is below the threshold, andwherein the capability comprises an indication of using the machine learning model when the apparatus battery is above the threshold.
37. The apparatus of any one of claims 30-36, wherein the reference signal measurement comprises an L1 measurement, an L3 measurement, a reference signal received power (RSRP), a reference signal received quality (RSRQ), ora signal-to-interference-plus-noise ratio (SINR).
38. The apparatus according to any one of claims 1-37, wherein the apparatus comprises a user equipment (UE).
39. A method for causing performance of an apparatus, the method including storing instructions in memory that, when executed by at least one processor of the apparatus, cause the apparatus to perform according to any one of claims 1-38.
40. A non-transitory computer-readable medium comprising instructions stored thereon for execution by at least one processor of an apparatus to cause the apparatus to perform according to any one of claims 1 to 38.
