Method and apparatus for implementing AI / ML in 5G network

Abstract

The present disclosure relates to a 5G or 6G communication system for supporting a higher data transmission rate. There is a method for supporting artificial intelligence / machine learning (AI / ML) functionality in a wireless communication system. The method includes establishing a first radio resource control (RRC) connection between the UE and a base station (BS) by using a first signaling radio bearer (SRB); receiving a configuration message including bearer configuration information from the BS via the first SRB; establishing a second radio resource control (RRC) connection between the UE and the BS by using a second SRB; and receiving at least one RRC message including data related to the AI / ML function from the BS based on the second RRC connection, wherein the second SRB is configured based on the bearer configuration information.

Description

Technical Field

[0001] This application relates to providing machine learning (ML) functions in a 5G network, including configuration and reporting, as well as signaling procedures. Background Art

[0002] The 5G mobile communication technology defines wide frequency bands, enabling high transmission rates and new services, and can be implemented not only in the "sub-6 GHz" frequency bands such as 3.5 GHz, but also in the "above-6 GHz" frequency bands (including 28 GHz and 39 GHz) known as millimeter waves. In addition, it has been considered to implement 6G mobile communication technology (referred to as the super 5G system) in the terahertz (THz) frequency band (e.g., the 95 GHz to 3 THz frequency band) in order to achieve a transmission rate fifty times faster than that of 5G mobile communication technology and an ultra-low latency that is one-tenth of that of 5G mobile communication technology.

[0003] At the beginning of the development of 5G mobile communication technology, in order to support services and meet the performance requirements related to enhanced mobile broadband (eMBB), ultra-reliable low-latency communication (URLLC), and massive machine type communication (mMTC), standardization of the following technologies has been underway: beamforming and massive MIMO to mitigate radio wave path loss in millimeter waves and increase radio wave transmission distance; supporting parameter sets (e.g., operating multiple subcarrier spacings) to effectively utilize millimeter wave resources and dynamically operate time slot formats; initial access technologies for supporting multi-beam transmission and broadband; the definition and operation of BWP (bandwidth part); new channel coding methods (such as LDPC (low-density parity-check) codes for large data transmission and polarization codes for highly reliable transmission of control information); L2 preprocessing; and network slicing for providing dedicated networks dedicated to specific services.

[0004] Currently, in view of the services supported by 5G mobile communication technology, discussions on the improvement and performance enhancement of the initial 5G mobile communication technology are underway, and physical layer standardization of technologies such as the following already exists: V2X (vehicle-to-everything) that assists autonomous vehicle driving decisions based on information about the position and status of a vehicle transmitted by the vehicle and is used to improve user convenience, NR-U (new radio unlicensed) aiming to make system operations comply with various regulatory requirements in unlicensed frequency bands, NR UE energy saving, non-terrestrial network (NTN), which is UE-satellite direct communication for providing coverage in areas where terrestrial network communication is unavailable, and positioning.

[0005] In addition, technologies in the air interface architecture / protocol aspect are being continuously standardized. For example, Industrial Internet of Things (IIoT) is used to support new services through interoperability and integration with other industries, IAB (Integrated Access and Backhaul) is used to provide nodes for network service area expansion by supporting wireless backhaul links and access links in an integrated manner, including mobility enhancements such as conditional handover and DAPS (Dual Active Protocol Stack) handover, and two-step random access (2-step RACH for NR) for simplifying the random access process. System architectures / services for the following technologies are also being continuously standardized: 5G baseline architecture (e.g., service-based architecture or service-based interface) for combining network function virtualization (NFV) and software-defined network (SDN) technologies; and mobile edge computing (MEC) for receiving services based on UE location.

[0006] With the commercialization of 5G mobile communication systems, the exponentially growing connected devices will be connected to the communication network, and thus it is expected that the functions and performance of 5G mobile communication systems and the integrated operation of connected devices will need to be enhanced. For this purpose, new research related to the following technologies is planned: extended reality (XR) to effectively support AR (augmented reality), VR (virtual reality), MR (mixed reality), etc.; improving 5G performance and reducing complexity by leveraging artificial intelligence (AI) and machine learning (ML); AI service support; metaverse service support and drone communication.

[0007] In addition, this development of 5G mobile communication systems will not only serve as the basis for developing the following technologies: new waveforms for coverage in the terahertz band for 6G mobile communication technologies, multi-antenna transmission technologies such as full-dimensional MIMO (FD-MIMO), array antennas, and large antennas, metasurface-based lenses and antennas for improving the coverage of terahertz band signals, high-dimensional spatial multiplexing technologies using OAM (orbital angular momentum), and RIS (reconfigurable intelligent surface), but also serve as the basis for developing the following technologies: full-duplex technologies for increasing the frequency efficiency of 6G mobile communication technologies and improving the system network, AI-based communication technologies for achieving system optimization by leveraging satellites and AI (artificial intelligence) from the design phase and internalizing end-to-end AI support functions, and next-generation distributed computing technologies for implementing services with a complexity level exceeding the limits of UE operating capabilities by leveraging ultra-high-performance communication and computing resources. Summary of the Invention

[0008]

Technical Problem

[0009] Currently, there is a need to enhance the implementation of AI / ML in 5G networks.

[0010]

Technical Solution

[0011] In a first method of the present technology, a computer-implemented method for configuring a user equipment to support artificial intelligence / machine learning (AI / ML) functions in a wireless communication network is provided. The method includes: establishing a radio resource control (RRC) connection from the user equipment to the network via a first signaling radio bearer (SRB); receiving, at the user equipment via the first signaling radio bearer, a configuration message, where the configuration message includes bearer configuration information for a second radio bearer to support AI / ML functions in the network by enabling transmission of data related to AI / ML functions via the second radio bearer; establishing a radio resource control (RRC) connection from the user equipment to the network via the second radio bearer to enable RRC messages including data related to AI / ML functions to be transmitted from or received by the user equipment.

[0012] The user equipment can be any suitable device used by a user to connect to the network and is typically mobile, for example, a mobile device or a terminal. The wireless communication network can also be referred to as a mobile communication network or a radio communication network. Once the user equipment is connected to the network, the network can be considered to include the user equipment and other nodes, as described below.

[0013] The AI / ML functions in the network can include at least one AI / ML model located within the network and / or at the user equipment. For example, the AI / ML model can be used to perform network energy saving, load balancing, or mobility optimization within the network. The AI / ML functions in the network can include AI / ML model training, AI / ML model inference, and / or AI / model configuration. The AI / ML functions in the network can also include transmitting AI / ML data around the network and / or between the network and the user equipment.

[0014] The network can include multiple nodes (also referred to as base stations or cells) and a central server (e.g., an operations, administration, and management OAM server), and an AI / ML model can be present at one or more nodes and / or in the central server. The network can be a next-generation mobile communication system, hereinafter referred to as an NR or 5G network. The data related to AI / ML functions can include the input and output data of the AI / ML model located within the network or at the user equipment. The input can include one or more of the data for training the AI / ML model and the data used by the trained AI / ML model during inference. The output data can include the output data from the AI / ML model during inference and the feedback data from the AI / ML model.

[0015] The method may include: receiving measurement configuration information at a user equipment; and configuring the user equipment using the received measurement configuration information, wherein the measurement configuration information includes information on configuring the user equipment to collect and report measurement data required for an AI / ML model. The measurement configuration information may be received in a configuration message together with bearer configuration information for a second radio bearer. In other words, the bearer configuration information and the measurement configuration information may be received simultaneously. Alternatively, the measurement configuration information may be received in a second separate reconfiguration message. The measurement configuration information may be received from a network (e.g., from a node including the AI / ML model or elsewhere in the network).

[0016] Accordingly, configuring the user equipment using the measurement configuration information may include configuring the user equipment to collect the required measurement data. The measurement data may include at least one measurement (e.g., radio measurement, intra-frequency measurement, inter-frequency measurement, inter-RAT measurement, channel state information reference signal measurement, synchronization signal block measurement, synchronization information, or physical broadcast channel measurement). The measurement data may include location information of the user equipment, e.g., positioning, location measurement, and / or mobility measurement. As an example, when the model is predicting an optimized network energy saving decision, configuring the user equipment may include configuring radio measurements (e.g., RSRP, RSRQ, and SINR) related to the serving cell and neighboring cells associated with the user equipment, as well as cell-level and beam-level UE measurements. When the model is used for load balancing decisions, configuring the user equipment may include radio measurements (e.g., RSRP, RSRQ, and SINR) related to the serving cell and neighboring cells associated with the user equipment, as well as measurements tracking the mobility history of the user equipment. When the model is used for mobility optimization, configuring the user equipment may include configuring measurements tracking the mobility history of the user equipment, as well as radio measurements (e.g., RSRP, RSRQ, SINR) related to the serving cell and neighboring cells associated with the UE location information.

[0017] Measurement configuration information may include some or all of the following parameters: a measurement object that defines the object to be measured, a reporting configuration, a measurement identity that links each reporting configuration to the measurement object, a measurement configuration that defines any filtering to be applied, and a measurement gap that defines the period for which the user equipment makes reports. Each measurement object may have one or more reporting configurations. The reporting configuration may include a reporting criterion, which is the criterion for triggering the user equipment to send a measurement report to the network (e.g., directly to an AI / ML model within the network or to different locations in the network as specified in the reporting configuration). The criterion may be periodic or conditional, e.g., in response to an event such as receiving a message, or in response to the user equipment meeting a condition. For example, it may be determined whether the condition is met by comparing a measurement with a threshold, and the measurement for comparison may be related to one or more of RSRP (Reference Signal Received Power), RSRQ (Reference Signal Received Quality), or SINR (Signal-to-Interference-plus-Noise Ratio). By including the criterion for triggering a report, the frequency of messages being transmitted can be controlled.

[0018] Another way to control the frequency of transmitted messages is to configure the message structure for reporting input data. The measurement configuration information may also define a message structure for simultaneously reporting multiple pieces of data in a single message transmitted from the user equipment to the network, where the message structure includes sequence information for each piece of data in the message, whereby the receiver can sort these pieces of data in sequential order (e.g., chronological order). By including multiple pieces of AI / ML data information in this newly proposed message structure by the user equipment, frequent reporting can be avoided. The sequence information may be a serial number or timing information for each piece of AI / ML data. The number of AI / ML data to be included in the report message may also be configured by an RRC message. Each piece of data may be for the same target (measurement object) or for the same AI / ML model.

[0019] The first signaling radio bearer may be SRB1 and may be used for the first messages (RRC and / or non-access stratum (NAS) messages) between the user equipment and the network, which are used to establish and support the connection of the user equipment to the network. The second signaling radio bearer may be a standard signaling radio bearer (e.g., SRB2), a newly configured signaling radio bearer (e.g., SBR5), or other appropriate SRB number), or may be a data radio bearer. By using separate radio bearers to establish the connection, the messages transmitted via the first signaling radio bearer may take precedence over the messages transmitted via the second signaling radio bearer. The or each signaling radio bearer may be referred to as a communication channel.

[0020] When using a newly configured signaling radio bearer SBR5 and / or a newly configured data radio bearer to send and / or receive messages of AI / ML-related information, the network may configure the priorities of these messages. The method may include: receiving, at a user equipment, priority ranking configuration information from the network, the priority ranking configuration information defining a lower priority for messages including data related to AI / ML functions than for other messages transmitted between the user equipment and the network. The user equipment may be configured using the priority ranking configuration information, i.e., the user equipment may be configured to implement the priority ranking configuration information. The priority ranking configuration information may include logical channel priority ranking parameters (i.e., the allocation of priorities and priority bit rates (PBR)) to control the priority ranking of data transmission. By prioritizing other messages, especially those related to establishing a connection within the network, the transmission of potentially large amounts of AI / ML data can be controlled to avoid affecting the network's control of the user equipment.

[0021] Data security between the network and the user equipment is very important. Thus, the method may include activating access stratum security before establishing an RRC connection via a second radio bearer. Once the AS security is activated, all RRC messages on standard signaling radio bearers (e.g., SRB1, SRB2, SRB3, and SRB4) are typically integrity protected and encrypted, for example, by PDCP (Packet Data Convergence Protocol). However, when the second radio bearer is a newly configured signaling radio bearer, the method may include: receiving security configuration information that defines whether the user equipment is to apply integrity protection and / or encryption to messages between the user equipment and the network; and configuring the user equipment according to the security configuration information. The security configuration information defines whether the message is not integrity protected and not encrypted, not integrity protected but encrypted, integrity protected but not encrypted, or integrity protected and encrypted. In other words, RRC integrity protection and encryption can be activated and deactivated based on the received configuration information. Using an option that does not include both integrity protection and encryption can reduce the processing burden at the user equipment and avoid over-security protection in the network.

[0022] The configuration message received at the user equipment via a first signaling radio bearer may be any suitable RRC message, e.g., an RRCReconfiguration, RRCSetup, RRCResume, RRCRelease message, or a newly defined RRC message. Configuration information, measurement configuration information, priority ranking configuration information, and security configuration information may be transmitted simultaneously in the same configuration message. Alternatively, separate messages may be used for some or each type of configuration information.

[0023] Each cell within a wireless communication network can broadcast system information that indicates whether the cell supports AI / ML capabilities, e.g., indicating whether the cell has AI / ML that needs to be trained or can be used for inference. When a user equipment obtains the system information, the user equipment can receive a configuration message and / or establish an RRC connection with the network via a second radio bearer through a cell that supports AI / ML capabilities. When the user equipment has previously connected to the network using a specific cell, the user equipment can select a cell that supports AI / ML capabilities and establish a new connection to the network via the selected cell. The user equipment can be configured to send capability information to the network (e.g., to the selected cell), where the capability information includes the user equipment's capability to support the AI / ML capabilities of the cell (network).

[0024] Once the user equipment is configured, for example, using some or all of configuration information, measurement configuration information, prioritization configuration information, and security configuration information, a method is also provided for supporting AI / ML capabilities within the network by the user equipment by transmitting data related to the AI / ML capabilities to the network (e.g., a cell within the network that has an AI / ML model). The user equipment may have been configured to measure the data being transmitted prior to transmission, as described above. The user equipment can also receive messages from the network to support the AI / ML capabilities, e.g., trigger measurement and reporting of AI / ML data and / or reconfigure the AI / ML capabilities at the user equipment.

[0025] The methods described above focus on the steps performed by the user equipment, but it should be understood that corresponding steps are also performed by the network, particularly by nodes within the network. Thus, according to another aspect of the present technology, a method for configuring a user equipment to support artificial intelligence / machine learning (AI / ML) capabilities in a wireless communication network is described. The method includes: establishing a first radio resource control (RRC) connection from the user equipment to the network using a first signaling radio bearer; transmitting, via the first signaling radio bearer, a configuration message from the network to the user equipment, where the configuration message includes configuration information for configuring a second radio bearer to transmit data related to the AI / ML capabilities between the user equipment and the network, thereby configuring the user equipment to support the AI / ML capabilities in the network; and receiving, at the network, an acknowledgement that the user equipment has been configured to enable the transmission or reception, by the user equipment, of at least one RRC message including data related to the AI / ML capabilities using the second data bearer.

[0026] Once the user equipment is configured, according to another aspect of the present technology, a method for applying AI / ML functions within a network is described. For example, a node within the network having an AI / ML model can receive AI / ML data from the user equipment and can use the received AI / ML data to train the AI / ML model. Alternatively, a node within the network having an AI / ML model can receive AI / ML data from the user equipment and can infer an output from the AI / ML model.

[0027] The or each AI / ML model in the system can be a pre-trained general ML model that can be obtained from a data store, and / or can be a general model trained to full precision by a central server on data accessible to the central server. The AI / ML model in the form of a neural network includes multiple layers, where each layer has a set of general weights. One or more local models can also exist within the system, and each local model can be a similar model with a set of local weights for each layer. Each local model can be trained using a set of local data samples that can be stored in a database on the user equipment and / or measured by the user equipment.

[0028] In any arrangement, the AI / ML data collected or measured can include the location information of the user equipment and at least one of one or more measurements from the user equipment and / or one or more nodes within the network. The one or more measurements can include one or more of radio measurements, intra-frequency measurements, inter-frequency measurements, inter-RAT measurements, channel state information reference signal measurements, synchronization signal block measurements. As described above, the location information can include at least one of location, location measurement, and mobility measurement. Similarly, in any arrangement, at least one configuration message can further include: prioritization configuration information for configuring the user equipment to define a lower priority for messages including data related to AI / ML functions than other messages transmitted between the user equipment and the network; and / or security configuration information for configuring the user equipment to apply integrity protection and / or encryption to messages between the user equipment and the network.

[0029] The user equipment has three states. In the first state, the user equipment is connected to the network and the connection is active (i.e., messages are being transmitted between the user equipment and the network). Once the user equipment has been connected to the network, the user equipment can transition to a second state where the connection is inactive. There is also a third state before the connection is established and the user equipment is idle. These states can be referred to as RRC_Connected, RRC_INACTIVE, and RRC_IDLE. Appropriate RRC messages from the network to the user equipment will cause the user equipment to transition between states. For example, a release message (e.g., RRC_release) received by the user equipment from the network will cause the user equipment (from the connected or inactive state) to transition to the idle state. When the user equipment is in the connected state, messages with AI / ML data can be transmitted between the user equipment and the rest of the network to support AI / ML functions within the network, as described above.

[0030] In a related method of the present technology, a computer-readable storage medium is provided, which includes instructions that, when executed by a processor, cause the processor to execute any of the methods described herein.

[0031] As those skilled in the art will understand, the present technology can be embodied as a system, a method, or a computer program product. Therefore, the present technology can take the form of an all-hardware embodiment, an all-software embodiment, or an embodiment combining software and hardware aspects. All of the above methods can be considered computer-implemented methods.

[0032] In addition, the present technology can take the form of a computer program product embodied in a computer-readable medium, on which computer-readable program code is embodied. The computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. The computer-readable medium can be, for example but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing.

[0033] The computer program code for performing the operations of the present technology can be written in any combination of one or more programming languages, including object-oriented programming languages and traditional procedural programming languages. The code components can be embodied as procedures, methods, etc., and can include sub-components, which can take the form of instructions or instruction sequences at any level of abstraction from direct machine instructions of a native instruction set to high-level compiled or interpreted language constructs.

[0034] Embodiments of the present technology also provide a non-transitory data carrier carrying code that, when implemented on a processor, causes the processor to execute any of the methods described herein.

[0035] The technology also provides processor control code to implement the above - mentioned method, for example, on a general - purpose computer system or a digital signal processor (DSP). The technology also provides a carrier carrying the processor control code to implement any one of the above - mentioned methods at runtime, especially on a non - transitory data carrier. The code can be provided on a carrier such as a disk, a microprocessor, a CD - ROM or a DVD - ROM, on a programmable memory such as non - volatile memory (e.g., flash memory) or read - only memory (firmware), or on a data carrier such as an optical or electrical signal carrier. The code (and / or data) for implementing embodiments of the technology described herein can include source code, object code, or executable code in a conventional programming language (interpreted or compiled), such as Python, C, or assembly code, code for setting or controlling an ASIC (application - specific integrated circuit) or an FPGA (field - programmable gate array), or code in a hardware description language such as Verilog(RTM) or VHDL (very high - speed integrated circuit hardware description language). As those skilled in the art will understand, such code and / or data can be distributed among multiple interconnected components that communicate with each other. The technology can include a controller that includes a microprocessor, a working memory, and a program memory coupled to one or more of the components of the system.

[0036] Those skilled in the art will also be clear that all or part of the logical method according to an embodiment of the present technology can be appropriately embodied in a logical device including logical elements to perform the steps of the above - mentioned method, and such logical elements can include, for example, components in a programmable logic array or an application - specific integrated circuit, such as logic gates. Such a logical arrangement can also be embodied in an enabling element for temporarily or permanently establishing a logical structure in such an array or circuit using, for example, a virtual hardware descriptor language, which can be stored and transmitted using a fixed or transportable carrier medium.

[0037] In an embodiment, the present technology can be implemented in the form of a data carrier having functional data thereon, the functional data including functional computer data structures to enable the computer system to execute all the steps of the above - mentioned method when loaded into and operated on a computer system or a network.

[0038] The above - mentioned method can be executed in whole or in part on a device (i.e., a user device in the form of an electronic device).

[0039] According to another aspect of the present technology, there is provided a user equipment, which includes: at least one receiver for receiving messages from a wireless communication network; at least one transmitter for transmitting messages to the wireless communication network; and a processor configured to establish a radio resource control (RRC) connection from the user equipment to the network via a first signaling radio bearer of the at least one receiver and the at least one transmitter; and configure the user equipment using configuration information received in a configuration message via the first signaling radio bearer at the user equipment. The configuration message includes bearer configuration information for a second radio bearer to support an AI / ML function in the network by enabling transmission of data related to the AI / ML function via the second radio bearer; and configuring the user equipment includes establishing a radio resource control (RRC) connection from the user equipment to the network via the second radio bearer so that RRC messages including data related to the AI / ML function can be transmitted from or received by the user equipment.

[0040] The model can be processed by an artificial intelligence dedicated processor designed in a hardware structure designated for artificial intelligence model processing. The artificial intelligence model can be obtained through training. Here, "obtained through training" means obtaining a predefined operation rule or an artificial intelligence model configured to perform a desired feature (or purpose) by training a basic artificial intelligence model with multiple pieces of training data according to a training algorithm. The artificial intelligence model can include multiple neural network layers. Each of the multiple neural network layers includes multiple weight values and performs neural network calculations by calculating between the calculation results of the previous layer and the multiple weight values.

[0041] As described above, the present technology can be implemented using an AI model. Functions associated with AI can be executed by a non-volatile memory, a volatile memory, and a processor. The processor can include one or more processors. At this time, the one or more processors can be a general-purpose processor such as a central processing unit (CPU), an application processor (AP), etc., a graphics processing unit only, such as a graphics processing unit (GPU), a vision processing unit (VPU), and / or a dedicated AI processor such as a neural processing unit (NPU). The one or more processors control the processing of input data according to predefined operation rules or an artificial intelligence (AI) model stored in the non-volatile memory and the volatile memory. The predefined operation rules or the artificial intelligence model are provided through training or learning. Here, provided through learning means obtaining a predefined operation rule or an AI model with desired characteristics by applying a learning algorithm to multiple learning data. The learning can be performed in the device itself that executes AI according to the present disclosure, and / or can be implemented by a separate server and / or system.

[0042] An AI model can be composed of multiple neural network layers. Each layer has multiple weight values and performs layer operations through the calculations of the previous layer and the operations of multiple weights. Examples of neural networks include, but are not limited to, convolutional neural networks (CNNs), deep neural networks (DNNs), recurrent neural networks (RNNs), restricted Boltzmann machines (RBMs), deep belief networks (DBNs), bidirectional recurrent deep neural networks (BRDNNs), generative adversarial networks (GANs), and deep Q networks.

[0043] A learning algorithm is a method that uses multiple training data to train a predetermined target device (e.g., a node) to cause, allow, or control the target device to make a determination or prediction. Examples of learning algorithms include, but are not limited to, supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Reference will now be made to the drawings, and the implementation of the present technology will be described only by way of example, in which:

[0045] Figure 1 is a schematic diagram of an LTE system according to an embodiment disclosed herein;

[0046] Figure 2 shows a radio protocol structure for an LTE system according to an embodiment disclosed herein;

[0047] Figure 3 is a schematic diagram of a wireless communication system according to an embodiment disclosed herein;

[0048] Figure 4 shows a radio protocol structure for a wireless communication system according to an embodiment disclosed herein;

[0049] Figure 5 is a schematic block diagram of a functional framework 500 for implementing artificial intelligence or machine learning in a radio access network according to an embodiment disclosed herein;

[0050] Figure 6 shows a state diagram of a user equipment according to an embodiment disclosed herein;

[0051] Figure 7 shows a method by which a user equipment can obtain system information from a network according to an embodiment disclosed herein;

[0052] Figure 8 is a flowchart showing a process by which a UE switches from RRC idle mode to RRC connected mode according to an embodiment disclosed herein;

[0053] Figure 9is a flowchart showing an example use case in which AI / ML model training is completed at a central server and AI / ML model inference is completed at one or more nodes of a network, as disclosed herein;

[0054] Figure 10 is a flowchart showing another example use case in which AI / ML model training is completed at a node and AI / ML model inference is completed at one or more nodes of a network, as disclosed herein;

[0055] Figure 11 is a flowchart summarizing a method of configuring a user equipment to support AI / ML functionality, as disclosed herein;

[0056] Figure 12 is a block diagram of a system according to an embodiment disclosed herein, the system including nodes in a network and a user equipment connected to the nodes;

[0057] Figure 13 is a block diagram showing a terminal (or user equipment (UE)) according to an embodiment disclosed herein; and

[0058] Figure 14 is a block diagram showing a base station (BS) according to an embodiment disclosed herein.

[0059] It can be noted that, to the extent possible, the same reference numerals have been used to denote the same elements in the figures. Further, those of ordinary skill in the art will understand that the elements in the figures are shown for simplicity and may not necessarily be drawn to scale. For example, the sizes of some elements in the figures may be exaggerated relative to other elements to help improve understanding of aspects of the present invention. Additionally, one or more elements may have been represented in the figures by conventional symbols, and the drawings may show only specific details relevant to understanding the embodiments of the present invention so as not to obscure the drawings with details that would be readily apparent to those of ordinary skill in the art benefited by the description herein. Detailed Description

[0060] To meet the requirements of 5G networks for key performance and the needs of an unprecedentedly growing number of mobile subscribers, millions of base stations (BSs) are being deployed. Such rapid growth brings problems of high energy consumption, CO2 emissions, and operating expenditure (OPEX). Therefore, energy saving is an important use case, having mechanisms operating on different time scales and can involve different layers of the network.

[0061] Cell activation / deactivation is an energy-saving solution in the spatial domain that uses traffic offloading in a hierarchical structure to reduce the energy consumption of the entire radio access network (RAN). When the expected traffic is below a fixed threshold, cells can be turned off, and the served UEs can be offloaded to new target cells. Efficient energy consumption can also be achieved in other ways, such as reducing the load, modifying the coverage area, or other RAN configuration adjustments. The optimal energy-saving decision depends on many factors, including the load conditions of different RAN nodes, RAN node capabilities, KPI / QoS (Quality of Service) requirements, the number of active UEs and UE mobility, cell utilization, etc.

[0062] However, identifying actions aimed at improving energy efficiency is not an easy task. Incorrectly turning off cells can severely degrade network performance because the remaining active cells need to serve additional traffic. Incorrect traffic offloading actions can lead to a decrease in energy efficiency rather than an increase. Current energy-saving solutions are vulnerable to potential problems.

[0063] An example problem is inaccurate cell load prediction because currently, energy-saving decisions rely on the current traffic load without considering future traffic loads. Another example is the goal conflict between system performance and energy efficiency. Maximizing the key performance indicators (KPIs) of the system usually comes at the expense of energy efficiency. Similarly, the most energy-efficient solution may affect system performance. Therefore, it is necessary to balance and manage the trade-off between the two. Another example is the conventional energy-saving related parameter adjustment. Energy-saving related parameter configuration is set through traditional operations, for example, based on different thresholds of cell load for cell on / off, which is a somewhat rigid mechanism because it is difficult to set reasonable thresholds. Another example is an action that may result in a local (e.g., limited to a single RAN node) increase in energy efficiency while causing an overall (e.g., involving multiple RAN nodes) decrease in energy efficiency.

[0064] In addition to the need for energy saving, the rapid traffic growth and the use of multiple frequency bands in commercial networks make it challenging to direct traffic in a balanced manner. To address this problem, load balancing has been proposed. The goal of load balancing is to evenly distribute the load between cells and between regions of a cell, or to transfer a portion of the traffic from a congested cell or a congested area of a cell, or to offload users from one cell, cell region, carrier, or RAT to improve network performance. This can be accomplished by optimizing handover parameters and handover actions. The automation of such optimization can provide a high-quality user experience, while increasing system capacity and also minimizing human intervention in network management and optimization tasks.

[0065] However, the optimization of load balancing is not an easy task. Currently, load balancing decisions relying on the cell load status of the current / past state are insufficient. The traffic load and resource status of the network change rapidly, especially in cases with high mobility and a large number of connections, which may lead to ping-pong handovers between different cells, cell overload, and a reduction in the user service quality. When performing load balancing, it is difficult to ensure the overall network and service performance. For load balancing, user equipment (UE) in a congested cell can be offloaded to a target cell through a handover process or by adjusting the handover configuration. For example, if a UE with a time-varying traffic load is offloaded to a target cell, the target cell may become overloaded due to the newly arrived large amount of traffic. It is difficult to determine whether the service performance after the offloading action meets the expected target.

[0066] In addition to the need for energy conservation and load balancing, mobility management is a solution to ensure service continuity during mobility by minimizing dropped calls, RLF, unnecessary handovers, and ping-pong. For future high-frequency networks, as the coverage range of a single node decreases, the frequency of UE handovers between nodes becomes higher, especially for high-mobility UEs. Additionally, for applications characterized by strict quality of service (QoS) requirements (such as reliability, latency, etc.), QoE is sensitive to handover performance, making it necessary for mobility management to avoid unsuccessful handovers and reduce latency during the handover process. However, for conventional methods, it is challenging to achieve almost zero-failure handovers with a trial-and-error-based approach. Unsuccessful handover situations are the main cause of packet loss or additional latency during the mobility period, which is undesirable for packet-loss-intolerant and low-latency applications. Additionally, the effectiveness of feedback-based adjustment may be weak due to the randomness and volatility of the transmission environment. In addition to the baseline situation of mobility, the optimization areas of mobility also include dual connectivity, CHO, and DAPS, each of which has additional aspects to be addressed in the optimization of mobility.

[0067] There are various mobility-related contingencies. One example is a too-late handover within the system. In this contingency, a radio link failure (RLF) occurs after the UE has stayed in a cell for a long time; the UE attempts to re-establish a radio link connection in a different cell. Another example is a too-early handover within the system. In this contingency, an RLF occurs shortly after successfully handovering from a source cell to a target cell, or a handover failure occurs during the handover process; the UE attempts to re-establish a radio link connection in the source cell. Another example is a handover to the wrong cell within the system. In this contingency, an RLF occurs shortly after successfully handovering from a source cell to a target cell, or a handover failure occurs during the handover process; the UE attempts to re-establish a radio link connection in a cell other than the source cell and the target cell. Another example is any potential problem during an otherwise successful handover.

[0068] Therefore, the present applicant has recognized the need for improvements in 5G networks.

[0069] Broadly speaking, the present technology generally relates to supporting AI / ML capabilities in 5G networks. This includes AI / ML configuration / reporting, signaling procedures, and operations at the user equipment (UE). To reduce the UE processing burden, new security protection (integrity protection or encryption) mechanisms are proposed, different from the mandatory security protection for signaling radio bearers (SRBs). To avoid affecting network control of the UE, a new prioritization mechanism is introduced by defining rules for SRBs or introducing new SRBs or allocating separate data radio bearers (DRBs). To avoid frequent reporting, a new message structure is designed to report multiple data for the same target in chronological order in one message. To support UE mobility for AI / ML capabilities, radio resource control (RRC) procedures are proposed. To increase the reliability of AI / ML capabilities, a retransmission mechanism is proposed.

[0070] Definitions of Terms, Symbols, and Acronyms

[0071] Acronyms

[0072]

[0073]

[0074] Data collection: Data collected from network nodes, management entities, or user equipment (UE), serving as the basis for AI / ML model training, data analysis, and inference. Data collection is the function of providing input data to the model training and model inference functions. AI / ML algorithm-specific data preparation (e.g., data preprocessing and cleaning, formatting, and transformation) is not performed in the data collection function. Examples of input data can include measurements (or responses or information or reports) from the UE or different network entities, feedback from Actors, outputs from AI / ML models.

[0075] Training data: Data required as input to the AI / ML model training function.

[0076] Inference data: Data required as input to the AI / ML model inference function.

[0077] AI / ML (Artificial Intelligence / Machine Learning) Model: A data-driven algorithm that applies machine learning techniques to generate a set of outputs consisting of prediction information and / or decision parameters based on a set of inputs. It also refers to AI / ML (model) training or inference or training data or prediction information or decision parameters or a set of inputs. AI / ML data refers to AI / ML (model) training or inference or training data or prediction information or decision parameters or a set of inputs used for AI / ML or the AI / ML model. AI / ML configuration includes the type of AI / ML model, model inference, model training, algorithms, or how to construct RRC (Radio Resource Control) messages (including AI / ML data) for response (or reporting) (e.g., what information should be included), or how to send RRC messages from a transmitter (UE or gNB) to a receiver (gNB or UE) (e.g., periodicity, timer-related information, etc.). AI / ML data can refer to measurement reports (e.g., for beams or SSB (Synchronization Signal Block) or CSI (Channel State Information) or CRS (Cell-Specific Reference Signal)) and CSI reports. AI / ML configuration can include measurement configuration (e.g., for beams or SSB or CSI or CRS) and CSI measurement configuration (e.g., aperiodic CSI, periodic CSI, etc.). AI / ML functionality means all this information (e.g., AI / ML data and AI / ML configuration), measurement configuration and reporting, requests / reports for training data, and the training and inference processes described below.

[0078] AI / ML Training: An online or offline process of training an AI / ML model by learning the features and patterns that best represent the data and obtaining a trained AI / ML model for inference. Model training is the function that performs AI / ML model training, validation, and testing, which can generate model performance metrics as part of the model testing process. If needed, the model training function is also responsible for data preparation (e.g., data preprocessing and cleaning, formatting, and transformation) based on the training data delivered by the data collection function.

[0079] Model Deployment / Update: Used to initially deploy a trained, validated, and tested AI / ML model to the model inference function, or to deploy an updated model to the model inference function.

[0080] AI / ML Inference: The process of using a trained AI / ML model to make predictions or guide decisions based on the collected data and the AI / ML model. Model inference is the function that provides the AI / ML model inference output (e.g., predictions or decisions). When applicable, the model inference function can provide model performance feedback to the model training function. If needed, the model inference function is also responsible for data preparation (e.g., data preprocessing and cleaning, formatting, and transformation) based on the inference data delivered by the data collection function.

[0081] Output: The inference output of the AI / ML model generated by the model inference function. The details of the inference output are use-case specific.

[0082] Model performance feedback: It can be used to monitor the performance of the AI / ML model (when available).

[0083] Actor: This is the function that receives the output from the model inference function and triggers or executes the corresponding actions. An actor can trigger actions against other entities or itself.

[0084] Feedback: Information that may be required to derive training data, inference data, or to monitor the performance of the AI / ML model and its impact on the network by updating KPIs (Key Performance Indicators) and performance counters.

[0085] Federated learning (also known as collaborative learning) is a machine learning technique for training algorithms on multiple decentralized edge devices or servers while keeping local data samples without exchanging them. This approach contrasts with traditional centralized machine learning techniques that upload all local datasets to a single server and more classical decentralized methods that typically assume local data samples are identically distributed. Federated learning enables multiple actors (or UEs) to build a common and robust machine learning model without sharing data, thus addressing key issues such as data privacy, data security, data access rights, and access to heterogeneous data. Applications of federated learning span many industries, including defense, telecommunications, IoT, and pharmaceuticals.

[0086] Split learning is a newly developed technique that allows participating entities to train a machine learning model without sharing any raw data. In principle, both federated learning and split learning have similar properties in terms of the benefit of generating data privacy effects.

[0087] A signaling radio bearer (SRB) is defined as a radio bearer (RB) used only for transmitting RRC (Radio Resource Control) and NAS (Non-Access Stratum) messages. More specifically, the following SRBs are defined:

[0088] SRB0 is used for RRC messages using the CCCH (Common Control Channel) logical channel.

[0089] SRB1 is used for the first RRC message (which may include piggybacked NAS messages) and NAS messages before the establishment of SRB2, and all of them use the DCCH logical channel. It can be used for the second RRC message including AI / ML related information (e.g., AI / ML data, training data, update information, response, offset of AI / ML, etc.), but the priority of the second RRC information may be lower than that of the first RRC message (e.g., piggybacked NAS messages, NAS messages before the establishment of SRB2, RRC messages for setting (re)connection and (re)configuration, i.e., RRCSetupRequest, RRCSetup, RRCSetupComplete, RRCResumeRequest, RRCResume, RRCResumeComplete, RRCReconfiguration, RRCReconfigurationComplete, RRCReestablishment, RRCReestablishmentRequest, etc.), because the first RRC message is much more important than the second RRC message for supporting the connection management and mobility support of the UE. When the first message and the second message are generated together or simultaneously (or coexist in the buffer), the first message can be prioritized over the second RRC message and can be submitted to the lower layer (e.g., PDCP layer) and sent first.

[0090] SRB2 is used for NAS messages and RRC messages including recorded measurement information or AI / ML related information (e.g., AI / ML data, training data, update information, response, offset of AI / ML, etc.), and all of them use the DCCH logical channel. SRB2 has a lower priority than SRB1 and can be configured by the network after the activation of AS (access stratum) security;

[0091] SRB3 is used for specific RRC messages that all use the DCCH logical channel when the UE is in (NG)EN-DC or NR-DC, or RRC messages including AI / ML related information (e.g., AI / ML data, training data, update information, response, offset of AI / ML, etc.);

[0092] SRB4 is used for specific RRC messages that all use the DCCH logical channel including application layer measurement report information, or RRC messages including AI / ML related information (e.g., AI / ML data, training data, update information, response, offset of AI / ML, etc.). SRB4 can only be configured by the network after the activation of AS security.

[0093] SRBx (e.g., SRB5) is used for RRC messages that include AI / ML related information (e.g., AI / ML data, training data, update information, responses, offsets of AI / ML, etc.), and they all use the DCCH logical channel. SRBx can be configured by the network after AS security activation. However, in cases where SRBx does not need to perform encryption and integrity protection on RRC messages, SRBx can be configured by the network before AS security activation. When SRBx is configured to send / receive RRC messages for AI / ML related information, for SRBx, the network can configure logical channel prioritization parameters (i.e., the allocation of priority and priority bit rate (PBR)) for the LCP process performed in the MAC entity (or layer) to control the prioritization of data transmission. In the present invention, SRBx is a new SRB configured for the AI / ML function, which can be named SRB5.

[0094] DRB (Data Radio Bearer): This is used for AI / ML related information (e.g., AI / ML data, training data, update information, responses, offsets of AI / ML, etc.), and they all use the DCCH logical channel. This DRB can be configured by the network after AS security activation. However, in cases where the DRB does not need to perform encryption and integrity protection on AI / ML data, the DRB can be configured by the network before AS security activation. When the DRB is configured to send / receive AI / ML data, for the DRB, the network can configure logical channel prioritization (LCP) parameters (i.e., the allocation of priority and priority bit rate (PBR)) for the LCP process performed in the MAC entity (or layer) to control the prioritization of data transmission.

[0095] NAS messages: In the downlink, piggybacking of NAS messages is only used for one related (i.e., combined success / failure) procedure: bearer establishment / modification / release. In the uplink, piggybacking of NAS messages is only used to transmit initial NAS messages during connection setup and connection recovery. NAS messages transmitted via SRB2 are also included in the RRC message. However, this RRC message does not include any RRC protocol control information.

[0096] AS (Access Stratum) Security: Once AS security is activated, all RRC messages on SRB1, SRB2, SRB3, and SRB4 (including those containing NAS messages) are integrity protected and encrypted by PDCP (Packet Data Convergence Protocol). NAS by default independently applies integrity protection and encryption to NAS messages. However, even when AS security is activated, some RRC messages (e.g., RRC messages for AI / ML data) may not be integrity protected or encrypted by PDCP, depending on the configuration (through the configuration of RRC messages) or according to predefined rules (e.g., no integrity protection (integrity verification) or no encryption (decryption) for specific RRC messages only) or according to the indication from RRC to PDCP on whether to perform encryption or integrity protection. This allows RRC messages to be without integrity protection and without encryption, RRC messages without integrity protection but encrypted, and RRC messages with integrity protection but not encrypted, in order to reduce the processing burden of the UE and avoid over - security protection in the RAN.

[0097] For operations of shared spectrum channel access, SRB0, SRB1, and SRB3 are assigned the highest - priority channel access priority class (CAPC) (i.e., CAPC = 1), while the CAPC for SRB2 is configurable.

[0098] System Architecture

[0099] Figure 1 Shows a Long - Term Evolution (LTE) system to which the AI / ML techniques described below can be applied. Refer to Figure 1 , the radio access network of the LTE system 10 includes next - generation base stations 16a, 16b, 16c, and 16d, a Mobility Management Entity (MME) 12, and a Serving Gateway (S - GW) 14. The User Equipment 18 (hereinafter referred to as UE or terminal) accesses the external network through eNBs 16a, 16b, 16c, and 16d and the S - GW 14. The next - generation base stations are also referred to as evolved Node Bs or cells, hereinafter referred to as eNBs, Node Bs, cells, or base stations.

[0100] In Figure 1In this, eNBs 16a, 16b, 16c, and 16d correspond to the existing Node Bs of UMTS (Universal Mobile Telecommunications System). The eNBs are connected to the UEs 18 via radio channels and play a more complex role than the existing Node Bs. In the LTE system 10, since all user services related to real-time services such as Voice over IP (VoIP) via the Internet Protocol are served through a shared channel, a device for performing scheduling by collecting status information (such as the buffer status, available transmit power status, and channel status of the UE) is required, and the eNBs 16a, 16b, 16c, and 16d are responsible for this function of the device. Generally, one eNB controls multiple cells. For example, to implement a transmission rate of 100 Mbps, the LTE system 10 uses Orthogonal Frequency Division Multiplexing (OFDM) in a 20 MHz bandwidth as the radio access technology. Additionally, the LTE system 10 adopts an Adaptive Modulation and Coding Scheme (hereinafter referred to as AMC) for determining the modulation scheme and channel coding rate based on the channel status of the UE. The S-GW 14 is a device for providing data bearers and generating or removing data bearers under the control of the MME 12. The MME 12 can be responsible for various control functions and the mobility management function of the UE and is connected to multiple base stations.

[0101] Figure 2 illustrates such as Figure 1 the radio protocol structure in the LTE system as shown. Referring to Figure 2 , the radio protocols of the LTE system respectively include the Packet Data Convergence Protocol (PDCP) 220 and 230, the Radio Link Control (RLC) 222 and 232, and the Media Access Control (MAC) 224 and 234 in the UE 218 and the eNB 216. The Packet Data Convergence Protocol (PDCP) 226 and 236 are used to perform operations such as IP header compression / recovery.

[0102] The main functions of the PDCP are summarized as follows: header compression and decompression only for Robust Header Compression (ROHC); transfer of user data; in-sequence delivery of upper layer protocol data units (PDUs) during the PDCP reconstruction process for RLC acknowledged mode (AM); sequence reordering (for split bearers in DC (RLC AM only supported): PDCP PDU routing for transmission and PDCP PDU reordering for reception); duplicate detection of lower layer service data units (SDUs) during the PDCP reconstruction process for RLC AM; retransmission of PDCP SDUs at handover for RLC AM and retransmission of PDCP PDUs during the PDCP data recovery process for split bearers in DC; encryption and decryption; and timer-based SDU discard in the uplink.

[0103] The radio link control (hereinafter referred to as RLC) 222 or 232 performs automatic repeat request (ARQ) operations by configuring the Packet Data Convergence Protocol (PDCP) protocol data unit (PDU) or the RLC service data unit (SDU) to an appropriate size. The main functions of the RLC include the transfer of upper layer PDUs and RLC re-establishment. Only for AM data transfer, the main functions of the RLC include the ARQ function (error correction via ARQ), re-segmentation of RLC data PDUs, and protocol error detection. Only for unacknowledged mode (UM) and AM data transfer, the main functions of the RLC include concatenation, segmentation, and reassembly of RLC SDUs, re-ordering of RLC data PDUs, duplicate detection, and RLC SDU discard.

[0104] The MAC 224, 234 is connected to a plurality of RLC layer devices configured in a UE 218. The MAC 224, 234 may perform operations of multiplexing RLC PDUs into MAC PDUs and demultiplexing RLC PDUs from MAC PDUs. The main functions of the MAC are summarized as follows: mapping between logical channels and transport channels, multiplexing MAC SDUs belonging to one or different logical channels into transport blocks (TBs) transmitted to the physical layer on a transport channel / demultiplexing MAC SDUs belonging to one or different logical channels from transport blocks (TBs) transmitted from the physical layer on a transport channel, scheduling information reporting, error correction via hybrid automatic repeat request (HARQ), priority handling between logical channels of a UE, priority handling between UEs via dynamic scheduling, multimedia broadcast service (MBMS) identification, transport format selection, and padding.

[0105] The physical layer 226, 236 may perform the following operations: channel encoding and modulation of upper layer data, shaping upper layer data into orthogonal frequency division multiplexing (OFDM) symbols, transmitting OFDM symbols via a radio channel, or demodulating OFDM symbols received via a radio channel, channel decoding OFDM symbols, and transmitting OFDM symbols to the upper layer.

[0106] Figure 3 The structure of a next generation mobile communication system 300 to which the present disclosure can be applied is shown. Refer to Figure 3 , the radio access network of a next generation mobile communication system (hereinafter referred to as NR or 5G) includes a new radio node 326 and a new radio core network 320 (NR CN). The user terminal 318 accesses an external network via the NR gNB 326 and the NR CN 320. The new radio node B is hereinafter referred to as an NR gNB or an NR base station, and the user terminal may be a new radio user equipment and is hereinafter referred to as an NRUE or a UE.

[0107] In Figure 3In this, the NR gNB 326 corresponds to the evolved Node B (eNB) of the existing LTE system (for example, as shown in Figure 1 ). The NR gNB 326 is connected to the NR UE 312 via a radio channel and can provide high-quality services compared to the existing Node B. In the next-generation mobile communication system, since all types of user traffic are served through a shared channel, a device for performing scheduling by collecting status information (such as the buffer status of the UE, the available transmit power status, and the channel status) is required. In addition, the NR gNB 326 is responsible for this function of the device. Generally, one NR gNB 326 typically controls multiple cells. To implement ultra-high-speed data transmission compared to the existing LTE, the NR gNB 326 can have the existing maximum bandwidth or larger, and can additionally adopt beamforming technology using orthogonal frequency division multiplexing (hereinafter referred to as OFDM) as the radio access technology. In addition, the NR gNB 326 adopts an adaptive modulation and coding (AMC) scheme that determines the modulation scheme and the channel coding rate based on the channel status of the UE.

[0108] The NR CN 320 performs functions such as mobility support, bearer configuration, QoS configuration, etc. The NR CN 320 is a device responsible for various control functions and the mobility management function of the UE 318, and is connected to multiple base stations. In addition, the next-generation mobile communication system can also operate in combination with the existing LTE system 330, and the NR CN 320 can be connected to the MME 312 via a network interface. The MME 312 is connected to the eNB 316, that is, connected to the existing base station.

[0109] Figure 4 shows a radio protocol structure of the next-generation mobile communication system such as Figure 3 shown. Referring to Figure 4 , the radio protocols of the next-generation mobile communication system respectively include the NR SDAP 420 and 430, NR PDCP 422 and 432, NR RLC 424 and 434, and NR MAC 426 and 436 in the UE 318 and the NR base station 326.

[0110] The main functions of NR SDAP 420 and 430 may include one of the following functions: transmission of user plane data; mapping between QoS flows and data radio bearers (DRBs) for both the downlink (DL) and the uplink (UL); marking the QoS flow ID in both DL and UL packets; and mapping reactive QoS flows to DRBs for UL SDAP PDUs. For the SDAP layer device, for each PDCP layer device, each bearer, and each logical channel, the UE 318 can be configured via an RRC message regarding whether to use the header of the SDAP layer device (or new layer device) or the function of the SDAP layer device (or new layer device). When the SDAP header is configured, a NAS reactive QoS reactive configuration 1-bit indicator (NAS reactive QoS) and an AS QoS reactive configuration 1-bit indicator (AS reactive QoS) using the SDAP header are used to indicate that the UE implements an update or reconfiguration of the mapping information related to the QoS flows and data bearers of the uplink and the downlink. The SDAP header may include QoS flow ID information indicating QoS. The QoS information can be used as data processing priority information, scheduling information, etc., to support smooth services.

[0111] The main functions of NR PDCP 422 and 432 may include some of the following functions: header compression and decompression (only ROHC); transmission of user data; in-sequence delivery of upper layer PDUs; out-of-sequence delivery of upper layer PDUs; reordering of PDCP PDUs for reception; duplicate detection of lower layer SDUs; retransmission of PDCP SDUs; encryption and decryption; and timer-based SDU discard in the uplink. The reordering function of the NR PDCP device refers to the function of reordering the PDCP PDUs received from the lower layer in sequence based on the PDCP sequence number (SN). The reordering function may include the function of transmitting data to the upper layer in the reordered sequence, the function of transmitting data to the upper layer directly without considering the sequence, the function of reordering the sequence and recording the lost PDCP PDUs, the function of providing a status report on the lost PDCP PDUs to the transmitting side, and the function of requesting retransmission of the lost PDCP PDUs.

[0112] The main functions of NR RLC 424 and 434 may include some of the following functions: transmission of upper layer PDUs; in-sequence delivery of upper layer PDUs; out-of-sequence delivery of upper layer PDUs; error correction via ARQ; concatenation, segmentation, and reassembly of RLC SDUs; re-segmentation of RLC data PDUs; reordering of RLC data PDUs; duplicate detection; protocol error detection; RLC SDU discard; and RLC reconstruction.

[0113] The in-sequence delivery function of the NR RLC device refers to the function of transmitting the RLC SDUs received from the lower layer to the upper layer in the received order. If an RLC SDU is initially segmented into multiple RLC SDUs and received, the in-sequence delivery function may include the function of reassembling and transmitting the multiple RLC SDUs. The in-sequence delivery function may include the functions of reordering the received RLC PDUs based on the RLC SN or PDCP SN, reordering the sequence and recording the lost PDCP PDUs, providing a status report on the lost RLC PDUs to the transmitting side, or requesting retransmission of the lost RLC PDUs. Alternatively, the in-sequence delivery function of the NR RLC device may include the function of transmitting only the RLC SDUs before the lost RLC SDU to the upper layer in sequence in the case of loss of an RLC SDU, or transmitting all the RLC SDUs received before the start of the timer to the upper layer in sequence even if there are lost RLC SDUs in the case of timer expiration, or transmitting all the RLC SDUs received so far to the upper layer in sequence even if there are lost RLC SDUs in the case of expiration of a predetermined timer. Additionally, the RLC PDUs may be processed in the order in which the RLC PDUs are received (in the order of arrival regardless of the sequence number or serial number) and may be transmitted to the PDCP device in out-of-sequence delivery. The in-sequence delivery function may include the functions of receiving the segments stored in the buffer or the segments to be received later, reconfiguring the segments in a complete RLC PDU, processing the RLC PDU, and transmitting the RLC PDU to the PDCP device. The NR RLC layer may not include the concatenation function, and the concatenation function may be performed by the NR MAC layer or may be replaced by the multiplexing function of the NR MAC layer.

[0114] The out-of-sequence delivery function of the NR RLC device refers to the function of directly transmitting the RLC SDUs received from the lower layer to the upper layer regardless of the order. If an RLC SDU is initially segmented into multiple RLC SDUs and received, the out-of-sequence delivery function may include the functions of reassembling and transmitting the multiple RLC SDUs, as well as storing the RLC SN or PDCP SN of the received RLC PDUs, reordering the sequence, and recording the lost RLC PDUs.

[0115] The NR MACs 426 and 436 can be connected to multiple NR RLC layer devices configured in a UE 318. The main functions of the NR MAC can include some of the following functions: mapping between logical channels and transport channels; multiplexing / demultiplexing of MAC SDUs; scheduling information reporting; error correction via HARQ; prioritization between logical channels of a UE; prioritization between UEs by means of dynamic scheduling; MBMS service identification; transport format selection; and padding.

[0116] The NR PHY layers 428 and 438 can perform the following operations: channel encoding and modulation of upper layer data, shaping upper layer data into OFDM symbols, transmitting OFDM symbols via a radio channel, or demodulating and channel decoding OFDM symbols received via a radio channel, and delivering OFDM symbols to the upper layer.

[0117] Principles of security protection

[0118] As described above, in these two example networks, there is security protection from encryption or integrity protection. Encryption means not only the encryption operation but also the decryption operation, because if data is encrypted at the transmitter, the data should be decrypted at the receiver. Similarly, integrity protection means the integrity verification operation as well as the integrity protection operation, because if data is integrity protected at the transmitter, the data should be integrity verified at the receiver.

[0119] When implementing this technology, the following high-level principles can be applied. AS security includes integrity protection and encryption of RRC signaling (SRB) and user data (DRB). The RRC processes the configuration of AS security parameters as part of the AS configuration: integrity protection algorithms and encryption algorithms. If integrity protection and / or encryption are enabled for a DRB and two parameters (i.e., keySetChangeIndicator and nextHopChainingCount), the UE uses these two parameters to determine the AS security key during synchronous reconfiguration (key change), connection reconstruction, and / or connection recovery.

[0120] The integrity protection algorithm is common for SRB1, SRB2, SRB3 (if configured), SRB4 (if configured), SRBx (if configured), and DRB that are configured with integrity protection having the same keyToUse value. The encryption algorithm is common for SRB1, SRB2, SRB3 (if configured), SRB4 (if configured), SRBx (if configured), and DRB that are configured with the same keyToUse value. Neither integrity protection nor encryption applies to SRB0. It should be noted that all DRBs associated with the same PDU session have the same encryption enable / disable setting and the same integrity protection enable / disable setting.

[0121] RRC integrity protection and encryption are always activated together, i.e., in one message / process. RRC integrity protection and encryption for SRBs are never deactivated. However, it is possible to switch to the "NULL" encryption algorithm (nea0). For SRBx (if configured), RRC integrity protection and encryption can be activated and deactivated based on the configuration or indication of an RRC message (or MAC CE (control element) or PDCP control PDU (protocol data unit)) in order to reduce the processing burden on the UE. For SRBx (if configured), it is also possible to switch to the "NULL" encryption algorithm (nea0), and the "NULL" integrity protection algorithm (nia0) can be used.

[0122] The "NULL" integrity protection algorithm (nia0) is only used for SRBs and UEs in the restricted service mode, and when used for SRBs, integrity protection is disabled for DRBs. In the case of using the "NULL" integrity protection algorithm, the "NULL" encryption algorithm is also used. It should be noted that the lower layer discards RRC messages that fail the integrity protection check and indicates to RRC that the integrity protection verification check has failed.

[0123] The AS applies four different security keys: one for the integrity protection of RRC signaling (K RRCint ), one for the encryption of RRC signaling (K RRCenc ), one for the integrity protection of user data (K UPint ), and one for the encryption of user data (K UPenc ). All four AS keys are derived from the K gNB key. The K gNB key is based on the K AMF key processed by the upper layer. The integrity protection and encryption algorithms can only be changed through synchronous reconfiguration. The AS keys (K gNB , K RRCint , K RRCenc , K UPint and KUPenc ) Changes during synchronous reconfiguration (if masterKeyUpdate is included) and during connection re-establishment and connection recovery.

[0124] For each radio bearer, independent counters (COUNT used in the PDCP layer) are maintained for each direction. For each radio bearer, the COUNT is used as input for encryption and integrity protection. For a given security key, the same COUNT value is not allowed to be used more than once. The network is responsible for avoiding the reuse of COUNT with the same RB identity and the same key, for example, due to the transmission of a large amount of data, the release and establishment of new RBs, and the change of multiple endpoints of RLC-UM bearers and the change of multiple endpoints of RLC-AM bearers, where the PDCP reconstruction (COUNT reset) caused by only SN full configuration, while the key stream input (i.e., bearer ID, security key) at the MN has not been updated. To avoid such reuse, the network can, for example, use different RB identities for RB establishment, change the AS security key, or transition from RRC_CONNECTED to RRC_IDLE / RRC_INACTIVE and then to RRC_CONNECTED.

[0125] To limit signaling overhead, each message / packet includes a short sequence number (PDCP SN (sequence number)). Additionally, an overflow counter mechanism is used: hyperframe number (HFN used in the PDCP layer). The HFN needs to be synchronized between the UE and the network.

[0126] For each SRB, the value of the 5-bit BEARER parameter provided by RRC to the lower layer for deriving the input for encryption and integrity protection is the value of the corresponding srb-Identity with MBS filled with zeros.

[0127] For a UE equipped with sk-counter, keyToUse indicates whether the UE uses the master key (K gNB ) or the secondary key (S-K eNB or S-K gNB ) for a specific DRB. The secondary key is derived from the master key and the sk-Counter. Whenever the secondary key needs to be refreshed, for example, when the MN changes with K gNB or when changing to avoid COUNT reuse, a security key update is used. When the UE is in NR-DC, even if the secondary key (S-K gNB ) is not used to configure the DRB, the network can provide the sk-Counter to the UE configured with SCG to allow the configuration of SRB3. When using an SN-terminated MCG bearer, even if SCG is not configured, the network can provide the sk-Counter to the UE.

[0128] General AI / ML framework

[0129] Figure 5 is a schematic block diagram of a functional framework 500 for implementing artificial intelligence or machine learning in a radio access network (such as Figure 1 or Figure 3 those shown). Such a functional framework may be referred to as AI-enabled radio access network intelligence. Figure 5 It includes a data collection module 502 that provides training data to a model training module 504 and inference data to a model inference module 506. As shown by the information flow, after the model has been trained in the model training module 504, an update of the model is deployed to the model inference module 506. Optionally, as indicated by the dashed line, the model inference module 506 may provide model performance feedback to the model training module 504, and this model performance feedback may be used to further update the model.

[0130] Figure 5 An actor module 508 is also shown. The actor module 508 receives the output from the model inference module 506 and is used to provide feedback to the data collection module 502. As described in the above definition, the output from the model inference module 506 is the inference output of the AI / ML model. The feedback may include information required to derive training data, inference data, or monitor the performance of the AI / ML model.

[0131] The detailed AI / ML algorithms and models for use cases are implementation-specific. For example, considering the problems identified in the background section, ML techniques can be utilized to optimize energy-saving decisions by leveraging the data collected in the RAN network. The ML algorithm can predict the energy efficiency and load status of the next cycle, which can be used to make better decisions regarding cell activation / deactivation for energy saving. Based on the predicted load, the system can dynamically configure energy-saving policies (e.g., shutdown time and granularity, offloading actions) to maintain the balance between system performance and energy efficiency and reduce energy consumption. Additionally or alternatively, an AI / ML model-based solution can be introduced to improve load balancing performance. Based on various measurements and feedback, historical data, etc., collected from UEs and network nodes. The AI / ML model-based solution and predicted load can improve load balancing performance in order to provide a higher quality user experience and increase system capacity.

[0132] For further use, the mobility aspects of a self-organizing network (SON) that can be enhanced by using AI / ML include reducing the probability of unexpected events, UE location / mobility / performance prediction, and traffic steering. Predicting the location of the UE is a key part of mobility optimization since many RRM actions related to mobility (e.g., selecting a handover target cell) can benefit from the predicted UE location / trajectory. UE mobility prediction is also a key factor in optimizing early data forwarding, especially for CHO. When the UE is served by certain cells, UE performance prediction is a key factor in determining which is the best mobility target for maximizing efficiency and performance. RAN intelligence can observe multiple handover events with associated parameters, use this information to train its ML model and attempt to identify the set of parameters that lead to successful handovers and the set of parameters that lead to unexpected events.

[0133] Regarding traffic steering, efficient resource handling can be achieved by adjusting the handover trigger point and selecting the best combination of Pcell / Pcell / Scell serving the user. Existing traffic steering can also be improved by providing the RAN nodes with mobility-related information. For example, before initiating a handover, the source gNB can use the feedback on UE performance collected for past successful handovers and received from neighboring gNBs. In two reported examples, the source RAN node of a mobility event or the RAN node acting as the master node (eNB for EN-DC, gNB for NR-DC) can use the feedback received from another RAN node as input to the AI / ML functions that support traffic-related decisions (e.g., selecting a target cell in the case of mobility, selecting a PSCell / Scell in another case), enabling future decisions to be optimized.

[0134] The location of the AI / ML functions in the current RAN architecture depends on the deployment and the specific use case. For example, one solution that can be considered to support AI / ML-based network energy savings includes having the AI / ML model training located in the operations, administration, and management (OAM) system and the AI / ML model inference located in the gNB. In this example, it is also allowed for the gNB to continue with model training based on the AI / ML model trained in the OAM. Alternatively, both the AI / ML model training and the AI / ML model inference are located in the gNB.

[0135] Detailed AI / ML algorithms and models (both training and inference) can be configured to the UE, gNB, or network entity along with other relevant parameters and configuration information via RRC (Radio Resource Control) messages or newly defined network messages (e.g., X2 messages, inter-node messages for the interfaces between network entities). The configuration using messages allows for various arrangements. For example, AI / ML model training can be located in the UE, and AI / ML model inference can be located in the gNB or OAM. Alternatively, AI / ML model training can be located in the gNB or OAM, and AI / ML model inference can be located in the UE. Alternatively, both AI / ML model training and AI / ML model inference are located in the UE. Training at the UE can implement federated learning or split learning as described above.

[0136] The architecture can include a split of the Central Unit (CU) and Data Unit (DU) (in other words, as one of the gNB implementation ways, a split CU / DU architecture can exist). In this example, one solution is that AI / ML model training is located in the OAM and AI / ML model inference is located in the gNB-CU. Another solution is that both AI / ML model training and model inference are located in the gNB-CU. The configuration using messages (e.g., for federated or split learning) allows for further solutions. For example, one solution is that AI / ML model training is located in the UE and AI / ML model inference is located in the gNB-CU or OAM. Another solution is that AI / ML model training is located in the gNB-CU or OAM and AI / ML model inference is located in the UE. Another solution is that both AI / ML model training and AI / ML model inference are located in the UE.

[0137] In all the various arrangements, the general principle is that if needed, the model training and model inference functions should be able to request specific information for training or executing the AI / ML algorithm and avoid receiving unnecessary information from the UE, gNB, or network entity by using signaling (i.e., sending request messages). The nature of such information depends on the use case and the AI / ML algorithm. The model inference function should only signal the output of the model to nodes that explicitly request the output of the model (e.g., via subscription or via configuration) or nodes that take actions based on the output from the model inference, e.g., by sending messages (e.g., RRC messages, inter-node messages, or newly defined messages). The AI / ML model used in the model inference function must be initially trained, validated, and tested by the model training function before deployment to ensure performance and avoid excessive signaling and save radio resources. In some cases, training can also be performed after deployment. During the AI / ML operation, user data privacy and anonymization can be respected.

[0138] As described in detail below, the proposed mechanism can operate in NG-RAN SA (Next Generation RAN Standalone operation mode) and thus in the Figure 3 network arrangement. Those skilled in the art will readily understand that the proposed mechanism can be extended to other suitable networks, such as EN-DC (Evolved Universal Terrestrial Radio Access (E-UTRA) - New Radio Dual Connectivity with E-UTRA connected to the EPC (Evolved Packet Core Network)) and MR-DC (Multi-Radio Access Technology (RAT) Dual Connectivity).

[0139] User Equipment States and Transitions

[0140] As mentioned above, the AI / ML configuration information can be sent in the RRC message or other messages. Before considering in detail an example of the implementation, first consider the RRC procedures proposed for handling the AI / ML functions. Figure 6 An overview of the UE RRC state machine and state transitions in NR is shown. The UE has only one RRC state in NR at a time. As Figure 6 shown, when the RRC connection has been established, the UE is in the RRC_CONNECTED state or the RRC_INACTIVE state. If this is not the case, i.e., the RRC connection has not been established, the UE is in the RRC_IDLE state.

[0141] The RRC_IDLE state can be characterized by various features. For example, UE-specific DRX can be configured by the upper layer, and at the lower layer, the UE can be configured with discontinuous reception (DRX) for point-to-multipoint (PTM) transmissions for multicast and broadcast services (MBS) broadcasts. There can also be UE-controlled mobility based on network configuration. The UE can monitor short messages transmitted via the Downlink Control Information (DCI) together with the Paging - Radio Network Temporary Identifier (P-RNTI), and can monitor the paging channel for core network (CN) paging using the 5G-S-TMSI, unless the UE acts as a level 2 U2N remote UE. When the UE is configured by the upper layer for MBS multicast reception, the UE can monitor the paging channel for CN paging using the TMGI. The UE can perform neighbor cell measurements and cell (re)selection; obtain the System Information (SI) and can send an SI request (if configured); and for the UE with a recorded measurement configuration, it can perform the recording of available measurements and the recording of location and time. The UE can also perform idle / inactive measurements for the UE with an idle / inactive measurement configuration, and perform AI / ML functions (e.g., collect AI / ML data or measure AI / ML data or report AI / ML data) for the appropriately configured UE. When the UE is configured by the upper layer for MBS broadcast reception, the UE can obtain the MCCH change notification and MBS broadcast control information and data.

[0142] Similarly, the RRC_INACTIVE state can be characterized by various features. For example, UE-specific DRX can be configured by the upper layer or by the RRC layer, and at the lower layer, the UE can be configured with DRX for PTM transmission for MBS broadcast. There can be UE-controlled mobility based on network configuration, and the UE can store the UE inactive AS context. The RAN-based notification area can be configured by the RRC layer, and unicast data and / or signaling can be configured for small data transmission (SDT) to / from the UE via a radio bearer. The UE can monitor short messages sent together with the P-RNTI via DCI. During the SDT procedure, the UE can monitor the control channel associated with the shared data channel to determine whether data is scheduled for it, and when the SDT procedure is not in progress, the UE can monitor the paging channel for CN paging using the 5G-S-TMSI and the paging channel for RAN paging monitored using the full RNTI, unless the UE acts as an L2 U2N remote UE. If the UE is configured by the upper layer for MBS multicast reception, the UE can monitor the paging channel for paging using the TMGI when the SDT procedure is not in progress. The UE can perform neighbor cell measurements and cell (re)selection; and perform RAN-based reporting area updates periodically and when moving out of the configured RAN-based notification area. The UE can obtain system information when the SDT procedure is not in progress and can send an SI request (if configured). When the SDT procedure is not in progress, the UE can record available measurements together with location and time for the UE with a recorded measurement configuration; perform idle / inactive measurements for the idle / inactive measurement configuration UE, and perform AI / ML functions (e.g., collect AI / ML data or measure AI / ML data or report AI / ML data) for the configured UE. If the UE is configured by the upper layer for MBS broadcast reception, the UE obtains MCCH change notifications and MBS broadcast control information and data. The UE can also transmit SRS for positioning.

[0143] Similarly, the RRC_CONNECTED state can be characterized by various features. For example, the UE stores the AS context. There is unicast data transfer to / from the UE and MBS multicast data transfer to the UE. At the lower layer, the UE can be configured with UE-specific DRX or DRX for PTM transmission for MBS broadcast and / or DRX for MBS multicast. For a UE that supports carrier aggregation (CA), one or more SCell aggregated with the SPCell can be used to increase the bandwidth. For a UE that supports DC, one SCG aggregated with the MCG can be used to increase the bandwidth. There can be network-controlled mobility within NR, to / from E-UTRA, and to UTRA-FDD. There can be network-controlled mobility (path switching) between the serving cell and the L2 U2N relay UE, and vice versa. If configured, the UE can monitor short messages sent with DCI together with the P-RNTI. The UE can monitor the control channel associated with the shared data channel to determine whether data is scheduled for it. The UE can provide channel quality and feedback information. The UE can perform neighbor cell measurements and measurement reports and AI / ML functions (e.g., collect AI / ML data or measure AI / ML data or report AI / ML data) for the configured UE. The UE can obtain system information. The UE can perform immediate minimized drive test (MDT) measurements and available location reports. If the US is configured by the upper layer for MBS broadcast reception, the UE obtains MCCH change notifications and MBS broadcast control information and data.

[0144] Principles of RRC Control for AI / ML

[0145] The RRC protocol includes the following main functions: broadcast of system information, RRC connection control, and recovery from radio link failure. The broadcast of system information includes different types of information, including some or all of the NAS public information; information applicable to UEs in RRC_IDLE and RRC_INACTIVE (e.g., cell (re)selection parameters, neighbor cell information) and (also) information applicable to UEs in RRC_CONNECTED (e.g., common channel configuration information); earthquake and tsunami warning system (ETWS) notifications, commercial mobile alert service (CMAS) notifications; positioning assistance data, and an indication of whether the cell supports AI / ML functions. The system information can broadcast this indication. When the UE obtains the system information, the UE can perform cell (re)selection based on this, and if the system information broadcasts an indication of support for AI / ML functions, send UE capability information including the UE's support for AI / ML functions to the cell (or gNB or network).

[0146] RRC connection control may include establishment, modification, suspension, resume, and / or release of RRC connections, including, for example, assignment / modification of UE identifiers (C-RNTI, full I-RNTI, etc.). RRC connection control may include establishment, modification, suspension, resume, and / or release of SRBs (except SRB0). RRC connection control may include establishment, modification, suspension, resume, and / or release of RBs (DRB / MRB) carrying user data;

[0147] RRC connection control may include access prohibition; the access prohibition mechanism may allow UEs for the purpose of AI / ML functions to access the cell based on a newly defined indication in the system information. The access prohibition mechanism may not allow UEs for the purpose of AI / ML functions to access the cell based on a newly defined indication in the system information. RRC connection control may include initial AS security activation, i.e., initial configuration of AS integrity protection (SRB, DRB) and AS encryption (SRB, DRB). RRC connection mobility may include, for example, intra-frequency and inter-frequency handovers, path switching from the PCell to the target L2 U2N relay UE or from the L2 U2N relay UE to the target PCell, associated AS security processing (i.e., key / algorithm change), and specification of RRC context information transmitted between network nodes.

[0148] RRC connection control may include radio configuration control, including, for example, assignment and / or modification of ARQ configuration, HARQ configuration, and DRX configuration. In the case of DC, cell management may include, for example, addition, modification, and / or release of SCG cells or change of the PSCell. In the case of CA, cell management may include, for example, addition, modification, and / or release of SCell. RRC connection control may include QoS control, including assignment or modification of semi-persistent scheduling (SPS) configuration and configured grant configuration for downlink (DL) and uplink (UL) respectively, and assignment or modification of parameters for UL rate control in the UE, i.e., assignment of priorities and priority bit rates (PBR) for each RB of the UE and logical channels of the IAB-MT.

[0149] RRC connection control may include AI / ML function control, including assignment or modification of semi-persistent scheduling (SPS) configuration and configured grant configuration for DL and UL respectively, and assignment or modification of parameters for UL rate control in the UE, i.e., assignment of priorities and priority bit rates (PBR) for each RB of the UE and integrated access and backhaul mobile terminal (IAB-MT) logical channels.

[0150] Recovery from radio link failure may include inter-RAT mobility, including, for example, AS security activation and transmission of RRC context information. Recovery from radio link failure may include measurement configuration and reporting. For example, the configuration and reporting may include establishing, modifying, or releasing measurement configuration (e.g., intra-frequency, inter-frequency, and inter-RAT measurements; measurements for AI / ML functions; setup and release of measurement gaps and measurement reports). Recovery from radio link failure may include AI / ML functions and AI / ML configuration and reporting of measurements. This may include establishing, modifying, or releasing AI / ML configuration, including measurement configuration (e.g., intra-frequency, inter-frequency, and inter-RAT measurements, time information (measurement duration or measurement timing), or sequence number or cell identity (or physical cell identity) for reordering), and including measurement configuration (e.g., for beam or single sideband (SSB) or channel state information (CSI) or cell reference signal (CRS)). The AI / ML configuration and reporting of AI / ML functions may include setting and releasing AI / ML configuration including measurement configuration; and the AI / ML reporting includes measurement reports (e.g., for frequency or beam or SSB or CSI or CRS or cell identity (or physical cell identity)).

[0151] Recovery from radio link failure may include configuring the BAP entity and the BH RLC channel to support the IAB node. Recovery from radio link failure may include other functions, such as general protocol error handling, transmission of dedicated NAS information, and transmission of UE radio access capability information. Recovery from radio link failure may include support for self-configuration and self-optimization; support for measurement recording and reporting for network performance optimization; support for transmission of application layer measurement configuration and reporting; and support for AI / ML configuration and reporting.

[0152] The establishment of the RRC connection involves the establishment of SRB1. The network completes the establishment of the RRC connection before completing the establishment of the NG connection (i.e., before receiving the UE context information from the 5GC). Therefore, during the initial phase of the RRC connection, AS security is not activated. During this initial phase of the RRC connection, the network may configure the UE to perform measurement reporting or collect AI / ML data, but the UE only sends the corresponding measurement reports or AI / ML data after successful AS security activation. However, when AS security has been activated, the UE only accepts synchronous reconfiguration messages.

[0153] AI / ML data indicates AI / ML (model) training or inference or training data or prediction information or decision parameters or a set of inputs for AI / ML or an AI / ML model. For example, AI / ML data may indicate the measurement reports of the UE (e.g., for beam or SSB (synchronization signal block) or CSI or CRS (cell-specific reference signal)) and CSI (channel state information) reports or positioning information.

[0154] After receiving the UE context from the 5GC, the RAN uses the initial AS security activation procedure to activate AS security (both encryption and integrity protection). The RRC messages (commands and successful responses) used to activate AS security are integrity protected, while encryption starts after the process is completed. That is, the response to the message used to activate AS security is not encrypted, while subsequent RRC messages (e.g., for establishing SRB2, DRB, and multicast MRB or new RBs for supporting AI / ML functions (e.g., SRBx or SRB4 or DRB), or for transmitting / receiving AI / ML data (or AI / ML configuration)), e.g., RRCSetup or RRCResume or RRCReconfiguration messages are integrity protected and encrypted. In another embodiment, based on the security configuration, the RRC messages transmitted via the RB for transmitting / receiving AI / ML data (or AI / ML configuration) may not be integrity protected or encrypted, or neither. After initiating the initial AS security activation procedure or establishing SRB1, the network may initiate the establishment of SRB2 and DRB and / or multicast MRB or new RBs for supporting AI / ML functions (e.g., SRBx or SRB4 or DRB), i.e., the network may do so before receiving the confirmation of the initial AS security start from the UE. In any case, the network will apply encryption and integrity protection to the RRC reconfiguration messages for establishing SRB2, DRB, and / or multicast MRB or new RBs for supporting AI / ML functions (e.g., SRBx or SRB4 or DRB).

[0155] The network initiates the security mode command procedure for a UE in RRC_CONNECTED. In addition, when only SRB1 is established, i.e., before establishing SRB2, multicast MRB, and / or DRB or new RBs for supporting AI / ML functions (e.g., SRBx or SRB4 or DRB), the network applies this procedure. The network may initiate an RRC reconfiguration procedure for a UE in RRC_CONNECTED to establish RBs (other than SRB1 established during RRC connection establishment), e.g., SRB2, DRB, and / or multicast MRB or new RBs for supporting AI / ML functions (e.g., SRBx or SRB4 or DRB), which are only executed when AS security has been activated.

[0156] If the initial AS security activation and / or radio bearer establishment fails, the network shall release the RRC connection. Configurations with SRB2 without a DRB or multicast MRB or a new RB for supporting AI / ML functions (e.g., SRBx or SRB4 or DRB) or with a DRB or multicast MRB without SRB2 are not supported (i.e., SRB2 and at least one DRB or multicast MRB or a new RB for supporting AI / ML functions (e.g., SRBx or SRB4 or DRB) must be configured in the same RRC reconfiguration message, and it is not allowed to release all DRBs and multicast MRBs and new RBs for supporting AI / ML functions (e.g., SRBx or SRB4 or DRB) without releasing the RRC connection). For IAB-MT, configurations with SRB2 without any DRB / MRB / or new RB for supporting AI / ML functions (e.g., SRBx or SRB4 or DRB) are supported.

[0157] To establish a new RB for supporting AI / ML functions (e.g., SRBx or SRB4 or DRB), RRC messages transmitted via SRB1 (e.g., RRCReconfiguration message) include the AI / ML configuration and the configuration of the new SRB.

[0158] The AI / ML configuration includes the type of the AI / ML model (e.g., an identifier of the model or model function); model inference; model training; the algorithm or how to construct the RRC message for response (or reporting) that includes AI / ML data (e.g., what information should be included) or how to send the RRC message from the transmitter (UE or gNB) to the receiver (gNB or UE) (e.g., periodicity, timer-related information, etc.). The AI / ML configuration may include the measurement configuration of the UE (e.g., for beam or SSB (synchronization signal block) or CSI or CRS (cell-specific reference signal)) and the CSI measurement configuration (e.g., aperiodic CSI, periodic CSI, etc.) or the positioning configuration.

[0159] When the AI / ML data or AI / ML configuration indicates an AI / ML model, this means the transmission (or delivery) of the AI / ML model via RRC messages through SRB1 or a new RB for supporting AI / ML functions (e.g., SRBx or SRB4 or DRB).

[0160] AI / ML model transfer (or AI / ML model delivery) may mean delivering an AI / ML model over the air interface, whether it is the parameters of the model structure known at the receiving end, a new model with parameters, or the identity of the model or function. This function may indicate UE measurements (e.g., for beam or SSB (Synchronization Signal Block) or CSI or CRS (Cell-Specific Reference Signal)) and CSI measurements (e.g., aperiodic CSI, periodic CSI, etc.) or positioning. The delivery may include the full model or a partial model. This may be a general term for delivering an AI / ML model from one entity to another entity in any way. An entity may mean a network node / function (e.g., gNB, LMF (Location Management Function), etc.), UE, proprietary server, etc.

[0161] When a new RB (e.g., SRBx or SRB4 or DRB) is established to support AI / ML functions, the UE (or gNB) may send AI / ML data to the gNB (or UE) via the RB, and vice versa. If the RB is an SRB (e.g., SRBx or SRB4 or SRB5), RRC messages including AI / ML data may be transmitted and received in the UE or gNB, and RRC messages may be newly defined for AI / ML functions.

[0162] When transmitting AI / ML data (or AI / ML configuration) via SRB1 or a new RB, if RRC message segmentation is enabled based on the field rrc-SegAllowed received in the received RRC message, and the encoded RRC message (or AI / ML data) is larger than the maximum supported size of the PDCP SDU for uplink transmission or downlink transmission (e.g., 9 kilobytes), then the RRC message (or AI / ML data) may be segmented. When the RRC message is segmented, each segment includes its own sequence number and an indicator (to indicate whether it is the last segment).

[0163] The AI / ML configuration may be configured, and the AI / ML data may be transmitted and received as part of SON (Self-Organizing Network) processes, MDT (Minimization of Drive Tests) processes, UE assistance information, early idle / inactive measurements, RRM (Radio Resource Management) measurement reports, CSI (Channel State Information) reporting frameworks, LPP (LTE Positioning Protocol) provided location information.

[0164] Return Figure 6, as shown in the figure, when the RRC connection is released, the UE transitions from the RRC_CONNECTED state to the RRC_IDLE state. The release of the RRC connection is typically initiated by the network. This process can be used to redirect the UE to an NR frequency or an E-UTRA carrier frequency. The suspension of the RRC connection is initiated by the network. When the RRC connection is suspended, the UE stores the UE inactive AS context and any configurations received from the network and transitions to the RRC_INACTIVE state. The RRC message used to suspend the RRC connection is integrity protected and encrypted.

[0165] When the UE needs to transition from the RRC_INACTIVE state to the RRC_CONNECTED state, either the RAN notification area (RNA) update is performed by the RRC layer, or the recovery of the suspended RRC connection is initiated by RAN paging from the NG-RAN or, for SDT, by the upper layer. When the RRC connection is recovered, the network configures the UE according to the RRC connection recovery procedure based on the stored UE inactive AS context and any RRC configurations received from the network. The RRC connection recovery procedure reactivates the AS security and re-establishes the SRB and DRB and / or the multicast MRB or the new radio bearer RB (e.g., SRBx or SRB4 or DRB) for supporting AI / ML functions (if configured). When the new RB for supporting AI / ML functions is a DRB, the DRB can be an AM (acknowledged mode) DRB, which is configured with an AM RLC entity and uses the RLC AM mode to support lossless delivery and high reliability, where the AM mode supports the ARQ (automatic repeat request) mechanism.

[0166] After initiating the recovery procedure for SDT, the AS security (encryption and integrity protection) is reactivated for SRB2 (if configured for SDT) and SRB1. Additionally, for all DRBs configured for SDT configuration, the AS security is also reactivated (if security is configured). Furthermore, when the UE remains in the RRC_INACTIVE state, the PDCP entity of SRB1 and the PDCP entities of the radio bearers configured for SDT are re-established and recovered. When the UE is in the RRC_INACTIVE state and the SDT process is ongoing, the transmission and reception of data and / or signaling messages can occur on the radio bearers configured for SDT.

[0167] In response to a request to resume an RRC connection or in response to a resume procedure initiated for SDT, the network may resume a suspended RRC connection and move the UE to RRC_CONNECTED, or reject the resume request and move the UE to RRC_INACTIVE (using a wait timer), or directly re-suspend the RRC connection and move the UE to RRC_INACTIVE, or directly release the RRC connection and move the UE to RRC_IDLE, or instruct the UE to initiate NAS-level resume (in which case the network sends an RRC setup message).

[0168] System Information Acquisition

[0169] As described above, the UE applies a system information (SI) acquisition procedure to obtain AS, NAS, and positioning assistance data information. This procedure is applicable to UEs in RRC_IDLE, RRC_INACTIVE, and RRC_CONNECTED states. Figure 7 Fig. shows one way in which user equipment 718 may acquire system information from network 700.

[0170] In the first step S720, a UE 718 in the RRC_IDLE or RRC_INACTIVE mode receives information from the Management Information Base (MIB) from the network 700. This ensures that the UE 718 has a valid version of the MIB. In the second step S722, the UE 718 receives system information block types including at least SIB1 to SIB4 from the network 700. If the UE supports E-UTRA, it also receives SIB5. If the UE is configured for idle / inactive measurements, it also receives SIB11. If the UE is capable of NR sidelink communication / discovery and is configured by the upper layer to receive or transmit NR sidelink communication / discovery, it also receives SIB12. If the UE is capable of V2X sidelink communication and is configured by the upper layer to receive or transmit V2X sidelink communication, it receives SIB13 and SIB14. If the UE is configured by the upper layer to report disaster roaming related information, it receives SIB15. If the UE is capable of slice-based cell reselection and the UE receives NSAG information for cell reselection from the upper layer, it receives SIB16. If the UE is accessing NR via NTN access, it receives SIB19. Finally, if the US is capable of implementing AI / ML functions (such as AI / ML model training and model inference), it receives a new system information block type SIBxx (e.g., SIB22). SIBxx may broadcast whether the cell supports AI / ML model training or model inference or AI / ML functions, or broadcast information about AI / ML model training or model inference. Regardless of which RRC state the UE is in, a UE that is receiving or interested in receiving AI / ML functions via an SRB or DRB and is capable of supporting AI / ML functions should ensure that it has a valid version of SIBxx.

[0171] Once the management information and system type block information are shared, as shown in step S724, the UE 718 can send a message (SystemInformationRequest) requesting system information. In the next step S726, the network 700 responds with one or more messages (SystemInformationMessages) conveying the requested system information.

[0172] The detailed process of an example state transition

[0173] Figure 8 is a flowchart showing the process of a UE switching from the RRC idle mode to the RRC connected mode. This process can be in a next-generation mobile communication system (such as Figure 1 or Figure 3Execute in the RRC connection mode in the example shown. This process includes a method for configuring protocol layer devices or functions of the UE 818. As shown in the figure, there is communication between the UE 818 and the first node gNB1 in the network and between the UE 818 and the second node gNB2 in the network. The two nodes are only illustrative.

[0174] Each cell or node (e.g., gNB 1 or gNB 2) served by the base station can serve a very wide frequency band. First, the UE 818 can search the entire frequency band provided by the service provider (PLMN) in units of predetermined resource blocks (e.g., in units of 12 resource blocks (RBs)). That is, the UE 818 can start to discover the primary synchronization sequence (PSS) or the secondary synchronization sequence (SSS) in the entire system bandwidth in units of resource blocks. If the UE 818 discovers the PSS or SSS in units of resource blocks and then detects the signal, the UE can read the signal, analyze (decode) the signal, and identify the boundary between the subframe and the radio transmission resource frame (radio frame). If the UE 818 completes synchronization, the UE can read the system information of the cell where the UE currently resides. That is, in operations S800 and S802, the UE can identify information about the control resource set (CORESET) by identifying the master system information block (MIB) or the minimum system information (MSI), and identify the initial access bandwidth part (BWP) information by reading the system information. Above, the CORESET information refers to the position of the time / frequency transmission resources for transmitting control signals from the base station, and can be, for example, the position of the resources for transmitting the PDCCH channel. System information and SIB acquisition can be as Figure 7 detailed.

[0175] As described above, if the UE 818 synchronizes with the base station (gNB 1) for the downlink signal and is able to receive control signals, the next stage is to configure the RRC connection. For example, this can be achieved by performing a random access procedure in the initial BWP, including sending a preamble message from the UE to the base station at step S804. Then, at step S806, the base station may send a random access response received at the UE 818. Then, at step S808, the UE 818 sends an RRCConnectionRequest to request configuration of the RRC connection, and at step S810, receives an RRCConnectionSetup message from the base station. The UE may configure one or more of bearer setup information, cell group setup information, cell setup information, AI / ML configuration, per-layer device information (e.g., SDAP layer device (or new layer device), PDCP layer device, RLC layer device, MAC layer device, or PHY layer device) through the RRCConnectionSetup message. Once the RRC connection is configured or established at the UE 818, an RRCConnectionSetupComplete message confirming the completion of the setup is sent from the UE to the base station at step S812. The RRCConnectionSetup message may include configuration information for the PCell, Pscell, or multiple cells, and multiple partial bandwidths may be configured for each cell (PCell, Pscell, or Scell).

[0176] Steps S804 to S812 represent a process for establishing a basic RRC connection, for example, using SRB1. Once completed, at step S814, the base station (gNB 1) may transmit an RRC message (UECapabilityEnquiry) to the UE to identify the UE capabilities. The base station transmits an RRC message to the UE to identify the UE capabilities, for example, to identify: how many frequency bands the UE can read, whether the UE supports a function, or the frequency band region that the UE can read.

[0177] When receiving an RRC message that enquires about the UE capabilities, the UE performs a UE capability reporting process, and at step S816, may transmit an RRC message (UECapabilityInformation) including UE capability information related to the functions supported by the UE to the base station. The RRC message for reporting the UE capabilities (e.g., non-access stratum (NAS) message or access stratum (AS) message) may include some or multiple pieces of information for the AI / ML functions. Additionally, after identifying the UE capabilities, an appropriate partial bandwidth (BWP) or appropriate functions may be configured for the UE.

[0178] The base station can request the UE capabilities from the UE. In another method, the base station can ask the MME (Mobility Management Entity) or AMF (Access and Mobility Management Function) about the UE capabilities to identify the UE capabilities. This is because if the UE has accessed the MME or AMF previously, the MME or AMF may have the UE capabilities information.

[0179] For example, as a result of the received capabilities information, it may be necessary to reconfigure the UE, and thus the base station can send a reconfiguration message (e.g., RRCConnectionReconfiguration message) to the UE at step S818. In response to the reconfiguration message, the UE can configure bearer setup information, cell group setup information, cell setup information, AI / ML configuration, per-layer device information (e.g., SDAP layer device (or new layer device), PDCP layer device, RLC layer device, MAC layer device, or PHY layer device). The RRC message can include the configuration information of the PCell, Pscell, or multiple cells, and multiple partial bandwidths can be configured for each cell (PCell, Pscell, or Scell). When receiving the RRCConnectionReconfiguration message in which the UE's configuration information is received, the UE can apply the configuration information to the UE's bearers or layer devices. Once the reconfiguration is complete, at step S820, the UE sends a message to the base station confirming the completion of the configuration. This message can be an RRCConnectionReconfigurationComplete message. As indicated at step S822, the configuration of the connection is thus completed.

[0180] At step S824, the bi-directional arrow indicates that data can be transferred from the UE to the base station and vice versa. Multiple data transfer steps can be performed. As indicated at step S826, it may be necessary to reconfigure the UE, and thus the base station can send an RRCConnectionReconfiguration message to the UE at step S826. In response to the reconfiguration message, the UE can apply the configuration information to the UE's bearers or layer devices. Once the reconfiguration is complete, at step S828, the UE sends a message to the base station confirming the completion of the configuration. As indicated at step S830, the reconfiguration of the connection is thus completed. The reconfiguration information can relate to the same parameters as the configuration information sent at step S810 and the reconfiguration information sent at step S818.

[0181] Additionally, when the base station (gNB 1) or the network instructs the UE to switch to another cell or frequency, the base station (gNB 1) or the network may configure a handover message for the handover that includes configuration information of the target base station (gNBA 2). At step S832, the handover message (RRCConnectionReconfiguration message) is transmitted to the UE. Then, the UE may perform a handover process (e.g., either a synchronization process or a random access process to the target base station gNB 2 as shown at step S836) according to the handover settings. The UE may configure an RRCConnectionReconfigurationComplete message and transmit this message to the target base station at step S834 when the handover is successfully performed. The configuration information of the target base station gNB 2 may include bearer configuration information, cell group configuration information, cell configuration information, AI / ML configuration, per-layer device information (e.g., SDAP layer device (or new layer device), or PDCP layer device, or RLC layer device, MAC layer device or PHY layer device). Other configuration information as described above may also be included.

[0182] Figure 6 It is shown that an RRC message for resuming the connection can be sent to move from the inactive state to the connected state. The resume message may include some or all of the configuration information described above for the initial configuration or reconfiguration. Further details are provided below, including pseudocode for RRC connection control.

[0183] Example use cases

[0184] AI / ML model training at the central server and AI / ML model inference at the nodes

[0185] Figure 9 is a flowchart showing the communication between the UE 918 and the network, which includes a central server 930 (e.g., an Operations, Administration, and Management (OAM) server) and multiple nodes, including a first node 926 and a second node 928. In this example, AI / ML model training is performed at the OAM server 930, and AI / ML model inference is performed at one or more nodes. In this solution, the nodes use the AI / ML model to make decisions to achieve network energy saving, load balancing, mobility optimization, and other purposes. For UE configuration, the network may send an RRC message (e.g., RRCReconfiguration or RRCSetup or RRCResume or a newly defined RRC message) to the UE to configure the type of AI / ML configuration such as algorithms or models, related timers, parameters for measurement (e.g., beam, CSI, SSB, frequency), conditions, or the type of report (or collection), etc.

[0186] In a first optional step S930, an AI / ML model can be provided to (or assumed to be had by) the second node 928 (NG-RAN node 2), and the AI / ML model can provide input information to the first node 926 (NG-RAN node 1). At step S932 (step 1-1), UE capability information can be transmitted from the UE 918 to the network (specifically, the first node 926) upon request. For a UE in the CONNECTED state, this can be done as described above in Figure 8 For example, the network (or specifically, the first node 926) broadcasts system information including an indication of AI / ML function support, and the UE can report UE capabilities as described above. When the network requests the UE to report UE capability information by sending a UE CapabilityEnquiry message, the UE can send a UEcapabilityInformation message including the capability of AI / ML function support.

[0187] Then, at step S934 (step 1-2), configuration information is sent from the first node 926 to the UE 918, for example, as described above. The NG-RAN node 1 configures the UE with AI / ML configuration (e.g., based on the received capability information), and sends a configuration message (e.g., RRCReconfiguation, RRCSetup, RRCResume, RRCRelease message or a newly defined message) to the UE to perform AI / ML functions (e.g., measurement procedures and reporting). The NG-RAN node 1 can also configure the UE to provide AI / ML data (e.g., measurement and / or location information, RRM measurements, MDT measurements, speed, location, time information for measurement, measurement duration, sequence number for reordering training data, or CSI (e.g., aperiodic CSI or periodic CSI), or measurement results of frequency (or cell identity or beam or SSB)).

[0188] The AI / ML configuration information may include configuring one of the signaling radio bearers (SRBs), specifically SRB1, SRB2, SRB3, SRB4, or SRBx, such that the UE reports AI / ML data to the network via the SRB by sending appropriate RRC messages. In another embodiment, the AI / ML configuration information may include, for example, configuring one of the data radio bearers (DRBs) with a special logical channel identifier or bearer identifier, such that the UE reports AI / ML data to the network via the DRB by sending appropriate RRC messages. The AI / ML configuration information may include configuring radio resources (e.g., the time and frequency resources of the PUCCH (Physical Uplink Control Channel) or PUSCH (Physical Uplink Shared Channel)), such that the UE reports AI / ML data via the configured resources (e.g., UL grant or SPS (Semi-Persistent Scheduling) grant (i.e., radio resources) or configured grant).

[0189] The configuration information of the SRB or DRB may include AS security configuration, i.e., whether to perform encryption (or integrity protection). A large amount of AI / ML data may need to be reported to the network, i.e., frequent reporting may be required. To reduce the processing burden on the UE caused by encryption or integrity protection, one or both of the encryption function or integrity protection function may not be configured for the SRB or DRB. In other words, only encryption may be configured, or only integrity protection may be configured, or neither encryption nor integrity protection may be configured. In another alternative, the RRC layer may indicate to the PDCP layer via an indication whether each RRC message (or each piece of AI / ML data) requires integrity protection and whether each RRC message (or AI / ML data) requires encryption. It may be applied to the configured SRB or DRB for AI / ML data. In another alternative, the configured AS security (encryption or integrity protection) of the SRB (or DRB) for AI / ML data reporting may be activated or deactivated via an RRC message or MAC CE or PDCP control PDU or PDCCH DCI to control the UE processing burden.

[0190] As an alternative solution to the need to report a large amount of AI / ML data, the UE may avoid frequent reporting by including multiple pieces of AI / ML data information in the newly proposed message structure of one RRC message (e.g., a measurement report message or an RRC message for reporting). In other words, the UE may report a large amount of AI / ML data with a single message (e.g., an RRC message or MAC CE). The proposed message may include a sequence number or timing information for each piece of AI / ML data so that the receiver can sort them in chronological order. The number of AI / ML data to be included in the report message may be configured by an RRC message (e.g., in the AI / ML configuration). The detailed message structure is discussed below.

[0191] The configuration information may additionally or alternatively include prioritization information for AI / ML data. Given that it is very important to send / receive common RRC messages with low latency and high reliability due to the UE's connection management and mobility support, the AI / ML data may not be urgent, i.e., not time-sensitive. Therefore, a prioritization mechanism may be configured and used to prioritize other RRC messages over the reporting messages for AI / ML data.

[0192] The nature of the SRB (i.e., SRB1 or SRB2 or SRB3 or SRB4) may be configured and used to transmit / receive RRC messages including AI / ML data. These RRC messages may have a lower priority than the first RRC messages because the first RRC messages support the UE's connection management and mobility support and are thus much more important than the RRC messages for AI / ML data. These first RRC messages may include piggybacked NAS messages, NAS messages before the establishment of SRB2, RRC messages for establishing (re)connections and (re)configurations, including for example: RRCSetupRequest, RRCSetup, RRCSetupComplete, RRCResumeRequest, RRCResume, RRCResumeComplete, RRCReconfiguration, RRCReconfigurationComplete, RRCReestablishment, RRCReestablishmentRequest, etc. When the first RRC messages and the RRC messages for AI / ML data are generated together or simultaneously (or coexist in the buffer), the first RRC messages may then be prioritized over the RRC messages for AI / ML and may be submitted to the lower layer (e.g., PDCP layer) and sent first.

[0193] As an alternative, SRBx or DRB may be configured and used only to transmit / receive messages including AI / ML data. As another alternative, for AI / ML data through the configured RB (SRB or DRB), the network may assign / modify the semi-persistent scheduling (SPS) configuration and the configured grant configuration for DL and UL respectively, assign / modify the parameters of the UL rate control in the UE (i.e., assign the priority and PBR (LCP parameters of the LCP process in the MAC layer) for each RB and logical channel identifier of the UE), and assign a separate SCell configuration or a separate bandwidth part (BWP) configuration in order to control the prioritization between the messages including AI / ML data and other messages (or data).

[0194] The configuration information may include the configuration of AI / ML data and the AI / ML configuration (e.g., the feature CSI-MeasConfig or MeasConfig specified below).

[0195] When the UE starts collecting AI / ML data (or performing measurements) via an appropriate RRC message (e.g., RRCReconfiguration or RRCRelease), the network may configure conditions (e.g., thresholds, explicit indications, timers, etc.). These conditions can be configured regardless of the UE's state (RRC connection). The network may send an RRCReconfiguration message including the configuration conditions detailed below to the UE in the RRC connected state.

[0196] As an example, the threshold can be configured to enable the UE to know the conditions when the UE starts collecting AI / ML data (or performing measurements). When the conditions are met, the UE starts collecting AI / ML data. For example, if a specific metric is greater than (or less than) a threshold, and / or if another specific metric (e.g.) is greater than (or less than) another threshold, the UE may consider the conditions to be met. The specific metric can be RSRP (Reference Signal Received Power) or RSRQ (Reference Signal Received Quality) or SINR (Signal-to-Interference-plus-Noise Ratio) or time, etc. Based on the configuration, one or more than one threshold can be used to generate the conditions. The network may configure the periodicity with which the UE should collect AI / ML data (or perform measurements).

[0197] As another example, the network may configure the conditions or types or sizes via RRC messages to enable the UE to know which AI / ML data should be collected (or how much AI / ML data should be collected). The network may send an explicit indication to the UE via an appropriate RRC message (e.g., RRCReconfiguration or RRCRelease message) or MAC CE (Control Element) or PDCCH (Physical Downlink Control Channel) or DCI (Downlink Control Information) regarding when to start (or activate) or stop (or deactivate) collecting AI / ML data (or performing measurements) or which measurement object should be measured (e.g., CSI or beam or SSB or frequency or cell identity). When the UE receives the indication via the RRC message, the UE may start or stop collecting AI / ML data that meets the conditions configured by the network and store it.

[0198] As another example, the network may configure a timer (e.g., T3xx) for the UE via an RRC message (e.g., RRCReconfiguration or RRCRelease message). After configuring the timer, the UE may start the timer and collect AI / ML data (or perform measurements) while the timer is running. After the timer expires, the UE may stop collecting AI / ML data. The area information may be configured to cause the UE to collect AI / ML data (i.e., perform measurements) within the configured area. The configured area may be defined by a cell identifier (i.e., physical cell identifier or TAI (tracking area information)). When the UE is outside this area, the UE may stop collecting AI / ML data or may stop the timer.

[0199] The network may configure the conditions (e.g., thresholds, explicit indications, timers, etc.) under which a UE (in RRC connected state) reports AI / ML data (e.g., reports measurement results) and the radio resources (e.g., time / frequency resources) on which the UE sends the AI / ML data (or reports measurement results) via an RRC message (e.g., RRCReconfiguration or RRCRelease). The network may send an RRCReconfiguration message to the UE in the RRC connected state that includes the following configurations.

[0200] For example, the network may configure the periodicity with which the UE should report AI / ML data (or report measurement results). The UE may periodically send (or report) AI / ML data to the network via an RRC message or MAC CE or configured grant for PUCCH or PUSCH (i.e., radio resources). As another example, after receiving an explicit indication from the network, the UE may report AI / ML data (or report measurement results) via an RRC message or MAC CE or configured grant for PUCCH or PUSCH (i.e., time / frequency radio resources). The network may send an explicit indication to the UE via an RRC message (e.g., RRCReconfiguration or RRCRelease message) or MAC CE (control element) or PDCCH (physical downlink control channel) or DCI (downlink control information) as to when or whether or where (e.g., radio resources for reporting) to report AI / ML data (or perform measurements) or which measurement object (e.g., CSI or beam or SSB or frequency or cell identifier) should be measured or of a change in the target AI / ML data.

[0201] As another example, the network may configure a timer (e.g., T3xx) for the UE via an RRC message (e.g., RRCReconfiguration or RRCRelease message). After the timer is configured, the UE may start the timer. Whenever the timer expires, the UE may report AI / ML data (or report measurement results) via an RRC message or a MAC CE or a configured grant for PUCCH or PUSCH (i.e., time / frequency radio resources). After reporting (or sending AI / ML data), the UE may restart the timer.

[0202] As another example, a threshold may be configured to enable the UE to know the conditions under which the UE reports AI / ML data (or performs measurements). When the conditions are met, the UE reports AI / ML data. For example, if a specific metric is greater than (or less than) a threshold, and / or if another specific metric (e.g.) is greater than (or less than) another threshold, the UE may consider the conditions to be met. The specific metric may be RSRP (Reference Signal Received Power) or RSRQ (Reference Signal Received Quality) or SINR (Signal-to-Interference-plus-Noise Ratio) or time, etc. Based on the configuration, one or more thresholds may be used to generate the conditions.

[0203] At step S926 (step 2), as proposed above, the UE collects AI / ML data or indicated measurements based on the configured conditions or periodically (e.g., for the configured radio resources) or after receiving an explicit indication or when the timer expires. Examples of UE measurements are related to RSRP, RSRQ, and SINR of the serving cell and neighboring cells. As shown at step S938 (step 3-1), there is an optional request for reporting measurements from the first node to the UE. As proposed above, the UE sends AI / ML data or a measurement report message to the NG-RAN node 1 (at step S940 - step 3-2). The network may optionally send an RRC message (e.g., UEInformationRequest message) including an indication of the AI / ML data report, and the UE may report via an RRC message (e.g., UEInformationResponse message). Alternatively, the UE may send based on the configured conditions or periodically (e.g., for the configured radio resources) or when the timer expires, rather than only when receiving an explicit indication. In this example, only one UE is shown, but it should be understood that the first node (NG-RAN) may collect AI / ML data from more UEs or other nodes (NG-RAN).

[0204] At step S942 (step 4), the first node (NG-RAN node 1) sends AI / ML data or UE measurement reports and other input data for model training to OAM 930. The AI / ML data can be included in the inter-node message between the NG-RAN node and OAM. At step S944 (step 5), the second node (NG-RAN node 2 with an optional AI / ML model) also sends AI / ML data or input data for model training to OAM 930. NG-RAN node 2 sends AI / ML data or input data for training, where the input data for training includes the required input information from NG-RAN node 2. If NG-RAN node 2 executes the AI / ML model, the transmitted AI / ML data or input data for training can include the corresponding inference results from NG-RAN node 2. The AI / ML data can be included in the inter-node message between the second node (NG-RAN node 2) and OAM. In each of steps S942 and S944, the inter-node message can also include an indication of whether the AI / ML data is for model training or model inference or model performance feedback or model deployment or update. For this purpose, a new inter-node message can be defined. In another embodiment, traditional inter-node messages can be used for this purpose by introducing a new container for the AI / ML data and new indication / parameters in the message.

[0205] At step S946 (step 6), the model is trained at OAM 930. The AI / ML model can also be trained using AI / ML data or required measurements and input data from other nodes to achieve network energy saving / load balancing / mobility optimization.

[0206] At step S948 (step 7), OAM 930 deploys or updates the AI / ML model into the NG-RAN nodes. For simplicity, only the deployment to the first node is shown. The or each NG-RAN node can also continue model training based on the AI / ML model received from OAM. OAM 930 can send an AI / ML model deployment message to deploy the trained / updated AI / ML model into the NG-RAN nodes. Alternatively, the AI / ML model or update parameters can be included in the inter-node message between the or each NG-RAN node and OAM. The inter-node message can also include an indication for indicating whether this is for model deployment or model update. For this purpose, a new inter-node message can be defined. In another embodiment, traditional inter-node messages can be used for this purpose by introducing a new container for the AI / ML model and new indication / parameters in the message.

[0207] At step S950 (step 8), the second node (NG-RAN node 2) sends AI / ML data or the required input data to the first node (NG-LAN node 1) for model inference. NG-RAN node 1 receives AI / ML data or input information from the adjacent NG-RAN node 2 for model inference. The AI / ML data can be included in the inter-node message between the NG-RAN nodes. The inter-node message can also include an indication of whether the AI / ML data is for model training or model inference or model performance feedback or model deployment or update. For this purpose, the inter-node message can be newly defined. In another embodiment, the traditional inter-node message can be used for this purpose by introducing a new container for the AI / ML data and new indication / parameters in the message.

[0208] As shown at step S952 (step 9-1), there is an optional request for the reported measurement from the first node to the UE. As proposed above, the UE sends AI / ML data or a measurement report message to NG-RAN node 1 (at step S954 - step 392). The same process as explained with respect to steps S938 and S940 can be followed.

[0209] At step S956 (step 10), NG-RAN node 1 generates a model inference output (e.g., energy saving policy, load balancing, handover policy, mobility optimization, etc.). The inference can be based on the AI / ML data and / or local input of the first node (NG-RAN node 1) and any AI / ML data or input received from the second node (NG-RAN node 2). The first node (NG-RAN node 1) performs model inference and generates a prediction or decision. The AI / ML data or the required measurements are used in the model inference to output predictions, e.g., UE trajectory prediction, target cell prediction, target NG-RAN node prediction, etc.

[0210] At step S958 (step 11), there is an option step where the first node (NG-RAN node 1) sends model performance feedback to the OAM if applicable. The AI / ML data can be included in the inter-node message between the NG-RAN node and the OAM. The inter-node message can also include an indication of whether the AI / ML data is for model training or model inference or model performance feedback or model deployment or update. For this purpose, the inter-node message can be newly defined. In another embodiment, the traditional inter-node message can be used for this purpose by introducing a new container for the AI / ML data and new indication / parameters in the message.

[0211] At step S960 (step 12), which is shown sequentially after the previous step but can also be simultaneous, the first node (NG-RAN node 1) performs any action based on the model inference output, e.g., a network energy saving action or a load balancing action or a mobility optimization and / or handover procedure. If the output is a handover policy, NG-RAN node 1 can select the target cell (e.g., another node) most suitable for each UE before performing the handover for each UE. NG-RAN node 1 can take a load balancing action, e.g., a UE can move from its current node (NG-RAN node 1) to a different node (e.g., NG-RAN node 2). Based on the prediction, recommended action or configuration, NG-RAN node 1, the target NG-RAN node (represented by NG-LAN node 2 in this step of the flowchart), and the UE perform a mobility optimization and / or handover procedure to handover the UE from NG-RAN node 1 to the target NG-RAN node. AI / ML data (e.g., actions) can be included in the inter-node message between NG-RAN nodes (e.g., between NG-RAN node 1 and NG-RAN node 2). The inter-node message can also include an indication of whether the AI / ML data is used to convey an action or a decision. For this purpose, an inter-node message can be newly defined. In another embodiment, traditional inter-node messages can be used for this purpose by introducing a new container for AI / ML data and a new indication / parameter in the message.

[0212] At step S962 (step 13), the target node (e.g., NG-RAN node 2) can provide feedback to the OAM. Similarly, at step S964 (step 14), the first node (NG-RAN node 1) can provide feedback to the OAM. In both cases, AI / ML data (e.g., feedback) can be included in the inter-node message between NG-RAN nodes (e.g., between NG-RAN node 1 and NG-RAN node 2). The inter-node message can also include an indication of whether the AI / ML data is for model training or model inference or model performance feedback or model deployment or update. For this purpose, an inter-node message can be newly defined. In another embodiment, traditional inter-node messages can be used for this purpose by introducing a new container for AI / ML data and a new indication / parameter in the message.

[0213] AI / ML model training at the first node and AI / ML model inference at the same or different nodes

[0214] Figure 10It is a flowchart showing the communication between UE 1018 and multiple nodes, including the first node 1026 and the second node 1028. In this example, AI / ML model training is performed at the first node 1026, and AI / ML model inference is performed at one or more nodes. In this solution, the nodes use the AI / ML model to make decisions to achieve network energy saving, load balancing, mobility optimization, and other purposes. For UE configuration, the network can send RRC messages (e.g., RRCReconfiguration or RRCSetup or RRCResume or newly defined RRC messages) to the UE to configure the type of AI / ML configuration such as algorithms or models, related timers, parameters for measurement (e.g., beam, CSI, SSB, frequency), conditions, or the type of report (or collection), etc.

[0215] In the first optional step S1030, an AI / ML model can be provided to (or assumed to have) the second node 1028 (NG-RAN node 2), and the AI / ML model can provide input information to the first node 1026 (NG-RAN node 1). At step S1032 (step 1-1), UE capability information can be transmitted from UE 1018 to the network (specifically, the first node 1026) upon request. For a UE in the CONNECTED state, this can be done as described above in Figure 8 For example, the network (or specifically, the first node 1026) broadcasts system information including an indication of AI / ML function support, and the UE can report its UE capability as described above. When the network requests the UE to report UE capability information by sending a UE CapabilityEnquiry message, the UE can send a UEcapabilityInformation message including the capability of AI / ML function support.

[0216] Then, at step S1034 (step 1-2), configuration information is sent from the first node 1026 to the UE 1018, for example, as described above. The NG-RAN node 1 configures the UE with AI / ML configuration (e.g., based on the received capability information), and sends a configuration message (e.g., RRCReconfiguation, RRCSetup, RRCResume, RRCRelease message or newly defined message) to the UE to perform AI / ML functions (e.g., measurement procedures and reporting). The NG-RAN node 1 can also configure the UE to provide AI / ML data (e.g., measurement and / or location information, RRM measurements, MDT measurements, speed, location, time information for measurement, measurement duration, sequence number for reordering training data, or CSI (e.g., aperiodic CSI or periodic CSI), or measurement results of frequency (or cell identity or beam or SSB)). The details of the AI / ML configuration can be the same as those regarding Figure 9 described.

[0217] At step S1026 (step 2), as proposed above, the UE collects AI / ML data or indicated measurements based on configured conditions or periodically (e.g., for configured radio resources) or after receiving an explicit indication or when a timer expires. Examples of UE measurements are related to the RSRP, RSRQ, and SINR of the serving cell and neighboring cells. As shown at step S1038 (step 3-1), there is an optional request from the first node to the UE for reporting measurements. As proposed above, the UE sends AI / ML data or a measurement report message to the NG-RAN node 1 (at step S1040 - step 3-2). The network can optionally send an RRC message (e.g., UEInformationRequest message) including an indication for AI / ML data reporting, and the UE can report via the RRC message (e.g., UEInformationResponse message). Alternatively, the UE can send based on configured conditions or periodically (e.g., for configured radio resources) or when a timer expires, rather than only when receiving an explicit indication. In this example, only one UE is shown, but it should be understood that the first node (NG-RAN) can collect AI / ML data from more UEs or other nodes (NG-RAN). When the UE is in the RRC connected state, the UE can perform AI / ML data collection and reporting based on configuration or a specific request.

[0218] At step S1042 (step 4), the second node (NG-RAN node 2) sends AI / ML data or UE measurement reports and other input data for model training to the first node (NG-RAN node 1). If NG-RAN node 2 executes an AI / ML model, the transmitted AI / ML data or input data for training may include corresponding inference results from NG-RAN node 2. The AI / ML data may be included in the inter-node message between the nodes. The inter-node message may also include an indication of whether the AI / ML data is for model training or model inference or model performance feedback or model deployment or update. For this purpose, a new inter-node message may be defined. In another embodiment, the traditional inter-node message may be used for this purpose by introducing a new container for the AI / ML data and new indication / parameters in the message.

[0219] At step S1044 (step 5), the model is trained at the first node (NG-RAN node 1) based on the collected data for configuration purposes. AI / ML data or required measurements and input data from other nodes may also be utilized to train the AI / ML model to achieve network energy saving / load balancing / mobility optimization.

[0220] At step S1046 (step 6), the second node (NG-RAN node 2) sends AI / ML data (e.g., required input data) to the first node (NG-LAN node 1) for model inference. NG-RAN node 1 receives AI / ML data (e.g., UE measurements and / or location information) from the adjacent NG-RAN node 2 for model inference. The AI / ML data may be included in the inter-node message between the NG-RAN nodes. The inter-node message may also include an indication of whether the AI / ML data is for model training or model inference or model performance feedback or model deployment or update. For this purpose, a new inter-node message may be defined. In another embodiment, the traditional inter-node message may be used for this purpose by introducing a new container for the AI / ML data and new indication / parameters in the message.

[0221] As shown at step S1048 (step 7), there is an optional request for reported measurements from the first node to the UE. As proposed above, the UE sends AI / ML data or measurement report messages to NG-RAN node 1 (at step S1050 - step 7-2). The same process as explained with respect to steps S1038 and S1040 may be followed.

[0222] At step S1052 (step 8), the NG-RAN node 1 generates model inference outputs (e.g., energy saving policies, load balancing, handover policies, mobility optimization, etc.). The inference can be based on the AI / ML data and / or local inputs of the first node (NG-RAN node 1) and any AI / ML data or inputs received from the second node (NG-RAN node 2). The first node (NG-RAN node 1) performs model inference and generates predictions or decisions. AI / ML data or required measurements are used in the model inference to output predictions, e.g., UE trajectory prediction, target cell prediction, target NG-RAN node prediction, etc.

[0223] At step S1054 (step 9), the first node (NG-RAN node 1) performs any actions according to the model inference output, e.g., network energy saving actions or load balancing actions or mobility optimization and / or handover procedures. If the output is a handover policy, the NG-RAN node 1 can select the target cell (e.g., another node) most suitable for each UE before performing the handover. The NG-RAN node 1 can take load balancing actions, e.g., the UE can move from its current node (NG-RAN node 1) to a different node (e.g., NG-RAN node 2). According to the prediction, recommended actions or configurations, the NG-RAN node 1, the target NG-RAN node (represented by the NG-LAN node 2 in this step of the flowchart), and the UE perform mobility optimization and / or handover procedures to hand over the UE from the NG-RAN node 1 to the target NG-RAN node. AI / ML data (e.g., actions) can be included in the inter-node messages between NG-RAN nodes (e.g., between NG-RAN node 1 and NG-RAN node 2). The inter-node messages can also include an indication of whether the AI / ML data is used to convey actions or decisions. For this purpose, inter-node messages can be newly defined. In another embodiment, traditional inter-node messages can be used for this purpose by introducing a new container for AI / ML data and new indications / parameters in the messages.

[0224] At step S1056 (step 10), the target node (e.g., NG-RAN node 2) can provide feedback to the first node (e.g., NG-RAN node 1). AI / ML data (e.g., feedback) can be included in the inter-node messages between NG-RAN nodes (e.g., between NG-RAN node 1 and NG-RAN node 2). The inter-node messages can also include an indication of whether the AI / ML data is for model training or model inference or model performance feedback or model deployment or update. For this purpose, inter-node messages can be newly defined. In another embodiment, traditional inter-node messages can be used for this purpose by introducing a new container for AI / ML data and new indications / parameters in the messages.

[0225] Other examples

[0226] In Figure 9 the arrangement, model training is performed in the central server 930, and model inference is performed in the node 926. In Figure 10 the arrangement, both model training and model inference are performed in the node 1026. It should be understood that other arrangements are also possible. For example, both model training and model inference can be performed in the UE. In this case, the AI / ML model (or the AL / ML model functions (e.g., training and inference)) is configured for the UE via appropriate RRC messages or newly defined messages. Such an arrangement can be used for federated learning or split learning. As another example, model training is performed in the UE, and model inference is performed at the node or the central server. As another example, model training is performed on the node or the central server, and model inference is performed in the UE.

[0227] In any arrangement, the UE is configured as needed to collect AI / ML data and / or receive AI / ML data from the network (e.g., from the node (NG-RAN) or the central server (OAM)). Based on the received / collected AI / ML data, the UE can train (or update) the AI / ML model and / or generate model inference output or take actions. The UE can also send the trained AI / ML model or inference output or updated parameters to the network (e.g., the node (NG-RAN) or the central server (OAM)). For such delivery, the AI / ML model or updated parameters or AI / ML data can be included in the RRC message (or NAS message) between the NG-RAN node (or OAM) and the UE. The inter-node message can also include an indication of whether this is for model training or model inference (or inference output) or model performance feedback or model deployment or update. For this purpose, a new RRC message (or NAS (non-access stratum) message) can be defined. In another embodiment, the traditional RRC message (or NAS message) can be used for this purpose by introducing a new container for the AI / ML model and new indication / parameters in the message. In other words, the inter-node message can indicate an Xn message (i.e., a message via the Xn interface between gNBs) or a NAS message or an RRC message.

[0228] Input, output, and feedback

[0229] As described in the above examples, the UE, node, and / or central server with the updated model require various inputs during the inference phase. Similarly, there can be more than one output during the inference phase. The input and output will depend on the nature of the model, e.g., whether the model is for network energy saving, load balancing, or mobility optimization. These examples also mention that feedback can be provided by different parts of the network, and details of the feedback types are provided below.

[0230] For example, when the model is predicting optimized network energy-saving decisions, the nodes of the model (NG-RAN), the central server, or the UE may require the following information as input data to achieve AI / ML-based network energy saving. The model can use inputs from local nodes (i.e., any node within the network that is associated with the UE and is not the central server), and these inputs can include some or all of the following: UE mobility / trajectory prediction, current and / or predicted energy efficiency, and / or current and / or predicted resource status. The model can use inputs from the UE, and these inputs can include UE location information (e.g., coordinates, serving cell ID, movement speed) and UE measurement reports (e.g., UE RSRP, RSRQ, SINR measurements, etc.). The UE location information can be interpreted through the gNB implementation when available. The UE measurement reports can include cell-level and beam-level UE measurements. The model can use inputs from adjacent nodes, and these inputs can include current and / or predicted energy efficiency, current and / or predicted resource status, and / or current energy state (e.g., active, high, low, inactive). If the model (e.g., at the gNB) requires existing UE measurements to achieve AI / ML-based network energy saving, RAN3 should reuse the existing framework (including MDT and RRM measurements).

[0231] The output generated from the AI / ML-based network energy-saving model can include: energy-saving policies, such as recommended cell activation / deactivation; handover policies, including recommended candidate cells for taking over traffic; predicted energy efficiency; predicted energy state (e.g., active, high, low, inactive); and / or the model output validity time. To optimize the performance of the AI / ML-based network energy-saving model, feedback can be collected from the NG-RAN nodes, including the resource status, energy efficiency, and UE performance (e.g., handovered UE) affected by the energy-saving actions of adjacent NG-RAN nodes, including: bit rate, packet loss, latency, and system KPIs (e.g., throughput, delay, RLF of the current and adjacent NG-RAN nodes).

[0232] As another example, when the model is used for load balancing decisions, the nodes (NG-RAN), central server, or UEs maintaining the model may require the following information as input data to achieve AI / ML-based network energy savings. The model can use inputs from local nodes, and these inputs can include UE mobility / trajectory predictions and current and / or predicted resource conditions, as in the previous example. The model can also use current and predicted UE traffic and predicted resource condition information from adjacent NG-RAN nodes of the local node. The model can use inputs from the UE, and these inputs can include UE location information (e.g., coordinates, serving cell ID, movement speed) and UE measurement reports (e.g., UE RSRP, RSRQ, SINR measurements, etc.), as in the previous example. The model can also use the following inputs from the UE: UE mobility history information. The model can use inputs from adjacent nodes, and these inputs can include current and / or predicted resource conditions, as in the previous example. The model can also use UE performance measurements at adjacent cells where traffic is offloaded.

[0233] Outputs generated from the AI / ML-based load balancing model can include some or all of the following: selection of the target cell for load balancing, predicted own resource condition information, predicted resource condition information of adjacent NG-RAN nodes, model output validity time, and predicted UEs (to be used internally by the RAN node) to be switched to the target NG-RAN node. To optimize the performance of the AI / ML-based load balancing model, feedback can be collected from the NG-RAN nodes, including some or all of the UE performance information from the target NG-RAN (for those UEs switched from the source NG-RAN node), updated resource condition information from the target NG-LAN, and system KPIs (e.g., throughput, latency, current and adjacent RLF).

[0234] As another example, when the model is used for mobility optimization, the nodes (NG-RAN), central server, or UEs maintaining the model may require the following information as input data to achieve AI / ML-based network energy savings. The model can use inputs from local nodes, and these inputs can include UE mobility / trajectory prediction, current and / or predicted resource conditions, and predicted UE traffic, as in the previous example. The model can use inputs from the UE, and these inputs can include UE location information (e.g., coordinates, serving cell ID, moving speed) and UE mobility history information, as in the previous example. The model can also use the following inputs from the UE: radio measurements related to the serving cell and neighboring cells associated with the UE location information, e.g., RSRP, RSRQ, SINR. The model can use inputs from neighboring nodes, and these inputs can include current and / or predicted resource conditions, as in the previous example. The model can also use historical information, location, QoS parameters, and performance information of UEs that have been handed over (e.g., loss rate, latency, etc.) from the neighbors of the UE, as well as past successful and unsuccessful UE handovers, including too early, too late, or handover to the wrong (sub-optimal) cell, based on the existing SON / RLF reporting mechanism.

[0235] Outputs generated from the AI / ML-based mobility optimization model can include some or all of the following: UE trajectory prediction (latitude, longitude, altitude, cell ID of the UE over a period of time in the future), estimated arrival probability and associated confidence intervals in the CHO, predicted handover target node, candidate cells, priorities, handover execution timing in the CHO, predicted resource reservation time window of the CHO, UE traffic prediction, and model output validity time. The predicted handover target node can be output together with the predicted confidence. It should also be noted that whether the UE trajectory prediction is an external output for the node hosting the model inference function should be discussed during the specification working phase. The UE traffic prediction will be used internally by the RAN node, and the details are left to the specification working phase. To optimize the performance of the AI / ML-based mobility optimization model, feedback can be collected from the NG-RAN nodes, including some or all of the following: QoS parameters such as throughput, packet delay of the handovered UEs, resource condition information updates from the target NG-RAN, and performance information from the target NG-RAN.

[0236] Measurement

[0237] Measurements include (or indicate) the collection of AI / ML data, measurement configurations include (or indicate) AI / ML configurations, measurement results include (or indicate) AI / ML information, events for measurement triggering include (or indicate) events for AI / ML collection, and measurement reports include (or indicate) AI / ML data reports.

[0238] The network may configure an RRC_CONNECTED UE to perform measurements. The network may configure the UE to report according to a measurement configuration, or to perform conditional reconfiguration evaluation according to a conditional reconfiguration. The measurement configuration is provided by dedicated signaling, i.e., using an appropriate RRC message, e.g., an RRCReconfiguration or RRCResume message.

[0239] The network may configure the UE to perform some or all of the following types of measurements: NR measurements; inter-RAT measurements on E-UTRA frequencies; inter-RAT measurements on UTRA-FDD frequencies; and NR side-link measurements for an L2 U2N relay UE.

[0240] The network may configure the UE to report some or all of the following measurement information based on SS / PBCH blocks: measurement results for each SS / PBCH block; measurement results for each cell based on an SS / PBCH block; and an SS / PBCH block index. The network may configure the UE to report some or all of the following measurement information based on CSI-RS resources: measurement results for each CSI-RS resource; measurement results for each cell based on a CSI-RS resource; and a CSI-RS resource measurement identifier.

[0241] The network may configure the UE to perform the following types of measurements on NR side-links and V2X side-links: CBR measurements. The network may configure the UE to report the following CLI measurement information based on SRS resources: measurement results for each SRS resource; and an SRS resource index. The network may configure the UE to report the following CLI measurement information based on CLI-RSSI resources: measurement results for each CLI-RSSI resource; and a CLI-RSSI resource index. The network may configure the UE to report the following Rx-Tx time difference measurement information based on CSI-RS or PRS for tracking: UE Rx-Tx time difference measurement results.

[0242] The measurement configuration includes some or all of the following parameters: measurement objects, reporting configurations, measurement identities, quantity configurations, and measurement gaps, which will be described in more detail below.

[0243] The Measurement Object (MO) is a list of objects that the UE should perform measurements on. For intra-frequency and inter-frequency measurements, the measurement object indicates the frequency / time location of the reference signal to be measured and the subcarrier spacing. Associated with this measurement object, the network can configure a cell-specific offset list, a list of "excluded" cells, and a list of "allowed" cells. Cells in the excluded list are not applicable for event evaluation or measurement reporting. Cells in the allowed list are the only cells applicable for event evaluation or measurement reporting. The measObjectId of the measurement object corresponding to each serving cell is indicated by servingCellMO within the serving cell configuration.

[0244] For inter-RAT E-UTRA measurements, the measurement object is a single E-UTRA carrier frequency. Associated with this E-UTRA carrier frequency, the network can configure a cell-specific offset list and a list of "excluded" cells. Cells in the excluded list are not applicable for event evaluation or measurement reporting. For inter-RAT UTRA-FDD measurements, the measurement object is a set of cells on a single UTRA-FDD carrier frequency. For NR sidelink measurements of L2 U2N relay UEs, the measurement object is a single NR sidelink frequency to be measured. For CBR measurements of NR sidelink communication, the measurement object is a set of transmit resource pools on a single carrier frequency for NR sidelink communication. For CBR measurements of NR sidelink discovery, the measurement object is a set of discovery-specific resource pools or transmit resource pools also used for NR sidelink discovery on a single carrier frequency for NR sidelink discovery. For CLI measurements, the measurement object indicates the frequency / time location of the SRS resource and / or CLI-RSSI resource, and the subcarrier spacing of the SRS resource to be measured.

[0245] A reporting configuration is a list of reporting configurations, where one or more reporting configurations can exist for each measurement object. Each measurement reporting configuration includes some or all of the following information. The configuration can include a reporting criterion, which is the criterion for triggering the UE to send a measurement report. This can be periodic or a single event description. The configuration can include the RS type indicating the RS that the UE uses for beam and cell measurement results (SS / PBCH block or CSI-RS). The configuration can include a reporting format, for example, the number of each cell and each beam that the UE includes in the measurement report (e.g., RSRP), and other associated information, such as the maximum number of cells to be reported and the maximum number of beams per cell.

[0246] In the case of conditional reconfiguration, some or all of the following information is included. The configuration may include an execution criterion, which is the criterion used by the UE for conditional reconfiguration execution. The configuration may include an RS type indicating the RS that the UE uses to obtain beam and cell measurement results (based on SS / PBCH blocks or based on CSI-RS), and this RS type is used to evaluate the conditional reconfiguration execution condition.

[0247] For a measurement report, there may be a list of measurement identities, where each measurement identity links a measurement object to a reporting configuration. By configuring multiple measurement identities, more than one measurement object can be linked to the same reporting configuration, and more than one reporting configuration can be linked to the same measurement object. The measurement identity is also included in the measurement report that triggers the report as a reference for the network. For conditional reconfiguration triggering, one measurement identity exactly links to one conditional reconfiguration triggering configuration. At most two measurement identities can be linked to one conditional reconfiguration execution condition.

[0248] The measurement configuration defines the measurement filtering configuration for all event evaluations and related reports and for the periodic reporting of measurements. For NR measurements, the network can configure up to two measurement configurations, and the configuration to be used is referenced in the NR measurement object. In each configuration, different filter coefficients can be configured for different measurement quantities, different RS types, and for each cell and each beam.

[0249] A measurement gap is a period during which the UE can perform measurements.

[0250] A UE in the RRC_CONNECTED state maintains a list of measurement objects, a list of reporting configurations, and a list of measurement identities according to the signaling and procedures in this specification. The list of measurement objects may include NR measurement objects, CLI measurement objects, inter-RAT objects, and L2 U2N relay objects. Similarly, the list of reporting configurations includes NR, inter-RAT, and L2 U2N relay reporting configurations. Any measurement object can be linked to any reporting configuration of the same RAT type. Some reporting configurations may not be linked to measurement objects. Similarly, some measurement objects may not be linked to reporting configurations.

[0251] The measurement process differentiates the following types of cells: NR serving cells, listed cells, and detected cells. NR serving cells are the SpCell and one or more SCell. Listed cells are the cells listed within the measurement object. Detected cells are cells that are not listed in the measurement object but are detected by the UE at the SSB frequencies and subcarrier spacings indicated by the measurement object.

[0252] For NR measurement objects, the UE makes measurements and reports on the serving cell / serving relay UE (for L2 U2N remote UEs), the listed cells, and / or the detected cells. For inter-RAT measurement objects for E-UTRA, the UE makes measurements and reports on the listed cells and the detected cells, and for RSSI and channel occupancy measurements, the UE makes measurements and reports on the configured resources on the indicated frequencies. For inter-RAT measurement objects for UTRA-FDD, the UE makes measurements and reports on the listed cells. For CLI measurement objects, the UE makes measurements and reports on the configured measurement resources (i.e., SRS resources and / or CLI-RSSI resources). For L2 U2N relay objects, the UE makes measurements and reports on the serving NR cell and the discovered L2 U2N relay UEs.

[0253] RRC message structure

[0254] As described above, there is a large amount of AI / ML data that needs to be reported to the network. One solution to avoid frequent reporting as described above is to create a messaging structure that allows the UE to include multiple pieces of AI / ML data information in one RRC message (e.g., a measurement report message or an RRC message for reporting). The proposed message may include a sequence number or timing information for each piece of AI / ML data so that the receiver can sort them in chronological order. The number of AI / ML data that should be included in the report message can be configured by the RRC message (e.g., in the AI / ML configuration). The detailed message structure is described below.

[0255] RRC messages may contain one or more of the following information elements: MeasConfig, ReportConfigNR, MeasResults, CSI-MeasConfig, CSI-ReportConfig, and MeasurementReport. These provide configuration information for measurements. For AI / ML implementation, the message structure and parameters can be extended to configure AI / ML configuration: MeasConfigAIML, ReportConfigNRAIML, MeasResultsAIML, CSI-MeasConfigAIML, CSI-ReportConfigAIML, MeasurementReportAIML. For example, in the following text, when the network configures AI / ML configuration for the UE (or network entity), MeasConfig, ReportConfigNR, MeasResults, CSI-MeasConfig, CSI-ReportConfig, and MeasurementReport can be regarded as MeasConfigAIML, ReportConfigNRAIML, MeasResultsAIML, CSI-MeasConfigAIML, CSI-ReportConfigAIML, and MeasurementReportAIML.

[0256] Unless otherwise explicitly stated, when the proposed procedure mentions a field, it refers to the field included in VarMeasConfig, that is, only the measurement configuration procedure covers the direct UE actions related to the received measConfig.

[0257] Furthermore, consider some of these key information elements: The message structure may include a measurement configuration information element (MeasConfig) that specifies measurements to be performed by the UE. This information element may include fields covering intra-frequency, inter-frequency, and inter-RAT mobility as well as measurement gap configuration. For example, when the field "interFrequencyConfig-NoGap-r16" is set to true, the UE may be configured to perform SSB-based inter-frequency measurements without a measurement gap when the inter-frequency SSB is fully contained within the UE's active DL BWP. There may also be fields in the information element that are lists of measurement identities to be removed, added, and / or modified (e.g., MeasIdToAddModList and MeasIdToRemoveList). There may also be fields in the information element that are lists of measurement objects to be removed, added, and / or modified (e.g., MeasObjectToAddModList and MeasObjectToRemoveList). There may also be fields in the information element that are lists of measurement report configurations to be removed, added, and / or modified (e.g., ReportConfigToAddModList and ReportConfigToRemoveList). There may also be one or more thresholds, e.g., a threshold for controlling when the UE needs to perform measurements on non-serving cells. These fields may also be referred to as information elements.

[0258] The message structure may include a measurement configuration identity information element (MeasId) for identifying the measurement configuration, i.e., the link between the measurement object and the report configuration. The MeasID in the MeasConfig information element indicates the type or identifier of the individual measurement. When the network configures an AI / ML configuration for the UE (or a network entity), the measID in MeasConfigAIML may indicate the type of AI / ML data to be collected (e.g., can be used as an AI / ML data identifier to configure an AI / ML configuration for multiple types of AI / ML data) or the AI / ML model (e.g., can be used as an AI / ML model identifier to configure an AI / ML configuration for multiple AI / ML models) or the same information for the measurement.

[0259] The MeasConfig information element includes a field that is a list of measurement objects to be removed, added, and / or modified. The MeasObject information element within the MeasConfig information element includes details of the target frequency (e.g., time / frequency radio resources). The MeasObjectID can be used to indicate each measurement object MeasObject. When the network configures an AI / ML configuration for a UE (or a network entity), the measurement object information element measObject in the configuration information element MeasConfigAIML can indicate an AI / ML model (e.g., an AI / ML model identifier can be used to configure the AI / ML configuration for multiple AI / ML models) or the same information for measurement. Similarly, there may be an information element (MeasObjectIDAIML) that can be used to indicate each measurement object (MeasObjectAIML) in the AI / ML configuration. The MeasObject information element can also include a field (cellIndividualOffset) for indicating a separate offset applicable to a specific cell, and a field (physCellId) for the cell identity of the cells in the cell list.

[0260] The MeasurementReport message can include the measurement results (MeasResults) of the identified measurement configuration (measID). When an AI / ML configuration is configured, the MeasurementReportAIML can include the AI / ML data of measIDAIML. The MeasurementReport message is used for the indication of measurement results. The signaling radio bearer can be SRB1 and / or SRB3, and a data radio bearer DRB (if configured) can be used. The RLC-SAP can be AM, the logical channel can be DCCH, and the direction of the message is from the UE to the network.

[0261] The Measured Results Information Element (MeasResults) includes the detailed information to be reported for each identified measurement configuration (measID) (e.g., Physical Cell Identifier, cell list, frequency list, beam, beam list, RSRP, RSRQ, SINR, etc.). The Measured Results Information Element (MeasResults) can cover the measured results of intra-frequency, inter-frequency, and inter-RAT mobility as well as the measured results of NR side link communication / discovery. When the network configures AI / ML configuration for the UE (or network entity), MeasResultsAIML includes the AI / ML data set, AI / ML data type, AI / ML data volume, time information, sequence number, etc. for each identified measurement configuration, and the measurement configuration includes AI / ML (measIDAIML). MeasResultsAIML can include the same measurement information as the Measured Results Information Element (MeasResults).

[0262] For AI / ML data, sequence number or time information can be included to report multiple data of the same target (e.g., MeasObject, AI / ML data, or AI / ML model configured as described above) in chronological order (i.e., time series). For example, Option 1 is to introduce a sequence number (or time information) for each measurement result and include the sequence number (or time information). Option 2 is to introduce a list with a sequence number (or time information), i.e., each measurement quantity of the measurement result can include the sequence number (or time information). These two options are shown in the code in the detailed part below. In another embodiment, rules can be defined that do not require a sequence number (or time information) for each result (or quantity), e.g., the first result in the report message is the first result in the time series and the second result is the second result. The rule can also be defined in the opposite way, i.e., the first result in the report message is the latest result in the time series and the second result is the second latest result. The rule can be defined in various ways to report multiple data of the same target.

[0263] The field description of the Measured Results information element (MeasResults) includes some or all of the following fields: a field indicating the coarse location information reported by the UE (coarseLocationInfo), a field indicating the packet ratio exceeding the configured delay threshold (excessDelay), and a field for the measured results of the measured cells with reference signals indicated in the serving cell measurement object (measResultServingMOList). These measurement results may include the measurement results of the SpCell, the configured SCell, and the best neighboring cell within the measured cell with reference signals indicated on each serving cell measurement object. The measured results may include SFTD measurements (i.e., the timing difference between the PCell of the LTE network and the PSCell of the 5G NR network), such as measResultSFTD-EUTRA and measResultSFTD-NR. There may also be sidelink measurements, e.g., measResultsSL, sl-MeasResultsCandRelay, and sl-MeasResultsServingRelay.

[0264] The field description of the Measured Results information element (MeasResults) includes some or all of the following fields depending on the network type: the measured results of UTRA-FDD cells (measResultUTRA-FDD), the physical cell identifier of the E-UTRA cell or UTRA-FDD cell being reported (measResultUTRA-FDD or physcellID). There may also be a field description for the measured results related to the NR (New Radio) network.

[0265] Measurements can also be made in the RRC_IDLE and RRC_INACTIVE states. These measurements can be described by the information element (MeasResultIdleNR), in which case the information element covers the NR (New Radio) network. The field description may include fields indicating the following: NR carrier frequency (carrierFreq), idle / inactive measurement results of NR cells (measIdleResultNR), measured results of the serving cell (measResultServingCell), a list of idle / inactive measured results for the maximum number of reported best cells for a given NR carrier (measResultsPerCellListIdleNR), and / or beam-level measurements (resultsSSB-Indexes).

[0266] There is also an information element (ReportConfigNR) that includes details for reporting, such as which type of results should be reported, or periodic or time / frequency resources, or when to trigger reporting (i.e., events), etc. There is an identification information element (ReportConfigID) that can be used to indicate each reporting information element ReportConfigNR. When the network configures the AI / ML configuration for the UE (or network entity), the reporting information element can be labeled as ReportConfigNRAIML and can include details for reporting, such as which type of results should be reported, or periodic or time / frequency resources, or when to trigger reporting (i.e., events), etc. There is also an identification information element ReportConfigIDAIML that can be used to indicate each AI / ML reporting configuration ReportConfigNRAIML.

[0267] The reporting information element ReportConfigNR specifies the criteria for triggering one or more NR measurement reporting events: CHO (Conditional Handover), Conditional PSCell Addition (CPA), or Conditional PSCell Change (CPC) events, or L2 U2N relay measurement reporting events. For events labeled AN (where N equals 1, 2, etc.), the measurement reporting events and CHO, CPA, or CPC events are based on cell measurement results that can be derived based on SS / PBCH blocks or CSI-RS. The following table gives some examples of events that can trigger reporting:

[0268]

[0269]

[0270] Therefore, the reporting can be conditional, and there can be field descriptions of offsets or thresholds, including, for example, the offset value for triggering event A3 (a3 offset), and / or the thresholds (a4 threshold and a5 threshold) associated with the selected triggering quantity for each RS type for triggering event A4 or A5.

[0271] The message structure can also include an information element (ReportConfigToAddModList) that relates to a list of reporting configurations to be added or modified. The message structure can also include an information element (ServingCellConfig) for configuring (adding or modifying) the serving cell for the UE. The serving cell can be the SpCell or SCell of the MCG or SCG. Most of the parameters in this article are UE-specific, but some are cell-specific (e.g., in an additionally configured bandwidth part). Reconfiguration between PUCCH and PUCCH-less SCell is only supported using SCell release and addition.

[0272] The message structure may also include an information element (CSI-MeasConfig) that is used to configure CSI-RS (Channel State Information Reference Signal) for a serving cell that includes CSI-MeasConfig. On a serving cell that includes this information element (CSI-MeasConfig), the information element may also be used to configure channel state information reports to be transmitted on the PUCCH and channel state information reports on the PUSCH triggered by DCI received on the serving cell. This information element has various field descriptions, including a trigger state list (aperiodicTriggerStateList) for dynamically selecting one or more aperiodic and semi-persistent reporting configurations. There are also various lists for adding or modifying elements, including CSI-IM resources, CSI report settings, CSI resource settings, CSI-SB resources, and CSI-RS resources.

[0273] The message structure may also include an information element (CSI-ReportConfig) that is used to configure periodic or semi-persistent reports transmitted on the PUCCH in a cell that includes the information element CSI-ReportConfig. Alternatively, the information element may be used to configure semi-persistent or aperiodic reports transmitted on the PUSCH triggered by DCI received in a cell that includes CSI-ReportConfig. In this case, the cell on which the report is transmitted is determined by the received DCI.

[0274] The identifier of the measurement configuration (measID), the identifier of the measurement object (measObjectID), and the identifier of the reporting configuration (ReportConfigID) may be mapped in the identifier of the measurement configuration to be added or modified (MeasIdToAddMod). In this way, the measurement target, how to measure the target, how to report the results, which results should be reported, etc. can be defined. Similarly, the identifier of the AI / ML measurement configuration (measIDAIML), the identifier of the AI / ML measurement object (measObjectIDAIML), and the identifier of the AI / ML reporting configuration (ReportConfigIDAIML) may be mapped in the identifier of the AI / ML measurement configuration to be added or modified (MeasIdToAddModAIML). By mapping measIDAIML, measObjectIDAIML, and ReportConfigIDAIML in MeasIdToAddMod, the target data to be collected, how to measure the target (or collect the target), how to report the results, which results should be reported, etc. can be defined.

[0275] Split SRB handling

[0276] SRB4 is used for RRC messages including application layer measurement report information, and they all use the DCCH logical channel. SRB4 can only be configured by the network after AS security activation. The main purpose of SRB4 is for QoE (Quality of Experience) measurement. QoE is related to but different from Quality of Service (QoS), and QoS embodies the concept that can measure, improve, or even guarantee hardware and software characteristics. For SRB4, splitting the SRB can be useful for improving reliability.

[0277] As explained in the previous example, there can be RBs specifically configured for AI / ML functions, e.g., SRBx. This RB is used for RRC messages including AI / ML function or configuration related information, and they all use the DCCH logical channel. SRBx can only be configured by the network after AS security activation. The main purpose of SRBx is for AI / ML functions. For SRBx, splitting the SRB can be useful for improving reliability. The benefit of splitting the SRB is to reduce latency and increase the reliability of data transmission at the cost of UE implementation complexity. For SRBx, considering that SRBx is mainly used for AI / ML functions, it may be more preferable to reduce UE implementation complexity rather than the benefit of splitting the SRB. Therefore, it is best to limit the network configuration, i.e., not allow the network to configure SRBx with split SRB for the UE. When the network configures SRBx with split SRB for the UE, the UE regards it as an error situation or declares an error (e.g., configuration error or configuration failure).

[0278] General method

[0279] Figure 11It is a flowchart showing an overview of the steps of the method described in detail above. This method configures a user equipment (UE) to support artificial intelligence / machine learning (AI / ML) functions in a wireless communication network. In the first step S1100, a first radio resource control (RRC) connection is established from the user equipment to the network. The connection can be established via a first signaling radio bearer (e.g., SRB1), as described above. Then, at step S1105, at least one configuration message is sent from the network to the user equipment. At step S1110, the configuration message is received at the user equipment via the first signaling radio bearer. The configuration message includes bearer configuration information for a second signaling radio bearer to support AI / ML functions in the network. Data transfer related to the AI / ML function can be completed via the second radio bearer. In the next step S1115, the user equipment can be configured using the information in the configuration message. Although shown as a separate step S1120, establishing a second radio resource control (RRC) connection from the user equipment to the network via the second signaling radio bearer is part of the configuration of the user equipment. The second signaling radio bearer can be a standard signaling radio bearer (e.g., SRB2), a newly configured signaling radio bearer (e.g., SBR5), or other appropriate SRB number), or it can be a data radio bearer.

[0280] Additional configuration information can be received in the same configuration message or in a separate message. As described above, the user equipment can also be configured using some or all of the configuration information, measurement configuration information, prioritization configuration information, and security configuration information. After configuring the user equipment, at step S1125, the network confirms that the configuration has been made, for example, by receiving an acknowledgment message from the user equipment. Then, the network and the user equipment can be used together to apply AI / ML functions in the network. For example, a node within the network having an AI / ML model can receive AI / ML data from the user equipment and can use the received AI / ML data to train the AI / ML model. As another example, a node within the network having an AI / ML model can receive AI / ML data from the user equipment and can infer an output from the AI / ML model.

[0281] The user equipment may have been configured to measure the transmitted data before transmission, as described above. Thus, as Figure 11 shown, at step S1130, the user equipment measures data related to the AI / ML function. The second RRC connection enables the transmission of an RRC message including data related to the AI / ML function from the user equipment, as shown at step S1135. The data is received at the network, as shown at step S1140. The data can also be sent from the network to the user equipment via the second RRC connection.

[0282] General system

[0283] Figure 12 is a block diagram of a system that includes node 1200 in a network and user equipment 1250 connected to node 1200. It should be understood that the system may include multiple local devices, which may also be referred to as client devices or user devices or user equipment, and the system may include multiple nodes or cells and OAM as described above.

[0284] User equipment 1250 may be any of the following: a smart phone, a tablet computer, a laptop computer, a computer or computing device, a virtual assistant device, a vehicle, a drone, an autonomous vehicle, a robot or robotic device, a robotic assistant, an image capture system or device, an augmented reality system or device, a virtual reality system or device, a gaming system, an Internet of Things device or a smart consumer device (such as a smart refrigerator). It should be understood that this is a non-exhaustive and non-limiting list of example devices. Device 1250 includes standard components, for example, at least one processor 1252 coupled to a memory 1254. There may also be a microphone 1256 for capturing speech and a user interface 1258 for capturing other user input. There may be other sensors 1260 for capturing other data (including the above-mentioned AI / ML data). There may also be at least one transmitter 1270 and at least one receiver 1272. It should be understood that there may be other standard components not shown for simplicity. Device 1250 may include one or more modules for collecting user data 1264 stored in a storage device 1262, and such a storage device may be an encrypted storage device. By way of example only, the modules may include microphone 1256, user interface 1258, and sensors 1260.

[0285] At least one processor 1252 may include one or more of the following: a microprocessor, a microcontroller, and an integrated circuit. For example, memory 1254 may include volatile memory for use as temporary memory, such as random access memory (RAM), and / or non-volatile memory for storing data, programs, or instructions, such as flash memory, read-only memory (ROM), or electrically erasable programmable ROM (EEPROM).

[0286] There may be a training module 1290 on the user equipment that trains an ML model 1206 to generate a local ML model 1280 stored on the user equipment. The local model parameters 1268 of the local ML model may be stored in the storage device 1262. As an alternative, user equipment 1250 may not train the ML model but only perform inference using the stored local ML model 1280.

[0287] The system can incorporate federated learning, and thus node 1200 (or OAM) is arranged to perform any pre-training steps required to generate the initially trained ML model 1206. For example, node 1200 receives reference training data (input x and output / label y) from database 1202. Node 1200 includes a training module 1204 that receives the reference data from database 1202 as input and outputs basic or full-precision model parameters (i.e., the set of weights or parameters that have been learned during the training process). As an alternative to federated learning where the training module 1290 is on the user device, when using the training module 1204 on the node, local data can be received from user device 1250 to train a local ML model for user device 1250 in a manner similar to that described regarding training on the user device. Then the personalized ML model generated on the node is sent to the user device for storage as the local ML model 1280. This can be referred to as distributed learning.

[0288] It should be understood that there are other components within node 1200 that are not shown for ease of reference. Node 1200 can also use the trained ML model 1206 for inference, for example, to infer an output based on an input received from device 1250 (such as an input like a measurement from a sensor, etc.).

[0289] Figure 13 is a block diagram showing a terminal (or user equipment (UE)) according to an embodiment disclosed herein.

[0290] As Figure 13 shown, a terminal according to an embodiment can include a transceiver 1310, a memory 1320, and a processor (or controller) 1330. The transceiver 1310, memory 1320, and processor (or controller) 1330 of the terminal can operate according to the communication method of the terminal described above. However, the components of the terminal are not limited thereto. For example, the terminal can include more or fewer components than those Figure 13 described. Additionally, the processor (or controller) 1330, transceiver 1310, and memory 1320 can be implemented as a single chip. Additionally, the processor (or controller) 1330 can include at least one processor. Furthermore, Figure 13 the UE of Figures 1 to 12 corresponds to the UE 18, UE 218, NR UE, UE 318, UE 718, UE 818, UE 918, UE 1018, UE 1118, or device 1350.

[0291] The transceiver 1310 collectively refers to the terminal station receiver and the terminal transmitter, and can transmit signals to or receive signals from the base station or another terminal. The signals transmitted to or received from the terminal may include control information and data. The transceiver 1310 may include an RF transmitter for up-converting and amplifying the transmitted signal, and an RF receiver for low-noise amplifying and down-converting the received signal. However, this is only an example of the transceiver 1310, and the components of the transceiver 1310 are not limited to the RF transmitter and the RF receiver.

[0292] In addition, the transceiver 1310 may receive signals through a wireless channel and output them to the processor 1330, and transmit the signals output from the processor (or controller) 1330 through the wireless channel.

[0293] The memory 1320 may store programs and data required for the operation of the terminal. In addition, the memory 1320 may store control information or data included in the signals obtained by the terminal. The memory 1320 may be a storage medium such as a read-only memory (ROM), a random access memory (RAM), a hard disk, a CD-ROM, and a DVD, or a combination of storage media.

[0294] The processor (or controller) 1330 may control a series of processes so that the terminal operates as described above. For example, the processor (or controller) 1330 may receive data signals and / or control signals, and the controller 1330 may determine the result of receiving the signals transmitted by the base station and / or other terminals.

[0295] Figure 14 is a block diagram showing a base station (BS) according to an embodiment disclosed herein.

[0296] As Figure 14 shown, the base station of the present disclosure may include a transceiver 1410, a memory 1420, and a processor (or controller) 1430. The transceiver 1410, the memory 1420, and the processor (or controller) 1430 of the base station may operate according to the communication method of the base station described above. However, the components of the base station are not limited thereto. For example, the base station may include more or fewer components than those Figure 14 described. In addition, the processor (or controller) 1430, the transceiver 1410, and the memory 1420 may be implemented as a single chip. In addition, the processor (or controller) 1430 may include at least one processor. In addition, Figure 14 the base station of Figures 1 to 12ENBs 16a to 16d, LTE eNB 216, eNB 316, NR gNB 326, gNB 1, NG-RAN Node 1 (926), NG-RAN Node 2 (928), NG-RAN Node 1 (1026), NG-RAN Node 2 (1026), MCG 1120, SCG 1122 or Node 1200.

[0297] The transceiver 1410 collectively refers to a base station receiver and a base station transmitter, and can transmit signals to / from a terminal, another base station, and / or a core network function (or entity). Signals transmitted to / from a base station can include control information and data. The transceiver 1410 can include an RF transmitter for upconverting and amplifying transmitted signals, and an RF receiver for low-noise amplifying and downconverting received signals. However, this is only an example of the transceiver 1410, and the components of the transceiver 1410 are not limited to the RF transmitter and the RF receiver.

[0298] In addition, the transceiver 1410 can receive signals through a wireless channel and output them to the processor 1430, and transmit signals output from the processor (or controller) 1430 through a wireless channel.

[0299] The memory 1420 can store programs and data required for the operation of the base station. In addition, the memory 1420 can store control information or data included in signals obtained by the base station. The memory 1420 can be a storage medium such as ROM, RAM, hard disk, CD-ROM, and DVD, or a combination of storage media.

[0300] The processor (or controller) 1430 can control a series of processes so that the base station operates as described above. For example, the processor (or controller) 1430 can receive data signals and / or control signals, and the processor (or controller) 1430 can determine the result of receiving signals transmitted by the terminal and / or the core network function.

[0301] Detailed procedures for RRC connection control

[0302] UE's reception of RRCSetup

[0303] Figure 8 Outlines how an RRC connection can be set up. The following pseudocode gives details of the steps performed by the UE when receiving RRCSetup:

[0304] 1> If an RRC Setup is received in response to an RRCReestablishmentRequest; or 1> If an RRC Setup is received in response to an RRCResumeRequest or an RRCResumeRequest1:

[0305] 2> If sdt-MAC-PHY-CG-Config is configured:

[0306] 3> Instruct the MAC entity to stop the cg-SDT-TimeAlignmentTimer (if it is running);

[0307] 3> Instruct the MAC entity to start the timeAlignmentTimer associated with the PTAG (if it is not running);

[0308] 2> If srs-PosRRC-InactiveConfig is configured:

[0309] 3> Instruct the MAC entity to stop the inactivePosSRS-TimeAlignmentTimer (if it is running);

[0310] 2> Discard any stored UE inactive AS context and suspendConfig;

[0311] 2> Discard any current AS security context, including K RRCenc key, K RRCint key, K UPint key and K UPenc key;

[0312] 2> Release the radio resources of all established RBs except SRB0 and broadcast MRBs, including releasing the RLC entity, the associated PDCP entity, and SDAP;

[0313] 2> Release the RRC configuration except for the default L1 parameter values, the default MAC cell group configuration, the CCCH configuration, and the broadcast MRBs;

[0314] 2> Indicate to the upper layer the fallback of the RRC connection;

[0315] 2> Discard any application layer measurement reports that have not been transmitted;

[0316] 2> Notify the upper layer to release all application layer measurement configurations;

[0317] UE's reception of RRCReconfiguration

[0318] Figure 8It also outlines how RRC reconfiguration can be performed. The following pseudocode gives details of the steps performed by the UE upon receiving an RRCReconfiguration message. These steps can also be performed based on conditional reconfiguration (CHO, CPA or CPC). As explained above, when an RRC message (RRCReconfiguration or RRCResume) includes configuration information, the UE can apply the AI / ML configuration. If the UE receives an AI / ML configuration in otherConfig in RRCReconfguration, the UE can include AI / ML data in subsequent RRC messages to enable the network to utilize this information for network management. When the UE receives a handover command message (i.e., an RRCReconfiguration including reconfigurationWithSync) from the network (or gNB or source gNB), the UE can send a handover complete message (i.e., RRCReconfigurationComplete) to the network (or gNB or target gNB). The UE can include an indication of the availability of the AI / ML data to be reported or the AI / ML data itself in the response message (RRCReconfigurationComplete) to inform the network. In addition to this, the UE can support retransmission of RRC messages including AI / ML data, i.e., the UE can retransmit before its successful delivery has been acknowledged (e.g., acknowledged from the source gNB).

[0319] 1> If the RRCReconfiguration (or RRCResume) message includes AIMLConfig (i.e., AI / ML configuration):

[0320] 2> Perform the AI / ML configuration process;

[0321] 1> If the received otherConfig includes AI / ML configuration:

[0322] 2> If the AI / ML configuration is set to setup, include the available AI / ML data for any subsequent measurement reports or any subsequent RLF reports and SCGFailureInformation; 1> Set the content of the RRCReconfigurationComplete message as follows:

[0323] 2> If the RRCReconfiguration includes reconfigurationWithSync in spCellConfig of MCG:

[0324] 3> If the UE has recorded measurements available for NR and if the RPLMN is included in the plmn-IdentityList stored in VarLogMeasReport:

[0325] 4> Include logMeasAvailable in the RRCReconfigurationComplete message;

[0326] 4> If the Bluetooth measurement results are included in the measurements recorded by the UE that can be used for NR:

[0327] 5> Include logMeasAvailableBT in the RRCReconfigurationComplete message;

[0328] 4> If the WLAN measurement results are included in the measurements recorded by the UE that can be used for NR:

[0329] 5> Include logMeasAvailableWLAN in the RRCReconfigurationComplete message;

[0330] 2> If the RRCReconfiguration includes reconfigurationWithSync in the spCellConfig of the MCG:

[0331] 3> If the UE has AI / ML data available for NR or if the RPLMN is included in the plmn-IdentityList stored in VarLogMeasReport:

[0332] 4> Include AIMLDataAvailable (an indication of the availability of AI / ML data) or the AI / ML data in the RRCReconfigurationComplete message; 1> If reconfigurationWithSync is included in the spCellConfig of the MCG or SCG and when the MAC of the NR cell group has successfully completed the above-triggered random access procedure; or

[0333] 2> Stop the timer T304 of this cell group (if it is running);

[0334] 2> Stop the timer T310 of the source SpCell (if it is running);

[0335] 2> Apply the parts of the CSI report configuration, scheduling request configuration, and sounding RS configuration that do not require the UE to know the SFN of the corresponding target SpCell (if any);

[0336] 2> After obtaining the SFN of the target SpCell, apply the parts in measurement and radio resource configuration that require the UE to know the corresponding target SpCell SFN (e.g., measurement gap, periodic CQI report, scheduling request configuration, sounding RS configuration) (if any);

[0337] 2> If reconfigurationWithSync is included in the masterCellGroup:

[0338] 3> If application layer measurement is configured and if an application layer measurement report container has been received from the upper layer and the lower layer has not yet confirmed the successful transmission of the message or at least one fragment of the message:

[0339] 4> Resubmit the MeasurementReportAppLayer message or all fragments of the MeasurementReportAppLayer message to the lower layer for transmission via SRB4;

[0340] 3> If AI / ML functions (e.g., AI / ML configuration) are configured and if an RRC message including a report for AI / ML data has been generated (or transmitted) and the lower layer has not yet confirmed the successful transmission of the message or at least one fragment of the message:

[0341] 4> Resubmit the RRC message including a report for AI / ML data or all fragments of the RRC message to the lower layer for transmission via the configured RB (SRB1 or SRB2 or SRBx or DRB);

[0342] 2> If reconfigurationWithSync is included in the masterCellGroup and the target cell provides SIB21:

[0343] 3> If the UE initiated the transmission of the MBSInterestIndication message during the last 1 second before receiving this RRCReconfiguration message; or

[0344] 3> If the RRCReconfiguration message is applied due to conditional reconfiguration execution and the UE has initiated the transmission of the MBSInterestIndication message after receiving this RRCReconfiguration message:

[0345] 4> Initiate the transmission of the MBSInterestIndication message;

[0346] 2> The process ends.

[0347] AI / ML Configuration

[0348] Figure 9 and Figure 10 are examples of how AI / ML data can be used in a network including a UE. The following pseudocode gives details of the steps performed by the UE when configuring an AI / ML configuration: 1> If measConfigToReleaseList (e.g., a list of configurations to be released) is included in the AIMLConfig (i.e., AI / ML configuration) within RRCReconfiguration or RRCResume:

[0349] 2> For each measConfigId (e.g., a target identifier for AI / ML data collection) value included in measConfigToReleaseList:

[0350] 3> Discard any AI / ML data for measConfigId;

[0351] 3> Consider itself not configured to send AI / ML data for measConfigId.

[0352] 1> If measConfigToAddModList (e.g., a list of configurations to be added and modified) is included in the AIMLConfig within RRCReconfiguration or RRCResume:

[0353] 2> For each measConfigId value included in measConfigToAddModList:

[0354] 3> Consider itself configured to send AI / ML data for measConfigrId;

[0355] 3> If pauseReporting (e.g., an indication of whether to stop (or deactivate) or start (or activate) AI / ML data reporting) is set to true:

[0356] 4> If at least one (but not all) of the segments of the segmented RRC message containing AI / ML data associated with measConfigId has been submitted to the lower layer for transmission:

[0357] 5> Submit the remaining segments of the RRC message to the lower layer for transmission;

[0358] 4> Pause submitting RRC messages including AI / ML data to the lower layer for the AI / ML configuration associated with measConfigId;

[0359] 4> Store any previously or subsequently received AI / ML data containers associated with the measConfigId for which no fragment or complete message has been submitted to the lower layer for transmission.

[0360] 3> Otherwise, if pauseReporting is set to false and if transmission of AI / ML data reporting has been previously paused for the AI / ML configuration associated with measConfigId:

[0361] 4> Submit an RRC message including the stored AI / ML data (if any) to the lower layer for the AI / ML configuration associated with measConfigId.

[0362] 4> Resume submitting RRC messages including AI / ML data to the lower layer for the AI / ML configuration associated with measConfigId.

[0363] UE Actions on Entering RRC_IDLE

[0364] If a UE in the RRC idle state does not support the AI / ML function, when the UE enters the RRC idle state from the RRC connected (or RRC inactive) mode, the process of discarding AI / ML data or releasing the AI / ML configuration is required. The following pseudocode gives the UE actions on entering RRC_IDLE:

[0365] 1> Reset MAC;

[0366] 1> Set the variable pendingRNA-Update to false (if it was set to true).

[0367] 1> If the UE is leaving RRC_INACTIVE:

[0368] 2> If entering RRC_IDLE is not triggered by receiving an RRCRelease message:

[0369] 3> Discard the cell reselection priority information provided by cellReselectionPriorities if it has been stored.

[0370] 3> Stop timer T320 (if it is running).

[0371] 1> Stop all running timers except T302, T320, T325, T330, T331, and T400.

[0372] 1> Discard the UE inactive AS context (if any).

[0373] 1> Release suspendConfig (if configured);

[0374] 1> Remove all entries (if any) from MCG and SCGVarConditionalReconfig;

[0375] 1> For each measId (e.g., the target identifier for AI / ML data collection), if the associated reportConfig (e.g., the configuration for AI / ML data reporting) has a reportType set to condTriggerConfig:

[0376] 2> For the associated reportConfigId:

[0377] 3> Remove the entry with the matching reportConfigId from the reportConfigList within VarMeasConfig;

[0378] 2> If the associated measObjectId is only associated with a reportConfig whose reportType is set to condTriggerConfig:

[0379] 3> Remove the entry with the matching measObjectId from the measObjectList within VarMeasConfig;

[0380] 2> Remove the entry with the matching measId from the measIdList within VarMeasConfig; 1> Discard K gNB Key, S-K gNB Key, S-K eNB Key, K RRCenc Key, K RRCint Key, K UPint Key and K UPenc Key (if any);

[0381] 1> Release all radio resources, including releasing RLC entities, BAP entities, MAC configurations of all established RBs (except broadcast MRBs) and associated PDCP entities, BH RLC channels, Uu relay RLC channels, PC5 relay RLC channels, and SRAP entities;

[0382] 1> Indicate to the upper layer the release of the RRC connection and the reason for the release;

[0383] 1> Notify the upper layer to release all application layer measurement configurations;

[0384] 1> Discard any application layer measurement reports that have not been submitted to the lower layer for transmission;

[0385] 1> Notify the upper layer (e.g., application layer or NAS layer) of the release of the AI / ML configuration or release the AI / ML configuration;

[0386] 1> Discard any AI / ML data (e.g., reports) that have not been submitted to the lower layer for transmission;

[0387] 1> Discard any fragments of the stored segmented RRC messages (e.g., fragments of AI / ML data);

[0388] 1> Unless when the UE enters RRC_IDLE triggered by inter-RAT cell reselection when the UE is in RRC_INACTIVE or RRC_IDLE, or when selecting an inter-RAT cell while T311 is running, or when selecting an E-UTRA cell for EPS fallback for IMS voice:

[0389] 2> Enter RRC_IDLE and perform cell selection;

[0390] UE's quantity configuration

[0391] The following pseudocode gives the UE actions for quantity configuration. The UE shall: 1> For each RAT for which the received quantityConfig includes parameters:

[0392] 2> Set the corresponding parameters in quantityConfig within VarMeasConfig to the values of the received quantityConfig parameters;

[0393] 1> For each measId (or list of AIML identifiers) included in the measIdList (or list of AIML identifiers) within VarMeasConfig:

[0394] 2> Remove the measurement report entry for that measId (or AIML identifier) from VarMeasReportList

[0395] (or list of AIML reports) if it is included;

[0396] 2> Stop the periodic reporting timer or timer T321 or timer T322, whichever is running, and reset the associated information (e.g., timeToTrigger) for that measId (or AIML identifier).

[0397] Details of various information elements

[0398] As described above, there are various information elements in the RRC message structure. The codes for these elements are provided below:

[0399] MeasConfig information element

[0400]

[0401]

[0402]

[0403]

[0404] MeasId information element

[0405]

[0406] MeasIdToAddModList information element

[0407]

[0408] MeasObjectNR information element

[0409]

[0410]

[0411]

[0412]

[0413]

[0414]

[0415] MeasObjectToAddModList information element

[0416]

[0417] MeasurementReport message

[0418]

[0419]

[0420] MeasResults information element

[0421]

[0422]

[0423]

[0424]

[0425]

[0426]

[0427]

[0428]

[0429]

[0430]

[0431]

[0432]

[0433]

[0434]

[0435]

[0436] MeasResultIdleNR information element

[0437]

[0438]

[0439]

[0440] ReportConfigNR information element

[0441]

[0442]

[0443]

[0444]

[0445]

[0446]

[0447]

[0448]

[0449]

[0450]

[0451]

[0452] ReportConfigToAddModList information element

[0453]

[0454] ServingCellConfig information element

[0455]

[0456]

[0457]

[0458]

[0459]

[0460] CSI-MeasConfig information element

[0461]

[0462]

[0463]

[0464]

[0465] CSI-ReportConfig information element

[0466]

[0467]

[0468]

[0469]

[0470]

[0471] Those skilled in the art will understand that, although the best mode contemplated for carrying out the technology and, where appropriate, other modes have been described previously, the technology should not be limited to the specific configurations and methods disclosed in that description of the preferred embodiments. Those skilled in the art will recognize that the technology has a wide range of applications and that embodiments can be modified in a wide variety of ways without departing from any inventive concept as defined in the appended claims.

[0472] Various combinations of alternative features have been described herein, and it will be understood that the features described can be combined in any suitable combination. In particular, the features of any one exemplary embodiment can be combined appropriately with the features of any other embodiment, unless such combination is mutually exclusive. Throughout this specification, the term "comprising" or "comprises" means including the specified components but not excluding the presence of other components.

[0473] Attention is directed to all documents and literature related to this application that are filed simultaneously with or before this specification and that are publicly available in conjunction with this specification, and the contents of all such documents and literature are incorporated herein by reference. All features disclosed in this specification (including any appended claims, abstract, and drawings) and / or all steps of any method or process so disclosed can be combined in any combination, except combinations in which at least some of such features and / or steps are mutually exclusive.

[0474] Unless otherwise expressly stated, each feature disclosed in this specification (including any appended claims, abstract, and drawings) can be replaced by an alternative feature serving the same, equivalent, or similar purpose. Thus, unless otherwise expressly stated, each feature disclosed is only one example of a general series of equivalent or similar features. The invention is not limited to the details of the foregoing embodiments. The invention extends to any novel feature or any novel combination of features disclosed in this specification (including any appended claims, abstract, and drawings), or to any novel step or any novel combination of steps of any method or process so disclosed.

Claims

1. A method performed by a user equipment (UE) for supporting artificial intelligence / machine learning (AI / ML) functions in a wireless communication system, the method comprising: Establishing a first radio resource control (RRC) connection between the UE and a base station (BS) by using a first signaling radio bearer (SRB); Receiving, via the first SRB, a configuration message from the BS that includes bearer configuration information; Establishing a second radio resource control (RRC) connection between the UE and the BS by using a second SRB; and Receiving, based on the second RRC connection, at least one RRC message from the BS that includes data related to the AI / ML function, wherein the second SRB is configured based on the bearer configuration information.

2. The method according to claim 1, wherein The AI / ML function is provided by at least one AI / ML model, and the method further comprises: Receiving measurement configuration information from the BS, the measurement configuration information being for configuring the UE to obtain and report measurement data for the at least one AI / ML model, wherein the measurement configuration information includes a reporting configuration that defines at least one reporting criterion for triggering the UE to transmit a report to the BS.

3. The method according to claim 1, the method further comprising: Receiving priority ranking configuration information from the BS, the priority ranking configuration information being for defining a lower priority for the at least one RRC message that includes data related to the AI / ML function than other messages between the UE and the BS.

4. The method according to claim 3, the method further comprising: Activating access stratum security before establishing the second RRC connection; Receiving security configuration information from the BS, the security configuration information being for applying at least one of integrity protection or encryption to messages between the UE and the BS; And Applying both integrity protection and encryption to the at least one RRC message that includes data related to the AI / ML function before transmitting the at least one RRC message.

5. A user equipment (UE) for supporting artificial intelligence / machine learning (AI / ML) functions in a wireless communication system, the UE comprising: At least one transceiver; A controller connected to the at least one transceiver and configured to: Establish a first radio resource control (RRC) connection between the UE and a base station (BS) by using a first signaling radio bearer (SRB); Receiving, via the first SRB, a configuration message from the BS that includes bearer configuration information; Establishing a second radio resource control (RRC) connection between the UE and the BS by using a second SRB; and Receiving, based on the second RRC connection, at least one RRC message from the BS that includes data related to the AI / ML function, wherein the second SRB is configured based on the bearer configuration information.

6. The UE according to claim 5, wherein, The AI / ML function is provided by at least one AI / ML model, and the controller is further configured to: Receiving measurement configuration information from the BS, the measurement configuration information being for configuring the UE to obtain and report measurement data for the at least one AI / ML model, wherein the measurement configuration information includes a reporting configuration defining at least one reporting criterion for triggering the UE to transmit a report to the BS.

7. The UE according to claim 5, wherein The controller is further configured to receive prioritization configuration information from the BS, the prioritization configuration information being for defining a lower priority for at least one RRC message including data related to the AI / ML function than other messages between the UE and the BS.

8. The UE according to claim 7, wherein, The controller is further configured to: Activate access stratum security before establishing the second RRC connection; Receive security configuration information from the BS, the security configuration information being for applying at least one of integrity protection or encryption to messages between the UE and the BS; and Apply both integrity protection and encryption to at least one RRC message including data related to the AI / ML function before transmitting the at least one RRC message.

9. A method performed by a base station BS for supporting artificial intelligence / machine learning AI / ML functions in a wireless communication system, the method comprising: Establishing a first radio resource control RRC connection between a user equipment UE and the BS by using a first signaling radio bearer SRB; Transmitting a configuration message including bearer configuration information to the UE via the first SRB; Establishing a second radio resource control RRC connection between the UE and the BS by using a second SRB; and Transmitting at least one RRC message including data related to the AI / ML function to the UE based on the second RRC connection, wherein the second SRB is configured based on the bearer configuration information.

10. The method according to claim 9, wherein, The AI / ML function is provided by at least one AI / ML model, and the method further comprises: Transmitting measurement configuration information to the UE, the measurement configuration information being for configuring the UE to obtain and report measurement data for the at least one AI / ML model, wherein the measurement configuration information includes a reporting configuration defining at least one reporting criterion for triggering the UE to transmit a report to the BS.

11. The method according to claim 9, the method further comprising: Transmitting prioritization configuration information to the UE, the prioritization configuration information being for defining a lower priority for at least one RRC message including data related to the AI / ML function than other messages between the UE and the BS.

12. A base station BS for supporting artificial intelligence / machine learning AI / ML functions in a wireless communication system, the BS comprising: At least one transceiver; A controller connected to the at least one transceiver and configured to: Establish a first radio resource control RRC connection between a user equipment UE and the BS by using a first signaling radio bearer SRB; Transmit a configuration message including bearer configuration information to the UE via the first SRB; Establishing a second Radio Resource Control (RRC) connection between the UE and the BS by using a second SRB; and Transmitting at least one RRC message including data related to the AI / ML function to the UE based on the second RRC connection, wherein the second SRB is configured based on the bearer configuration information.

13. The BS according to claim 12, wherein The AI / ML function is provided by at least one AI / ML model, and the controller is further configured to: Transmit measurement configuration information to the UE, the measurement configuration information being used to configure the UE to obtain and report measurement data for the at least one AI / ML model, wherein the measurement configuration information includes a reporting configuration defining at least one reporting criterion for triggering the UE to transmit a report to the BS.

14. The BS according to claim 12, wherein The controller is further configured to transmit priority ranking configuration information to the UE, the priority ranking configuration information being used to define a lower priority for the at least one RRC message including data related to the AI / ML function than other messages between the UE and the BS.

15. The BS according to claim 14, wherein The controller is further configured to: Activate access stratum security before establishing the second RRC connection; Transmit security configuration information to the UE, the security configuration information being used to apply at least one of integrity protection or encryption to messages between the UE and the BS; and Apply both integrity protection and encryption to the at least one RRC message including data related to the AI / ML function before transmitting the at least one RRC message.

Images