Handling AI / ML for communication links between user equipment and one or more network entities of a wireless communication network

Abstract

A user equipment, UE, for a wireless communication network is configured or preconfigured with at least one artificial intelligence / machine learning model, AI / ML model, or at least one AI / ML function for performing one or more tasks associated with data transmission on a communication link between the UE and one or more network entities of the wireless communication network. The UE is to monitor the communication link for one or more specific events, and the UE is to trigger one or more actions regarding usage of the one or more AI / ML models or AI / ML functions in response to detecting one or more of the specific events on the communication link, or in response to specific signaling from a network entity of the wireless communication network.

Description

[0001] manual

[0002] This invention relates to the field of wireless communication systems or wireless communication networks, and more specifically, to the use of at least one artificial intelligence / machine learning model (AI / ML model) or at least one AI / ML function in a wireless communication system to perform one or more tasks. Embodiments of the invention relate to improvements and enhancements in the processing of AI / ML models or AI / ML functions associated with performing one or more tasks on a communication link between a UE and one or more network entities of the wireless communication network.

[0003] Figure 1 is a schematic diagram of an example of a terrestrial wireless network 100. As shown in Figure 1(A), the terrestrial wireless network 100 includes a core network (CN) 102 and one or more radio access networks RAN1, RAN2, ... RAN N Figure 1(B) shows the Radio Access Network (RAN). n The example diagram may include one or more base stations gNB1 to gNB5, each base station serving a specific area around the base station, schematically represented by individual cells 1061 to 1065. The base stations are configured to serve users within the cells. One or more base stations may serve users in licensed and / or unlicensed frequency bands. The term base station (BS) refers to a gNB in ​​a 5G network, an eNB in ​​UMTS / LTE / LTE-A / LTE-A Pro, or a BS-only in other mobile communication standards. A BS may also include Integrated Access and Backhaul (IAB) nodes, such as IAB donors and / or IAB nodes, which include a central unit (CU) and a distributed unit (DU), and / or contain IAB-MTs, which include IAB mobile terminals (MTs). The term base station may also refer to an access point (AP) in any WiFi standard, such as an access point belonging to the IEEE 802.11 family. Users may be static or mobile devices. The wireless communication system may also be accessed by mobile or static IoT devices connected to a base station or a user. Mobile or stationary devices can include physical equipment, ground vehicles (such as robots or cars), air vehicles (such as manned or unmanned aerial vehicles (UAVs, the latter also known as drones), buildings, and other items or devices in which electronics, software, sensors, actuators, etc., are embedded, and possess network connectivity, enabling these devices to collect and interact with data across existing network infrastructure. Figure 1(B) shows an example diagram of five cells; however, RAN n It can include more or fewer such cells, and RAN nIt may also include only a single base station. Figure 1(B) shows two users, UE1 and UE2 (also referred to as user equipment), located in cell 1062 and served by base station gNB2; another user, UE3, is shown in cell 1064 served by base station gNB4. Arrows 1081, 1082, and 1083 schematically represent uplink / downlink connections for transmitting data from users UE1, UE2, and UE3 to base stations gNB2 and gNB4, or for transmitting data from base stations gNB2 and gNB4 to users UE1, UE2, and UE3. This can be implemented on licensed or unlicensed frequency bands. In addition, Figure 1(B) also shows two additional devices, 1101 and 1102, in cell 1064, such as IoT devices, which can be static or mobile devices. Device 1101 accesses the wireless communication system through base station gNB4 to receive and transmit data, as schematically shown by arrow 1121; device 1102 accesses the wireless communication system through user UE3, as schematically shown by arrow 1122. Each base station gNB1 to gNB5 can be connected to the core network 102, for example, via the S1 interface through its respective backhaul links 1141 to 1145, as schematically indicated by arrows pointing to "core" in Figure 1(B). The core network 102 can be connected to one or more external networks, which can be the Internet, or private networks such as intranets or any other type of campus network, such as private WiFi communication systems or 4G / 5G mobile communication systems. Furthermore, some or all of the base stations gNB1 to gNB5 can be interconnected, for example, via the S1 interface, X2 interface, or XN interface under the New Radio (NR) architecture, through their respective backhaul links 1161 to 1165, as schematically indicated by arrows pointing to "gNB" in Figure 1(B). The sidelink channel allows direct communication between UEs, also known as device-to-device (D2D) communication. The sidelink interface in 3GPP is named PC5. Note that the term User Equipment (UE) or User Terminal can also refer to a Station (STA) in any WiFi standard, such as a station belonging to the IEEE 802.11 series.

[0004] For data transmission, a physical resource grid can be used. A physical resource grid can include sets of resource elements onto which various physical channels and physical signals are mapped. For example, physical channels can include a Physical Downlink Shared Channel (PDSCH), a Physical Uplink Shared Channel (PUSCH), and a Physical Sidelink Shared Channel (PSSCH), which carry user-specific data, also known as downlink, uplink, and sidelink payload data; a Physical Broadcast Channel (PBCH) and a Physical Sidelink Broadcast Channel (PSBCH), which, for example, carry a Master Information Block (MIB) and one or more System Information Blocks (SIBs) and one or more Sidelink Information Blocks (SLIBs) (if supported); a Physical Downlink Control Channel (PDCCH, GC-PDCCH), a Physical Uplink Control Channel (PUCCH), and a Sidelink Control Channel (PSSCH), which, for example, carry Downlink Control Information (DCI), Uplink Control Information (UCI), and Sidelink Control Information (SCI); and a Physical Sidelink Feedback Channel (PSFCH), which carries PC5 feedback responses. The sidelink interface can support two-stage SCI, which refers to a first control area containing a portion of the SCI (also known as the first-stage SCI) and, optionally, a second control area containing a second portion of the control information (also known as the second-stage SCI).

[0005] For the uplink, physical channels may also include physical random access channels (PRACH or RACH), which the UE uses to access the network once synchronized and has acquired the MIB and SIB. Physical signals may include reference signals or symbols (RS), synchronization signals, etc. The resource grid may include frames or radio frames that have a specific duration in the time domain and a given bandwidth in the frequency domain. The frame may have a specific number of subframes of a predefined length (e.g., 1 ms). Each subframe may include one or more time slots of 12 or 14 OFDM symbols depending on the cyclic prefix (CP) length. Frames may also have fewer OFDM symbols, for example, when utilizing shortened transmission time intervals (sTTI) or mini-slot / non-slot-based frame structures that include only a few OFDM symbols.

[0006] Wireless communication systems can be any single-carrier or multi-carrier system using frequency division multiplexing, such as orthogonal frequency division multiplexing (OFDM) systems, or orthogonal frequency division multiple access (OFDMA) systems, or any other signal based on inverse fast Fourier transform (IFFT), with or without a configured cyclic prefix (CP), such as discrete Fourier transform extended OFDM (DFT-s-OFDM). Other waveforms can be used, such as non-orthogonal waveforms for multiple access, such as filter bank multicarrier (FBMC), generalized frequency division multiplexing (GFDM), or universal filtered multicarrier (UFMC). Wireless communication systems can operate, for example, according to 3GPP's LTE, LTE-Advanced, LTE-Advanced Pro, or 5G, or 5G-Advanced, or 6G, or 3GPP's NR (New Radio), or within LTE-U (LTE unlicensed) or NR-U (NR unlicensed), as specified in the LTE and NR specifications.

[0007] The wireless network or communication system depicted in Figure 1 can be a heterogeneous network with different coverage networks, such as a macrocell network, each macrocell including macro base stations, such as base stations gNB1 to gNB5, and a small cell base station network (not shown in Figure 1), such as femtocells or picocells. In addition to the terrestrial wireless networks described above, there are also non-terrestrial wireless communication networks (NTNs), which include, for example, spaceborne transceivers (such as satellites) and / or airborne transceivers (such as unmanned aerial systems). Non-terrestrial wireless communication networks or systems can operate in a similar manner to the terrestrial systems described above with reference to Figure 1, for example, according to LTE-advanced Pro or 5G or 5G-advanced or NR (New Radio) or a possible future 6G radio system.

[0008] In mobile communication networks, such as the LTE or 5G / NR networks described above with reference to Figure 1, UEs can communicate directly with each other via one or more sidelink (SL) channels (e.g., using PC5 / PC3 interfaces or Wi-Fi Direct). UEs communicating directly with each other via sidelinks can include vehicles communicating directly with other vehicles (V2V communication), vehicles communicating with other entities in the wireless communication network (V2X communication), such as roadside units (RSUs), roadside entities (e.g., traffic lights, traffic signs), or pedestrians. RSUs may have BS or UE functions depending on the specific network configuration. Other UEs may not be vehicle-related UEs and may include any of the aforementioned devices. Such devices can also communicate directly with each other using SL channels (D2D communication).

[0009] When considering two UEs that communicate directly with each other via a sidelink, the two UEs can be served by the same base station, allowing the base station to provide sidelink resource allocation configuration or assistance to the UEs. For example, both UEs can be within the coverage area of ​​one of the base stations depicted in Figure 1. This is called the "in-coverage" scenario. Another scenario is called the "out-of-coverage" scenario. Note that "out-of-coverage" does not necessarily mean that the two UEs are outside one of the cells depicted in Figure 1, but rather that these UEs:

[0010] - It may not be connected to the base station, for example, they are not in an RRC connected state, causing the UE not to receive any sidelink resource allocation configuration or assistance from the base station, and / or

[0011] - It may connect to a base station; however, for one or more reasons, the base station may not provide the UE with sidelink resource allocation configuration or assistance, and / or

[0012] - May connect to base stations that may not support NR V2X services, such as GSM, UMTS, LTE base stations or WiFi APs.

[0013] Figure 2(A) is a schematic representation of an in-coverage scenario where two directly communicating UEs are connected to a base station. The base station gNB has a coverage area schematically represented by circle 200, which essentially corresponds to the cell schematically represented in Figure 1. The directly communicating UEs include a first vehicle 202 and a second vehicle 204, both within the coverage area 200 of the base station gNB. Vehicles 202 and 204 are both connected to the base station gNB and are directly connected to each other via the PC5 interface. The scheduling and / or interference management of V2V services is assisted by the gNB via control signaling on the Uu interface, which is the radio interface between the base station and the UE. In other words, the gNB provides SL resource allocation configuration or assistance to the UEs and allocates resources to be used for V2V communication on the sidelink. This configuration is also known as Mode 1 configuration in NR V2X and Mode 3 configuration in LTE V2X. Therefore, in Mode 1, the UE (e.g., UE 202) is connected to the gNB via the Uu interface, and the gNB coordinates the resources of UE 202 for transmitting control and / or data to another UE (e.g., UE 204) via the SL interface (referred to as PC5 in NR).

[0014] Figure 2(B) is a schematic representation of an out-of-coverage scenario, where UEs communicating directly with each other are either not connected to a base station (although they may be physically located within a cell of the wireless communication network), or some or all of the UEs communicating directly with each other are connected to a base station but the base station does not provide SL resource allocation configuration or assistance. Three vehicles 206, 208, and 210 are shown communicating directly with each other via a side link (e.g., using a PC5 interface). V2V service scheduling and / or interference management are based on algorithms implemented between the vehicles. This configuration is also known as Mode 2 configuration in NR V2X and Mode 4 configuration in LTE V2X. As mentioned earlier, the out-of-coverage scenario in Figure 2(B) does not necessarily mean that the corresponding Mode 2 UE in NR or Mode 4 UE in LTE is outside the base station coverage area 200, but rather that the corresponding Mode 2 UE in NR or Mode 4 UE in LTE is not served by the base station, is not connected to a base station in the coverage area, or is connected to a base station but does not receive SL resource allocation configuration or assistance information from the base station. Therefore, in some cases, within the coverage area 200 shown in Figure 2(A), in addition to NR mode 1 UEs or LTE mode 3 UEs 202 and 204, there may also be NR mode 2 or LTE mode 4 UEs 206, 208, and 210. Furthermore, Figure 2(B) schematically illustrates out-of-coverage UEs using relay and network communication. For example, UE 210 can communicate with UE 212 via a sidelink, and UE 212 can connect to the gNB via the Uu interface. Therefore, UE 212 can relay information between the gNB and UE 210. Thus, SL-UEs (e.g., UEs 206-210) do not need to have connectivity to the gNB, and when transmitting from UE 206 to UE 208, sensing and access resource allocation or random access-based resource allocation are performed. However, UEs 206-210 need to have available basic configuration for successful data exchange. This information can be pre-configured or configured when the UE is within the gNB coverage area. For this purpose, the gNB can provide basic configuration (e.g., basic information), which can be transmitted via a broadcast channel (e.g., using a System Information Block (SIB)). The BS can also assist the Mode 2 UE by providing basic information about which resource pool (RP) to use, or it can act as a synchronization source.

[0015] Although Figures 2(A) and 2(B) show vehicle-mounted UEs, it should be noted that the described in-coverage and out-of-coverage scenarios also apply to non-vehicle-mounted UEs. In other words, any UE, such as a handheld device, can be in-coverage or out-of-coverage when communicating directly with another UE using the SL channel.

[0016] Typically, Mode 1 refers to operation supported by the RAN (Radio Access Point) including the base station, while Mode 2 refers to autonomous mode, where the UE communicates directly without base station support. In the context of WiFi, coordination by the WiFi access point (AP) can be described as similar to Mode 1 operation, while Mode 2 corresponds to WiFi autonomous mode. In the latter case, two WiFi devices can communicate directly without the assistance of a WiFi AP.

[0017] In mobile communication networks, such as the LTE / 4G, 5G / NR, or WiFi networks described above with reference to Figure 1, data transmission on the communication link between the UE and one or more network entities of the wireless communication network may be impaired by specific events. For example, beam link failure (BLF) or radio link failure (RLF) may occur. Furthermore, if communication is conducted using unlicensed frequency bands, data transmission may be impaired if the UE fails to access the channel due to listen-before-tell (LBT).

[0018] BLF indicates that the currently used beam cannot maintain reliable communication. For example, a 5G beam link failure (BLF) is a situation where the communication link between a 5G base station (gNB) and a user equipment (UE), or between two UEs, is interrupted due to a loss of beam alignment. This can be caused by various factors such as mobility, interference, congestion, or misconfiguration. A BLF between two UEs may occur when they communicate directly via a sidelink (e.g., through a PC5 interface). In this case, if the UE is equipped with multiple antennas, or antenna arrays or antenna panels (which may be the case if transmitting in a higher frequency band, such as FR2), they can utilize beam management. Beam management techniques can be applied to ensure beam alignment between the transmitting and receiving UEs.

[0019] When a beam link failure occurs, the UE attempts to restore the connection by scanning alternative beams from the same gNB or from an adjacent gNB. The gNB also assists the UE by providing a beam failure detection reference signal and configuration parameters. If the UE cannot find a suitable beam within a specific time, it declares a radio link failure (RLF) and performs a reconstruction procedure. In the case of a BLF on a side link, the exact RLF procedure may depend on whether the UE is operating within the gNB's coverage (Mode 1) or outside the gNB's coverage (Mode 2). In Mode 1, a similar gNB-assisted RLF procedure can be utilized. In the case of Mode 2, assistance from the network cannot be expected; the UE may try to change modes or must perform the reconstruction procedure itself, for example, by transmitting a wider beam to find a beampup link (BPL) between the two UEs.

[0020] When an RLF (Remote Link Failure) is triggered, it means that the communication link between the UE and the destination has been interrupted due to various reasons, such as poor signal quality, interference, mobility issues, or handover failure. Different actions are taken to recover from an RLF depending on the network configuration and the type of cell group where the failure occurred. For example:

[0021] If the UE is in NR or LTE standalone operation and the RLF occurs in the primary cell group (MCG) (the only cell group the UE is connected to), the UE declares the RLF and triggers the RRC re-establishment procedure. This procedure involves sending an RRCReestablishmentRequest message to the network, containing the reason for the RLF and the UE's identifier. The network then attempts to re-establish the connection with the UE by sending an RRCReestablishment message containing the UE's new configuration. If the re-establishment is successful, the UE resumes normal operation. If unsuccessful, the UE enters the RRC_IDLE state and performs cell selection.

[0022] - If the UE is in Multi-Radio Dual Connectivity (MR-DC) operation, meaning it is connected to two cell groups: the primary cell group (MCG) via the 4G base station (MN) and the secondary cell group (SCG) via the 5G base station (SN), different scenarios may occur depending on which cell group fails. For example:

[0023] If an RLF (Link Recovery Fault) occurs within an MCG (Multi-Channel Group), the UE uses the SCG (Signal Segment Group) to report the fault to the network and request a handover to another MCG. This allows the UE to quickly recover from an MCG failure without losing connection or data. This feature is called Fast MCG Link Recovery.

[0024] If an RLF (Radio Link Failure) occurs in an SCG (Signal Cell Group), the UE marks the radio link failure on the SCG cell and sends an SCGFailureInformation message to the network via the MCG (Mechanical Cell Group). This message contains the cause of the RLF and some measurement results from neighboring cells. The network then decides whether to reconfigure or release the UE's SCG. This function is called SCG connectivity.

[0025] LBT failure refers to a situation where a device attempting to transmit data on unlicensed spectrum fails to detect an empty channel before sending a signal. LBT stands for Listen-Before-Speak, a protocol that requires devices to sense channel conditions and avoid interfering with other transmissions. In some regions (such as Europe and Japan), LBT is mandatory for unlicensed spectrum operation, while in others (such as the United States and China) it is optional. 5G unlicensed operation allows 5G New Radio (NR) to operate on unlicensed spectrum bands (such as 5 GHz and 6 GHz). 5G unlicensed operation can be used in different modes, such as carrier aggregation with licensed bands, dual connectivity with licensed bands, or standalone operation only on unlicensed bands. 5G unlicensed operation can provide higher bandwidth and capacity for 5G services, but it also faces challenges such as coexistence with other technologies (such as Wi-Fi) and meeting regulatory requirements.

[0026] In the context of unlicensed 5G, an LBT (Local Level-to-Band) failure means that a 5G NR device cannot access the unlicensed channel due to the presence of other signals or noise. This can degrade the performance and reliability of unlicensed 5G transmissions. When a UE faces an LBT failure, it may take different actions depending on the scenario and configuration. Some possible actions might be:

[0027] If the UE is performing initial access to unlicensed spectrum, it will retry the LBT procedure until successful or the maximum number of attempts is reached. If access to the channel is still unsuccessful after the maximum number of attempts is reached, it will report a fault to the network and wait for further instructions.

[0028] If a UE transmits data on an unlicensed spectrum using carrier aggregation or dual connectivity with a licensed frequency band, it will suspend transmission on the unlicensed carrier and continue using the licensed carrier. It will also notify the network of an LBT failure and request new licenses for the unlicensed carrier.

[0029] If the UE transmits data on unlicensed spectrum in standalone mode, it will suspend transmission and wait for a new transmission opportunity. It will also notify the network of an LBT failure and request a new license for the unlicensed carrier.

[0030] If the UE transmits HARQ feedback on unlicensed spectrum, it will discard the feedback and wait for the network to retransmit the data. The network assumes that no feedback has been received and retransmits the data accordingly.

[0031] In wireless network or communication systems, artificial intelligence (AI) and machine learning (ML) can be used for specific tasks. For example, according to 3GPP, AI / ML technologies and data analytics can be incorporated into 5G system design to support certain tasks, such as supporting network automation, data collection for various network functions, network energy saving, resource allocation and scheduling optimization, network slicing management, load balancing, mobility optimization, AI / ML-based services, and AI / ML for the New Radio (NR) air interface. For instance, when considering the NR air interface, AI / ML models can be used for one or more of the following use cases:

[0032] - Channel State Information (CSI):

[0033] For example, AI / ML can be used for time-domain prediction.

[0034] - Beam Management (BM):

[0035] For example, AI for beam management in 5G involves using AI and ML technologies to improve the efficiency and reliability of wireless communications using directional beams. Beam management is the process of guiding, tracking, and selecting the optimal beam for each user and link in a 5G network. This is challenging due to factors such as user mobility, congestion and reflections, multi-user interference, increased antenna numbers, and the adoption of higher frequencies. AI and ML can provide valuable solutions to alleviate this complexity and minimize the overhead associated with beam management and selection, while maintaining system performance.

[0036] - Unlicensed frequency band operation:

[0037] For example, AI for 5G unlicensed spectrum channel access involves using AI and ML technologies to improve the efficiency and reliability of wireless communications using unlicensed spectrum. Unlicensed spectrum is the portion of radio frequency spectrum not allocated to any specific service or operator, and can be used by anyone who complies with certain rules and regulations. Unlicensed spectrum can provide 5G applications with more bandwidth, lower costs, and greater flexibility, especially when licensed spectrum is scarce or expensive. AI for 5G unlicensed spectrum channel access presents several challenges and opportunities, such as:

[0038] ○ Channel Access Methods: Different methods exist for accessing unlicensed channels, such as Listen-After-Speak (LBT), gap-based channel access, and contention-based random access. Each method has its own advantages and disadvantages in terms of latency, throughput, fairness, and overhead. AI and ML can help design, optimize, and adapt these methods based on network conditions and user needs.

[0039] ○ Spectrum Sharing and Coexistence: Unlicensed spectrum is shared by multiple users and technologies, such as Wi-Fi, Bluetooth, LTE-U, LAA, MulteFire, CBRS, NR, etc. This can lead to interference, congestion, and collisions between different transmissions. AI and ML can help enhance spectrum sharing and coexistence mechanisms, such as sensing, coordination, scheduling, power control, beamforming, etc., to improve spectrum efficiency and quality of service.

[0040] ○ Private Networks and Industrial IoT: Unlicensed spectrum can support the deployment of 5G private networks and industrial IoT applications, such as smart factories, warehouses, and mines. These applications have high requirements for reliability, security, and low latency. AI and ML can help customize and optimize network performance for these applications, such as intelligent load balancing, proactive network slicing, and anomaly detection.

[0041] - position:

[0042] For example, direct AI / ML localization methods (e.g., fingerprint localization) and AI / ML-assisted localization methods (e.g., the output of AI / ML model inference is additional measurements and / or enhancements to existing measurements) can be implemented.

[0043] AI / ML models can operate on one or both sides of the communication link, for example, on the gNB or network side (e.g., CN) and / or on the UE. Some AI / ML models may not be specified and left to implementation, while others (e.g., enabling AI / ML for the air interface) need to be specified.

[0044] It should be noted that the information in the above sections is only used to enhance the understanding of the background of the present invention, and therefore may contain information that does not constitute prior art known to those skilled in the art.

[0045] Based on the above, improvements or enhancements may be needed in the use of AI / ML models in wireless communication systems or networks.

[0046] Embodiments of the invention will now be described in more detail with reference to the accompanying drawings:

[0047] Figures 1(A)-1(B) illustrate a wireless communication network, wherein Figure 1(A) is an example schematic diagram of a terrestrial wireless communication network and Figure 1(B) is an example schematic diagram of a radio access network (RAN);

[0048] Figure 2(A) is a schematic diagram of the scene within the coverage area;

[0049] Figure 2(B) is a schematic diagram of the scene outside the coverage area;

[0050] Figure 3To illustrate the wireless communication system of this embodiment, the wireless communication system includes a transmitter (such as a base station) and one or more receivers (such as user equipment UE).

[0051] Figure 4 A user equipment (UE) according to an embodiment of the present invention is shown;

[0052] Figure 5 illustrates an embodiment of the present invention when AI / ML is used in beam management;

[0053] Figure 6 illustrates another embodiment of the present invention when AI / ML is employed in beam management; and

[0054] Figure 7 An example of a computer system is shown, on which the units or modules and method steps of the present invention can be executed.

[0055] Embodiments of the invention will now be described in more detail with reference to the accompanying drawings, wherein the same or similar elements have the same reference numerals.

[0056] In wireless communication system networks, such as the network described above with reference to Figure 1, one or more artificial intelligence / machine learning models (AI / ML models) or one or more AI / ML functions can be implemented in user equipment or user terminal (UE) to perform one or more tasks, such as one or more of the following:

[0057] - RAN access based on AI / ML models

[0058] - Network energy saving based on AI / ML models

[0059] - Load balancing based on AI / ML models

[0060] - Mobility optimization based on AI / ML models

[0061] - Use cases based on AI / ML models, such as:

[0062] ○ Channel State Information (CSI) feedback, such as CSI compression and / or CSI prediction, or

[0063] ○ Beam management, or

[0064] ○Location, such as direct AI / ML location (e.g., fingerprint location) and / or AI / ML assisted location,

[0065] - Mobility management based on AI / ML models, such as handover (HO) prediction and / or conditional handover (CHO) prediction.

[0066] - Modulation and coding strategy (MCS) selection based on AI / ML models

[0067] - Synchronization based on AI / ML models

[0068] - Encoding and / or decoding and / or pre-encoding based on AI / ML models,

[0069] - Modulation and / or demodulation based on AI / ML models

[0070] - Localization or ranging based on AI / ML models

[0071] - Joint Communication and Sensing (JSAC) based on AI / ML models.

[0072] - Feedback calculations based on AI / ML models, such as channel state information (CSI), channel quality indicator (CQI), preference matrix index (PMI), and rank indicator feedback.

[0073] - Interference management based on AI / ML models

[0074] - AI / ML model-based prediction of Quality of Experience (QoE) and / or Quality of Service (QoS).

[0075] - Network traffic forecasting based on AI / ML models.

[0076] When one or more AI / ML models or functions are implemented in wireless communication networks (such as 3GPP networks or WiFi networks), the overall operation of the network or the efficiency of certain functions within the network can be improved. For example, AI / ML can be used to enhance the air interface in 5G networks. The corresponding AI / ML models are trained on a training dataset when implemented (e.g., within a user equipment), and the trained AI / ML models are used to perform specific tasks. Furthermore, the corresponding AI / ML models can be generalized. The generalization of an AI / ML model describes how it adapts to new data, and is one of the key capabilities for evaluating model performance. For example, when considering 3GPP wireless communication networks, the following scenarios can be considered to verify the generalization performance of AI / ML models considering various scenarios / configurations:

[0077] - Scenario 1:

[0078] The AI / ML model is trained on a dataset from scenario #A / configuration #A, and then the AI / ML model performs inference or testing on the same dataset for scenario #A / configuration #A (i.e., the dataset for scenario #A / configuration #A).

[0079] - Scenario 2:

[0080] The AI / ML model is trained on a dataset from scenario #A / configuration #A, and then the AI / ML model performs inference or testing on datasets different from scenario #A / configuration #A (e.g., datasets from scenario #B / configuration #B or datasets from scenario #A / configuration #B).

[0081] - Scenario 3:

[0082] The AI / ML model is trained on a dataset constructed by mixing datasets from multiple scenarios / configurations (including a first scenario #A / configuration #A and a second dataset different from the first scenario / configuration, such as a dataset from scenario #B / configuration #B or scenario #A / configuration #B). The AI / ML model then performs inference or testing on datasets from a single scenario / configuration (e.g., scenario #A / configuration #A, or scenario #B / configuration #B, or scenario #A / configuration #B) among the multiple scenarios / configurations.

[0083] It should be noted that the number of multiple scenarios / configurations can be greater than 2. Furthermore, the proportion of dataset mixing can be reported.

[0084] While AI / ML can offer the aforementioned benefits, such as potential performance gains relative to non-AI methods, it also has some weaknesses. For example, when considering 3GPP networks, AI / ML can be used in PHY layer use cases such as beam prediction, CSI prediction, CSI compression, and positioning. Despite the potential performance gains compared to traditional methods, some weaknesses remain. For instance, when considering the aforementioned generalizations, AI / ML performance may degrade in certain situations. Its performance can even be catastrophic, leading to severe connectivity problems between the UE and its communication partners (e.g., gNB or access point AP). Such problems may manifest as BLF or even RLF. As mentioned above, BLF indicates that the currently used beam cannot maintain reliable communication. RLF is more severe because RLF indicates a complete communication failure, such as due to persistent transmission or reception failures, or a failed beam recovery process. Furthermore, if communication is conducted using unlicensed frequency bands, the UE may be unable to access the channel due to LBT failures, leading to communication problems between the gNB and the UE. While there may be some monitoring mechanisms to assess AI performance, such mechanisms may be too slow or may miss certain aspects, causing the link to fail even though the monitoring does not detect the problem.

[0085] In other words, there may be situations where AI / ML processes, which typically improve communication link performance, may fail to deliver good performance under certain circumstances. This can lead to, for example, RLF, BLF, LBT failures, or collisions on the communication link. For instance, if AI / ML is used for beam management, it may predict the wrong beam, resulting in a BLF. If this prediction fails in some location, AI / ML may cause persistent failures in the beam process even after recovery from the BLF via beam link recovery. In some cases, this can even lead to an RLF. When using AI / ML to intelligently select channels for LBTs to enable UE coordination, it's possible for AI / ML to select the wrong channel, i.e., causing an LBT failure, or "winning" an LBT but subsequently colliding with another UE performing its LBT at the same time and on the same channel. Therefore, AI / ML can actually significantly degrade performance.

[0086] Therefore, improvements or enhancements may be needed in the AI / ML models or AI / ML functions that handle one or more tasks associated with performing data transmission on the communication link between the UE and one or more network entities of the wireless communication network.

[0087] Embodiments of the present invention address the aforementioned problems by allowing user devices to either switch to or adhere to conventional processes (i.e., non-AI processes). The present invention provides a user equipment (UE) for a wireless communication network, configured or pre-configured with at least one artificial intelligence / machine learning model (AI / ML model) or at least one AI / ML function for performing one or more tasks associated with data transmission on a communication link between the UE and one or more network entities of the wireless communication network. The UE monitors the communication link to acquire one or more specific events, and in response to detecting one or more specific events on the communication link or in response to specific signaling from a network entity of the wireless communication network, the UE triggers one or more actions regarding the use of one or more AI / ML models or AI / ML functions.

[0088] This invention is advantageous because it avoids undesirable damage to the communication link or data transmission on the communication link due to one or more AI / ML models or functions failing to provide acceptable performance. If it is determined that the use of AI / ML may actually degrade performance on the communication link, the UE may discontinue the currently used AI / ML and further operate based on conventional (non-AI) procedures, or may consider modifying the currently used AI / ML.

[0089] Embodiments of the present invention can be implemented in a wireless communication system including a base station and a user (such as a mobile terminal or IoT device), as shown in FIG1. Figure 3This is a schematic representation of a wireless communication system 310, including a transmitter 300 (such as a base station) and one or more receivers 302, 304 (such as user equipment UE). The transmitter 300 and receivers 302, 304 can communicate via one or more wireless communication links or channels 306a, 306b, 308 (such as radio links). The transmitter 300 may include one or more antennas ANT. T The receivers 302 and 304 may include one or more antennas (ANT~UE~) or an antenna array with multiple antennas, signal processors 302a and 304a, and transceivers 302b and 304b, which are coupled to each other. The base station 300 and UEs 302 and 304 may communicate via corresponding first wireless communication links 306a and 306b (e.g., radio links using the Uu interface), while UEs 302 and 304 may communicate with each other via a second wireless communication link 308 (e.g., radio links using the PC5 or sidelink SL interface). When UEs are not served by the base station or are not connected to the base station (e.g., they are not in an RRC connection state), or more generally, when the base station does not provide SL resource allocation configuration or assistance, UEs may communicate with each other via a sidelink. Figure 3 Systems or networks Figure 3 One or more UEs 302, 304 and Figure 3 The base station 300 can be operated according to the teachings of the invention described herein.

[0090] According to aspect 1, a user equipment (UE) for a wireless communication network is provided.

[0091] The UE is configured or pre-configured with at least one artificial intelligence / machine learning model (AI / ML model) or at least one AI / ML function for performing one or more tasks associated with data transmission on the communication link between the UE and one or more network entities of the wireless communication network.

[0092] The UE is used to monitor the communication link for one or more specific events.

[0093] The UE is configured to: in response to detecting one or more specific events in a specific event on the communication link or in response to specific signaling from a network entity of the wireless communication network, trigger one or more actions regarding the use of one or more AI / ML models or AI / ML functions.

[0094] According to aspect 2 relating to aspect 1, the UE is configured to trigger one or more actions after the nth occurrence of the specific event or the nth occurrence of the specific signaling, where n is an integer greater than 0.

[0095] According to aspect 3 of aspect 2, the UE triggers one or more actions only when the n specific events or specific signaling occur within a predefined time window.

[0096] According to aspect 4, which involves one of aspects 1 to 3, the one or more specific events include one or more of the following:

[0097] - The communication link between the UE and one or more entities of the wireless communication network is impaired;

[0098] - Impairment of data transmission on the communication link between the UE and one or more entities of the wireless communication network.

[0099] According to aspect 5 of aspect 4, the UE is configured to: determine if a communication link is impaired if one or more of the following apply:

[0100] - Experience one or more communication link failures;

[0101] - The quality of the communication link degrades below the configured or pre-configured threshold, or degrades by more than the configured or pre-configured amount during a predefined time period, such as Reference Received Power (RSRP), Reference Received Quality (RSRQ), Radio Signal Strength Indicator (RSSI), Signal-to-Interference-plus-Noise Ratio (SINR), or Signal-to-Noise Ratio (SNR).

[0102] According to aspect 6 of aspect 5, one or more failures include one or more of the following:

[0103] - The number of consecutive failures;

[0104] - The number of failures during a predefined time period, such as the number of failures within a configured or pre-configured time interval;

[0105] - The percentage of failures during a predefined time period, such as the percentage of failures within a configured or pre-configured time interval.

[0106] According to aspect 7 relating to any one of aspects 4 to 6, the UE is configured to: determine that data transmission on the communication link is impaired if one or more of the following applies:

[0107] - One or more data transmissions failed;

[0108] - The ratio of successful data transmission to unsuccessful data transmission from the UE to one or more entities exceeds a configured or pre-configured threshold, such as the ratio of HARQ-ACK to HARQ-NACK.

[0109] - The signal strength of the radio signal including the data transmission drops below a configured or pre-configured threshold, or drops more than a configured or pre-configured amount during a predefined time period, such as Reference Received Power (RSRP), Reference Received Quality (RSRQ), Radio Signal Strength Indicator (RSSI), Signal-to-Interference-plus-Noise Ratio (SINR), or Signal-to-Noise Ratio (SNR).

[0110] - The number or proportion of data transmissions that conflict with other UEs during a predefined time period exceeds a configured or pre-configured threshold;

[0111] - The level of interference detected on the communication link exceeds the configured or pre-configured threshold;

[0112] - Network congestion exceeds the configured or pre-configured threshold;

[0113] - The number of failed RACH attempts exceeds the configured or pre-configured threshold;

[0114] - The number of failed SIB decoding attempts exceeds the configured or pre-configured threshold;

[0115] - The number of failed cell reselection attempts exceeds the configured or pre-configured threshold;

[0116] - Geographic region / location;

[0117] - Scene type, such as urban area, suburbs or rural area.

[0118] According to aspect 8 relating to any of the foregoing aspects, specific signaling from the network entity, such as from a neighboring UE or from a base station, is provided by the network entity when the network entity detects one or more specific events on other communication links between the network entity and the UE, or between the network entity and one or more other entities of the wireless communication network, such as damage to the other communication links or damage to data transmission on the other communication links.

[0119] According to aspect 9 of aspect 8, the UE is used to receive the specific signaling via unicast transmission, multicast or multi-cast transmission, or broadcast transmission.

[0120] According to aspect 10 relating to aspect 9, the specific signaling from the network entity, such as from a neighboring UE or base station, is multicast (e.g., in the Group Common Physical Downlink Control Channel (GC-PDCCH)) or broadcast, such as SIB, which instructs all UEs or a group of UEs to perform one or more actions regarding the use of one or more AI / ML models or AI / ML functions.

[0121] According to aspect 11 of aspect 10, one or more actions concerning the use of one or more AI / ML models or AI / ML functions include: deactivating all and / or some AI / ML models or functions in all UEs or specific UEs, for example, based on emergency trigger signaling.

[0122] According to aspect 12 relating to any of the foregoing aspects, the UE is configured to trigger one or more actions regarding the use of one or more AI / ML models or AI / ML functions only when one or more other configuration or pre-configuration conditions are met.

[0123] According to aspect 13 relating to aspect 12, the one or more configuration or pre-configuration conditions include one or more of the following:

[0124] - The specific location or area where the UE is located;

[0125] - The signal strength of the wireless signal including the data transmission drops below a configured or pre-configured threshold, or drops more than a configured or pre-configured amount during a predefined time period, such as Reference Received Power (RSRP), Reference Received Quality (RSRQ), Radio Signal Strength Indicator (RSSI), Signal-to-Interference-plus-Noise Ratio (SINR), or Signal-to-Noise Ratio (SNR).

[0126] - The percentage or number of LBT failures caused by communication from the base station serving the UE exceeds the configured or pre-configured threshold;

[0127] - There is no evacuation signal on one or more frequency bands where data transmission is performed.

[0128] According to aspect 14 of aspect 13, the threshold changes dynamically based on temporal conditions.

[0129] According to aspect 15 relating to any of the foregoing aspects, one or more actions include one or more of the following:

[0130] - Modifications to one or more currently used AI / ML models or AI / ML functions;

[0131] - Modify the communication link and continue to use one or more currently used AI / ML models or AI / ML functions for the modified communication link;

[0132] - Modify one or more AI / ML models or AI / ML functions currently in use and the communication link, and use the modified one or more AI / ML models or AI / ML functions for the modified communication link.

[0133] According to aspect 16 of aspect 15, modifications to one or more AI / ML models or AI / ML functions include one or more of the following:

[0134] - Deactivate some or all AI / ML models or functions currently in use;

[0135] - Switch to a different AI / ML model or AI / ML function to perform the one or more tasks, such as configuring or pre-configuring a fallback AI / ML model or fallback AI / ML function;

[0136] - Adapt some or all of the AI / ML models or functions currently in use, such as increasing quantization granularity, resetting parameters to initial or default values, performing retraining, or performing fine-tuning.

[0137] - Transition to a disconnected state, such as the RRC_IDLE state or the RRC_INACTIVE state, which allows the use of disconnected AI / ML models and functions;

[0138] - Switch to rollback to a non-AI process.

[0139] According to aspect 17 of aspect 16, for the non-connected state, the UE is used to use a more robust non-connected mode AI / ML model or AI / ML function compared to the connected mode AI / ML model or AI / ML function.

[0140] Based on aspect 18 of aspect 16 or 17, activate some or all of the currently used AI / ML models or AI / ML functions, including one or more of the following additional actions:

[0141] - Turn off AI / ML models or AI / ML functions;

[0142] - Stop executing the one or more of the tasks mentioned above;

[0143] - Switch to a different task;

[0144] - Perform the one or more tasks using conventional computing techniques.

[0145] According to aspect 19, which involves one of aspects 15 to 18,

[0146] The modifications to the communication link include one or more of the following:

[0147] - Perform carrier aggregation, for example, if the UE has already performed carrier aggregation, aggregate carriers in a lower frequency band;

[0148] - Switch to a carrier in another frequency band, such as a carrier in FR1, and deactivate a carrier in a higher frequency band;

[0149] - Switch to a wider beam, for example by temporarily deactivating AI beamforming;

[0150] - Switch to a different beam pattern / codebook that is more suitable for the current network scenario;

[0151] - Perform a handover to a new cell with a better communication link;

[0152] - Redirect the UE to a less congested cell or sector to balance network load and reduce the likelihood of RLF.

[0153] According to aspect 20 relating to any of the foregoing aspects, the UE is used to report the one or more actions to the wireless communication network, such as to the base station of the wireless communication network, for example via RRC, MAC-CE, or PHY layer signaling.

[0154] According to aspect 21 which relates to any of the foregoing aspects, one or more tasks include one or more of the following:

[0155] - RAN access based on AI / ML models;

[0156] - Network energy saving based on AI / ML models;

[0157] - Load balancing based on AI / ML models;

[0158] - Mobility optimization based on AI / ML models;

[0159] - Use cases based on AI / ML models, such as:

[0160] ○ Channel State Information (CSI) feedback, such as CSI compression and / or CSI prediction; or

[0161] ○ Beam management; or

[0162] ○Location, such as direct AI / ML location (e.g., fingerprint location) and / or AI / ML assisted location;

[0163] - Mobility management based on AI / ML models, such as switching HO prediction and / or conditional switching CHO prediction;

[0164] - Selection of modulation and encoding / decoding schemes (MCS) based on AI / ML models;

[0165] - Synchronization based on AI / ML models;

[0166] - Encoding and / or decoding and / or precoding based on AI / ML models;

[0167] - Modulation and / or demodulation based on AI / ML models;

[0168] - Localization or ranging based on AI / ML models;

[0169] - Joint Communication and Perception JSAC based on AI / ML models;

[0170] - Feedback computation based on AI / ML models, such as CSI / CQI / PMI / RI feedback;

[0171] - Interference management based on AI / ML models;

[0172] - Predictions of Quality of Experience (QoE) and / or Quality of Service (QoS) based on AI / ML models;

[0173] - Network traffic forecasting based on AI / ML models.

[0174] According to aspect 22 relating to any of the foregoing aspects, the UE includes one or more of the following: a power-limited UE, or a handheld UE (such as a UE used by a pedestrian, referred to as a Vulnerable Road User (VRU) Pedestrian UE P-UE), or a personal or handheld UE used by public safety personnel and emergency responders (referred to as a Public Safety UE PS-UE), or an IoT UE or an environmental IoT UE (e.g., a sensor, actuator, or UE deployed in a campus network to perform repetitive tasks and requiring periodic input from a gateway node), or a mobile terminal, or a fixed terminal, or a cellular IoT UE, or an industrial IoT UE IIoT, or a SL UE, or a vehicle-mounted UE, or a vehicle-mounted group leader UE (GL-UE), or a dispatch UE S-UE, or an IoT or narrowband IoT (NB-IoT) device, NTN UE, or WiFi device or WiFi site STA, or ground vehicle, or aircraft, or drone, or mobile base station, or roadside unit RSU, or building, or any other item / device (e.g., sensor or actuator) equipped with network connectivity to enable the item / device to communicate using the wireless communication network, or any other item / device (e.g., sensor or actuator) equipped with network connectivity to enable the item / device to communicate using the sidelink of the wireless communication network, or any network entity with sidelink capability.

[0175] According to aspect 23, a wireless communication network, such as a 3GPP system, includes one or more user equipment (UE) as described in any of the preceding aspects and one or more base stations (BS).

[0176] According to aspect 24 of aspect 23, the BS includes one or more of the following: a macro cell base station, or a small cell base station, or a centralized unit of a base station, or a distributed unit of a base station, or an integrated access backhaul (IAB) node, or a roadside unit (RSU), or a WiFi access point (AP), or a UE, or an SL UE, or a group leader UE (GL-UE), or a relay or remote radio head, or an AMF, or an SMF, or a core network entity, or a mobile edge computing (MEC) entity, or a network slice in an NR or 5G core context, or any transmit / receive point (TRP) that enables an item or device to communicate using the wireless communication network, the item or device being equipped with a network connection to communicate using the wireless communication network.

[0177] According to aspect 25, a method is provided for operating a user equipment (UE) for a wireless communication network, wherein the UE is configured or pre-configured with at least one artificial intelligence / machine learning model (AI / ML) or at least one AI / ML function for performing one or more tasks associated with data transmission on a communication link between the UE and one or more network entities of the wireless communication network, the method comprising:

[0178] Monitor the communication link for one or more specific events; and

[0179] In response to the detection of one or more specific events on the communication link, or in response to specific signaling from a network entity of the wireless communication network, one or more actions are triggered regarding the use of one or more AI / ML models or AI / ML functions.

[0180] According to aspect 26, a non-transitory computer program product is provided, comprising a computer-readable medium storing instructions that, when executed on a computer, perform the method according to aspect 25.

[0181] The present invention provides a computer program product, including instructions that, when executed by a computer, cause the computer to perform one or more methods according to the present invention.

[0182] Embodiments of the invention will now be described in more detail with reference to the accompanying drawings. It should be noted that the aspects or embodiments subsequently outlined and described may be combined such that some or all of the aspects / embodiments are implemented in one embodiment. This document refers to one or more AI / ML models and / or one or more AI / ML functions. It should be noted that when only an AI / ML model is mentioned, it should be understood that it also refers to an AI / ML function, and when only an AI / ML function is mentioned, it should be understood that it also refers to an AI / ML model. An AI / ML function may refer to an AI / ML enabled feature / feature group (FG) enabled by one or more configurations, wherein said one or more configurations may be supported based on one or more conditions indicated by UE capabilities. An AI / ML enabled feature refers to a feature that can use AI / ML. It should be noted that a UE may have one AI / ML model for that function, or a UE may have multiple AI / ML models for that function. Examples of use cases for AI / ML enabled features or feature groups include:

[0183] - Enhanced CSI feedback, such as reduced overhead, improved accuracy, and improved prediction.

[0184] - Beam management, such as time and / or spatial domain beam prediction, for overhead and latency reduction and improved beam selection accuracy.

[0185] - Enhanced positioning accuracy in different scenarios, including scenarios with severe non-line-of-sight (NLOS) conditions.

[0186] Other examples may include RAN access, Network Energy Saving (NES), resource management and load balancing, mobility enhancement and optimization (including handover (HO) management and / or prediction, Conditional Handover (CHO) management and / or prediction), Modulation and Coding Scheme (MCS) selection, MIMO precoder computation, conventional PHY layer signal processing (e.g., synchronization, channel coding or decoding, modulation or demodulation), location or ranging, Joint Communication and Sensing (JSAC), feedback computation of CSI / CQI or PMI / RI, conventional MIMO processing, equalization, network offloading, interference management, Quality of Experience (QoE) and / or Quality of Service (QoS) prediction, and / or network traffic forecasting. It is important to note that the AI / ML methods of the selected sub-use cases need to be sufficiently diverse to support the various requirements of the gNB-UE cooperation level.

[0187] AI / ML models operate based on identified models, where the models can be associated with specific configurations / conditions related to one or more UE capabilities of AI / ML-enabled features / FGs, as well as additional conditions (e.g., scenarios, sites, and datasets) determined / identified between the UE side and the network side.

[0188] Figure 4A user equipment (UE) according to an embodiment of the present invention is illustrated. The UE 400 includes a signal processing unit or signal processor 402 and one or more antennas or antenna arrays 404 for communicating with other network entities via an air interface. Figure 4 As shown, UE 400 can communicate with a base station or gNB 406 using Uu interface 408, and / or conduct side-link (SL) communication with another UE 410 using PC5 interface 412. As illustrated, UE 400 is configured or pre-configured with at least one artificial intelligence / machine learning model (AI / ML model) or at least one AI / ML function 414 to perform one or more tasks associated with data transmission on communication links 408 or 412 between UE 400 and another gNB 406 or between another UE 410.

[0189] In a connected state (such as the RRC_CONNECTED state), UE 400 can use a configured / pre-configured AI / ML model or function 414, also referred to as a "connected AI / ML model or function". The UE receives an activation signal from, for example, gNB 406, to activate one or more of the connected AI / ML models or functions. Therefore, one or more connected AI / ML models or functions include one or more of the configured or pre-configured AI / ML models or AI / ML functions 414. According to an embodiment, in a disconnected state (such as the RRC_INACTIVE state or the RRC_IDLE state), UE 400 can also utilize one or more of the configured / pre-configured AI / ML models or functions 414, also referred to as "disconnected AI / ML models or functions". Disconnected AI / ML models or functions include one or more of the configured or pre-configured AI / ML models or AI / ML functions 414. In other words, when in a disconnected state, UE 400 can activate one or more or all of the configured or pre-configured AI / ML models or functions 414 without any explicit or implicit signaling from the network side that would cause such activation, and UE 400 can also benefit from the advantages of implementing AI / ML methods in UE, such as improving potential transmissions that UE may send or receive even when in a disconnected state.

[0190] UE 400 monitors communication links, such as communication link 408 or communication link 412 when communicating with gNB 406 and / or UE 410, to acquire specific events, as illustrated at 416. When one or more of the specific events on communication links 408 and 412 are detected, or when specific signaling is received from a network entity (such as a communication partner, e.g., gNB 406 or UE 410), the UE triggers one or more actions regarding the use of one or more AI / ML models and AI / ML functions, as illustrated at 418.

[0191] Figure 5 illustrates an embodiment of the present invention implemented using AI / ML in beam management. UE 400 includes multiple antennas or an antenna array with multiple elements, allowing the UE to direct transmit or receive beams to a desired direction. As shown in Figure 5(A), UE 400 can form beams B1, B2, and B3 pointing in different directions (each with its main lobe). To select the beam for communication with base station 406, UE 400 employs an AI / ML model or function 414a, referred to as beam AI. It is assumed that beam AI 414a selects beam B1 for communication with base station 406. However, as shown, beam B1 is not pointed at base station 406, therefore UE 400 experiences a disruption in the communication link between UE 400 and base station 406. For example, beam B1 selected by AI / ML 414a may lead to BLF or even RLF, as shown at 420. In response to the detection of event 420, or in response to the detection of event 420 multiple times within a predefined time period, UE 400 decides to take action regarding the use of the AI / ML model. In the depicted embodiment, UE 400 decides to deactivate beam AI 414b, as shown at 422 in FIG5(B). Deactivating beam AI 414b causes UE 400 to either switch to a non-AI procedure, i.e., a conventional procedure that does not use the AI / ML model or functionality. Using the conventional non-AI method, UE 400 selects beam B3 pointing to gNB 406 for communication to initiate a beam recovery procedure or an RRC reconstruction procedure.

[0192] Figure 6 illustrates another embodiment of the present invention when employing AI / ML in beam management. Similar to Figure 5, UE 400 includes multiple antennas or an antenna array with multiple elements, allowing the UE to direct transmit or receive beams to a desired direction. As shown in Figure 6(A), UE 400 can form beams B1, B2, and B3 pointing in different directions (with their main lobes). To select the beam for communication with base station 406, UE 400 employs an AI / ML model or function 414a, referred to as beam AI. It is assumed that beam AI 414a selects beam B1 for communication with base station 406. However, as shown, beam B1 is not pointed at base station 406, therefore UE 400 experiences a disruption in the communication link between UE 400 and base station 406. For example, beam B1 selected by AI / ML 414a may lead to BLF or even RLF, as shown at 420. In response to the detection of event 420, or in response to the detection of event 420 multiple times within a predefined time period, UE 400 decides to take action regarding the use of the AI / ML model. In the depicted embodiment, UE 400 decides to adapt beam AI 414b, as shown at 424 in FIG6(B). Adapting beam AI 414b causes UE 400 to either switch to an alternative AI / ML model or modify beam AI 414b. Using the adapted beam AI 414b, UE 400 selects beam B3 pointing to gNB406 for communication to initiate a beam recovery procedure or an RRC reconstruction procedure.

[0193] Therefore, according to embodiments of the present invention, the degradation or impairment of communication between UE 400 and gNB 406 caused by performance degradation of beam AI 414a is overcome by deactivating or adapting beam AI 414a and selecting an appropriate beam B3 using, for example, a non-AI procedure or other AI procedure more suitable for the situation. This enables the UE to rebuild the link using the appropriate beam B3. This solves the performance degradation problem of the AI / ML model or function used to support the transmissions sent or received by UE 400 on the communication link, and the impairment on the link can be avoided or quickly resolved by taking appropriate actions, such as deactivating the relevant AI / ML model or switching back to a non-AI / ML procedure.

[0194] According to other embodiments, after adapting its operation, UE 400 may report the adaptation of beam AI to gNB 406, for example, after link reconstruction, as schematically shown at 426 in FIG6(B). It should be noted that in the embodiment of FIG5, UE 400 may also report the deactivation of beam AI 414a to gNB 406. In other words, UE 400 may report to gNB 406 that it has taken certain actions. For example, UE 400 may report that some or all AI / ML models / functions have been deactivated, or that it has switched to another AI / ML model or function, or that one or more or all of the AI / ML models / functions have been adapted (e.g., by resetting AI / ML). Signaling or reporting 422 may be provided using RRC signaling, or one or more MAC-CEs, or PHY layer signaling (such as UCI).

[0195] According to an embodiment, UE 400 may trigger one or more actions immediately after the first detection of a specific event (such as BLF or RLF), such as the deactivation of beam AI 414a described above. According to other embodiments, to avoid frequent reconfiguration of UE 400 regarding the use of implemented AI / ML, the corresponding actions regarding AI / ML use may only be taken when a specific event or specific signaling occurs multiple times (e.g., within a predefined time period). For example, UE 400 may trigger the one or more actions after the nth occurrence of a specific event or the nth occurrence of a specific signaling, where n is an integer greater than 0. According to an embodiment, UE 400 may trigger the one or more actions when the number of events / signaling during the predefined time period exceeds a threshold. According to an embodiment, the predefined time period or time mentioned herein may be the time or time window of a specific AI / ML model / function activity.

[0196] In other words, a trigger may not be activated after the first occurrence of a triggering factor or event, but rather after a certain number of occurrences. For example, the first BLF may not trigger activation, but the trigger can be activated based on the failure behavior within a specific time window. For instance, a trigger may be activated only after n BLFs (where n is an integer greater than 0) or after a specific frequency or periodicity of BLFs. As mentioned earlier, the time window can also be the time of activity of a specific AI model / feature.

[0197] According to an embodiment, specific events on communication links 408 and 412 involve damage or degradation of the actual communication links 408 and 412 between UE 400 and its communication partners (e.g., gNB 406 or UE 410). In addition to degradation of the actual communication links, specific events can also be damage or degradation of actual data transmission on communication links 408 and 412 between UE and gNB 406 and / or UE 410.

[0198] According to an embodiment, if one or more failures occur in the communication link, the UE 400 determines that the communication link is impaired. For example, if the number of continuous failures of the communication link within a specific time period exceeds a configured or pre-configured threshold, the communication link can be determined to be impaired. It should be noted that the threshold can be set to one or more failures. The threshold can be configured differently depending on various factors, such as geographic region / location, scenario type (urban / suburban / rural), site configuration, and / or can change dynamically, for example, depending on the time of day. Failures can include one or more of the following: beam link failure (BLF), radio link failure (RLF), listen-before-tell (LBT) failure, or handover failure from the source base station to the target base station of the currently serving UE. The number of continuous failures can include one or more of the following:

[0199] - The number of consecutive failures, such as n consecutive failures, where n is an integer greater than 0;

[0200] - The number of failures during a predefined time period, for example, n failures within a configured or pre-configured time interval, where n is an integer greater than 0;

[0201] - The percentage of failures during a predefined time period, such as the percentage of failures within a configured or pre-configured time interval.

[0202] According to other embodiments, if the quality of the communication link degrades to below a configured or pre-configured threshold, for example, during a predefined time period, or degrades by more than a configured or pre-configured amount during a predefined time period, then UE 400 determines that the communication link is impaired. The quality of the communication link can be determined using one or more of the following: Reference Signal Received Power (RSRP), or Reference Signal Received Quality (RSRQ), or Radio Signal Strength Indicator (RSSI), or Signal-to-Interference Plus Noise Ratio (SINR), or Signal-to-Noise Ratio (SNR).

[0203] According to other embodiments, UE 400 determines that data transmission on the communication link is impaired if one or more of the following apply:

[0204] - One or more data transfers fail, for example, within a predefined time period.

[0205] For example, if the number of unsuccessful data transmissions from the UE to one or more of the entities exceeds a configured or pre-configured threshold, data transmission is determined to be impaired. It should be noted that the threshold can be set to one or more unsuccessful data transmissions. For example, if the number of Hybrid Automatic Repeat Request Negative Acknowledgments (HARQ-NACK) exceeds the threshold, data transmission on the communication link is determined to be impaired.

[0206] For example, if the number of successful data transmissions from the UE to one or more of the entities decreases below a configured or pre-configured threshold, data transmission is determined to be impaired. It should be noted that the threshold can be set to one or more successful data transmissions. For example, if the number of Hybrid Automatic Repeat Request Acknowledgments (HARQ-ACK) drops below the threshold, data transmission on the communication link is determined to be impaired.

[0207] - The ratio of successful to unsuccessful data transmissions from the UE to one or more entities, for example, exceeding a configured or pre-configured threshold within a predefined time period, such as the ratio of HARQ-ACK to HARQ-NACK.

[0208] - The signal strength of radio signals, including those used for data transmission, may drop below a configured or pre-configured threshold within a predefined time period, or drop by an amount exceeding a configured or pre-configured threshold within a predefined time period, such as Reference Signal Received Power (RSRP), Reference Signal Received Quality (RSRQ), Radio Signal Strength Indicator (RSSI), Signal-to-Interference-plus-Noise Ratio (SINR), or Signal-to-Noise Ratio (SNR).

[0209] - The number or percentage of data transmissions that collide with other UEs within a predefined time period exceeds a configured or pre-configured threshold. Note that the threshold can be set to one or more collisions.

[0210] - The level of interference detected on the communication link exceeds the configured or pre-configured threshold.

[0211] - Network congestion exceeds the configured or pre-configured threshold.

[0212] For example, the number of failed RACH attempts within a predefined time period exceeds a configured or pre-configured threshold. Note that the threshold can be set to one or more failed RACH attempts.

[0213] For example, the number of failed SIB decoding attempts within a predefined time period exceeds a configured or pre-defined threshold. Note that the threshold can be set to one or more failed SIB decoding attempts.

[0214] For example, the number of failed cell reselection attempts within a predefined time period exceeds a configured or pre-configured threshold. It's important to note that the threshold can be set to one or more failed cell reselection attempts.

[0215] So far, the present invention has been described with reference to embodiments in which UE 400 triggers one or more actions regarding the use of one or more AI / ML models or AI / ML functions in response to the detection of one or more specific events on a communication link. The invention is not limited to such embodiments. According to another embodiment, UE 400 triggers one or more actions regarding the use of one or more AI / ML models or AI / ML functions in response to specific signaling from a network entity of the wireless communication network. For example, the specific signaling may originate from a network entity with which UE 400 communicates, such as from a neighboring UE 410 or from base station 406. The signaling may be provided by the network entity upon detecting one or more specific events on another communication link between the network entity and the UE, or on another communication link between the network entity and one or more other entities of the wireless communication network, such as damage to the other communication link or damage to data transmission on the other communication link.

[0216] According to an embodiment, UE 400 receives the specific signaling via unicast, multicast, or broadcast transmission. For example, the specific signaling from a network entity (e.g., from a neighboring UE or from a base station) can be multicast (e.g., in a Group Common PDCCH (GC-PDCCH)) or broadcast (e.g., SIB), instructing all UEs or a group of UEs to perform one or more actions regarding the use of one or more AI / ML models or AI / ML functions. In other words, UE 400 can receive a broadcast or multicast emergency deactivation signal that can be used to activate the AI ​​of all or some UEs due to a problem on the communication link. According to an embodiment, the UE can also switch to RRC_IDLE after an emergency is triggered. For example, in some cases, the AI / ML in UEs in a specific area may perform catastrophically due to environmental changes, and the currently used AI / ML scheme may not be adequately adapted. This will result in performance degradation for almost all UEs in that area. Therefore, in such scenarios, the network entity can use emergency trigger signaling to deactivate all and / or some AI / ML models or functions in all and / or some UEs. Compared to state-of-the-art techniques that require deactivating certain AI / ML models or functions individually for each UE, the proposed method allows network entities to respond rapidly to changing circumstances. Therefore, the proposed method prevents further degradation due to AI / ML issues and avoids more serious problems, such as simultaneous communication link failures across many UEs.

[0217] In some scenarios, triggering may be activated, such as due to RLF, BLF, etc.; however, the probability that the fault is caused by an AI / ML fault may be low. For example, if the UE loses connection after tunneling, or if the LBT fault is caused by a large amount of communication from the gNB, the performance degradation on / on the communication link is unlikely to be due to AI / ML performance degradation. Furthermore, certain unlicensed frequency bands may be withdrawn due to evacuation signals, leading to LBT faults and / or HARQ-NACK. In these cases, although a trigger is activated, taking action may be detrimental. Therefore, according to an embodiment, the UE 400 triggers one or more actions regarding the use of one or more AI / ML models or AI / ML functions only when one or more additional configuration or pre-configuration conditions are met. The one or more configuration or pre-configuration conditions may include one or more of the following:

[0218] - The specific location or area where the UE is located, that is, the UE 400 needs to be located in a predefined location where the present invention is enabled.

[0219] - This includes situations where the signal strength of radio signals used for data transmission drops below a configured or pre-configured threshold, for example, within a predefined time period, or drops by an amount exceeding a configured or pre-configured threshold within a predefined time period, such as the Reference Signal Received Power (RSRP), Reference Signal Received Quality (RSRQ), Radio Signal Strength Indicator (RSSI), Signal-to-Interference Plus Noise Ratio (SINR), or Signal-to-Noise Ratio (SNR). Therefore, additional conditions act as filters, ensuring that when RLF is considered as a trigger for action, these actions are performed only when, for example, the SNR falls below a specific threshold.

[0220] - The percentage or number of LBT failures due to communication by the base station serving the UE exceeds a configured or pre-configured threshold. This threshold can change dynamically depending on timing conditions, such as time of day. Therefore, LBT failures due to the gNB are not considered. For example, the gNB may use a channel, causing an LBT failure at the UE. Since the gNB has a fair sharing medium mechanism, this may be intentional on the part of the gNB and therefore is not a triggering condition for performing the one or more actions mentioned above.

[0221] - There is no evacuation signal on one or more frequency bands where data transmission is performed. Some frequency bands may have shared access. For example, they are also used for military purposes. Then there are often signals that are reserved for military use, causing all non-military equipment to not use the frequency band for a specific period of time. Obviously, this is not a failure of the AI, therefore, although such an event will certainly trigger an RLF, it should not change the AI.

[0222] An example RRC configuration for dynamic AI deactivation can be as follows:

[0223]

[0224] The `TriggerSignal` parameter configures which trigger event is monitored for emergency actions. The `TriggerRSRPThreshold` parameter describes the RSRP range within which one or more actions are taken. For example, this might be beneficial if the RSRP is very low. In this case, the trigger event is likely activated by poor channel conditions, and taking action on the AI / ML process may not improve the situation. The `TriggerOnlyOnEvacSignalAbsence` parameter configures whether to omit triggering when an evacuation signal is present. For example, certain frequency bands may be shared with the military. With military users using the band, an evacuation signal can be transmitted, causing all non-military users to not use the band for a specific period. For example, the presence of such a signal in this case could lead to a persistent LBT failure or RLF. However, taking action on the AI / ML process may not be beneficial because the communication link problem is caused by evacuation. The `MinimumTimeToLastReport` parameter can define the time to the last measurement or performance report regarding AI / ML performance or the last trigger. If the time between the activation of the trigger and the last report or trigger is less than MinimumTimeToLastReport, the UE may not take any action, as it expects the network entity to be aware of the potential problem and be able to take appropriate action.

[0225] In the embodiment described above with reference to FIG. 5, the action taken by UE 400 is to deactivate beam AI 414a and perform beam selection using a non-AI procedure. The invention is not limited to this embodiment; rather, UE 400 may also perform other actions. According to embodiments, the one or more actions may include modifying one or more of the currently used AI / ML model or AI / ML function, as shown with reference to FIG. 6. For example, the modification may include one or more of the following:

[0226] - Deactivate some or all of the currently used AI / ML model or AI / ML function. For example, deactivating some or all of the currently used AI / ML model or AI / ML function may include one or more of the following further actions: shutting down the AI / ML model or AI / ML function, stopping the execution of the one or more tasks, switching to a different task, or using conventional computing techniques to perform the one or more tasks.

[0227] - Switch to a different AI / ML model or AI / ML function to perform the one or more tasks, for example, configure or pre-configure a fallback AI / ML model or fallback AI / ML function.

[0228] - Adapt some or all of the AI / ML models or functions currently in use, such as increasing quantization granularity, resetting parameters to initial or default values, performing retraining, or performing fine-tuning.

[0229] - Transition to a disconnected state, such as RRC_IDLE or RRC_INACTIVE. The disconnected state may or may not allow the use of disconnected AI / ML models and functions. In disconnected mode, the UE can switch to disconnected mode AI / ML, which may be more robust than connected mode AI / ML. The UE may or may not perform a RACH using disconnected mode AI to transition back to the RRC_CONNECTED state.

[0230] - Switch to the standby non-AI process. Since AI is an additional top-level function, all functions can also be performed without AI (or at least without AI known to 3GPP, such as internal chip components). This is called the standby process.

[0231] According to other embodiments, the one or more actions may include modifying the communication link and continuing to use one or more currently used AI / ML models or AI / ML functions on the modified communication link. For example, modifying the communication link may include one or more of the following:

[0232] - Perform carrier aggregation operations, such as aggregating carriers in a lower frequency band if the UE has already performed carrier aggregation.

[0233] - Switch to a carrier in another frequency band, such as a carrier in FR1, and deactivate a carrier in a higher frequency band.

[0234] - Switch to a wider beam, for example by temporarily deactivating AI beamforming.

[0235] - Switch to a different beam pattern / codebook that is more suitable for the current network scenario.

[0236] - Perform a handover to a new cell with a better communication link.

[0237] - Redirect the UE to a less congested cell or sector to balance network load and reduce the likelihood of RLF.

[0238] According to other embodiments, the one or more actions may include modifying one or more currently used AI / ML models or AI / ML functions and communication links, and using the one or more modified AI / ML models or AI / ML functions on the modified communication links.

[0239] General

[0240] The various embodiments of the present invention have been described in detail above. Different embodiments and technical solutions can be implemented individually or by combining two or more embodiments or solutions.

[0241] According to embodiments, the wireless communication system may include a terrestrial network, a non-terrestrial network, or a network or network segment that uses an airborne vehicle or spacecraft as a receiver, or a combination thereof. Furthermore, the wireless communication system may be a system or network different from the aforementioned 4G or 5G mobile communication systems; rather, embodiments of the present invention can also be implemented in any other wireless communication network, for example, in a private network, such as an intranet or any other type of campus network, or in a WiFi communication system.

[0242] According to embodiments of the present invention, the user equipment includes one or more of the following: a power-limited UE, or a handheld UE, such as a UE used by pedestrians, referred to as a vulnerable road user (VRU) or pedestrian UE (P-UE), or a personal or handheld UE used by public safety personnel and first responders, referred to as a public safety UE (PS-UE), or an IoT UE, such as a sensor, actuator, or UE located in a campus network to perform repetitive tasks and requiring input from a gateway node at periodic intervals, a mobile terminal, or a fixed terminal, or a cellular IoT-UE, or a vehicle-mounted UE, or a vehicle-mounted group leader (GL) UE, or a sidelink relay, or an IoT or narrowband IoT (NB-IoT) device, or a wearable device such as a smartwatch, fitness tracker, smart glasses, or a ground vehicle, or an air vehicle, or a drone, or a mobile base station, or a roadside unit (RSU), or a building, or any other item or device with network connectivity that enables the item / device to communicate using the wireless communication network, such as a sensor or actuator, or any other item or device with network connectivity that enables the item / device to communicate using a sidelink (a sidelink of the wireless communication network), such as a sensor or actuator, or a Wi-Fi device such as a station (STA), access point (AP), node or mesh node, or mesh point, or Mesh AP, or any network entity that supports a sidelink.

[0243] According to embodiments of the present invention, the network entity includes one or more of the following: a macro cell base station, or a small cell base station, or a central unit of a base station, an integrated access and backhaul (IAB) node, or a distributed unit of a base station, or a roadside unit (RSU), or a Wi-Fi device, such as an access point (AP) or a mesh AP, or a remote radio head, or an AMF, or an MME, or an SMF, or a core network entity, or a mobile edge computing (MEC) entity, or a network slice in an NR or 5G core context, or any transmit / receive point (TRP) that enables an item or device to communicate using the wireless communication network, wherein the item or device is provided with network connectivity for communicating using the wireless communication network.

[0244] Although some aspects of the described concepts are described in the context of the apparatus, these aspects also clearly represent descriptions of the corresponding methods, where a block or device corresponds to a method step or a feature of a method step. Similarly, aspects described in the context of a method step also represent descriptions of corresponding blocks, items, or features of the corresponding apparatus.

[0245] The various elements and features of this invention can be implemented in hardware using analog and / or digital circuits, in software by executing instructions through one or more general-purpose or special-purpose processors, or as a combination of hardware and software. For example, embodiments of this invention can be implemented in a computer system or other processing system environment. Figure 7 An example of a computer system 600 is shown. Units or modules executed by these units, as well as method steps, can be performed on one or more computer systems 600. The computer system 600 includes one or more processors 602, such as dedicated or general-purpose digital signal processors. The processors 602 are connected to a communication infrastructure 604, such as a bus or network. The computer system 600 includes main memory 606, such as random access memory (RAM), and secondary memory 608, such as a hard disk drive and / or a removable storage drive. The secondary memory 608 allows computer programs or other instructions to be loaded into the computer system 600. The computer system 600 may also include a communication interface 610 to allow software and data to be transferred between the computer system 600 and external devices. Communication can be in the form of electronic, electromagnetic, optical, or other signals that can be processed by the communication interface. Communication can use wires or cables, optical fibers, telephone lines, cellular telephone links, RF links, and other communication channels 612.

[0246] The terms "computer program medium" and "computer-readable medium" are generally used to refer to tangible storage media, such as removable storage units or hard disks installed in hard disk drives. These computer program products are means for providing software to computer system 600. The computer program (also referred to as computer control logic) is stored in main memory 606 and / or auxiliary memory 608. The computer program may also be received via communication interface 610. When executed, the computer program enables computer system 600 to implement the present invention. Specifically, when executed, the computer program enables processor 602 to implement the processes of the present invention, as described herein. Thus, such a computer program can represent a controller of computer system 600. In the case of implementing this disclosure using software, the software may be stored in a computer program product and loaded into computer system 600 using a removable storage drive or an interface (such as communication interface 610).

[0247] The implementation in hardware or software can be executed using digital storage media, such as cloud storage, floppy disks, DVDs, Blu-rays, CDs, ROMs, PROMs, EPROMs, EEPROMs, or FLASH memories, which store electronically readable control signals that cooperate with or are capable of cooperating with a programmable computer system to execute the corresponding methods. Therefore, the digital storage medium can be computer-readable.

[0248] Some embodiments of the invention include a data carrier having electronically readable control signals capable of cooperating with a programmable computer system to perform one of the methods described herein.

[0249] Typically, embodiments of the present invention can be implemented as a computer program product having program code that, when run on a computer, performs a method. The program code may, for example, be stored on a machine-readable medium.

[0250] Other embodiments include a computer program stored on a machine-readable medium for performing one of the methods described herein. In other words, an embodiment of the method of the present invention is therefore a computer program having program code that, when run on a computer, performs one of the methods described herein.

[0251] Another embodiment of the method of the present invention is therefore a data carrier or digital storage medium, or a computer-readable medium, on which a computer program for performing one of the methods described herein is recorded. Another embodiment of the method of the present invention is therefore a data stream or signal sequence representing a computer program for performing one of the methods described herein. The data stream or signal sequence may, for example, be configured to be transmitted via a data communication connection (e.g., via the Internet). Another embodiment includes a processing device, such as a computer or programmable logic device, configured or adapted to perform one of the methods described herein. Another embodiment includes a computer on which a computer program for performing one of the methods described herein is mounted.

[0252] In some embodiments, a programmable logic device (e.g., a field-programmable gate array) may be used to perform some or all of the functions of the methods described herein. In some embodiments, the field-programmable gate array may cooperate with a microprocessor to perform one of the methods described herein. Generally, these methods are preferably performed by any hardware device.

[0253] The above embodiments are merely illustrative of the principles of the present invention. It should be understood that modifications and variations of the arrangements and details described herein will be readily apparent to those skilled in the art. Therefore, the intent is limited only by the scope of the forthcoming patent claims and not by the specific details presented through the description and explanation of the embodiments herein.

Claims

1. A user equipment (UE) for a wireless communication network, The UE is configured or pre-configured with at least one artificial intelligence / machine learning model (AI / ML model) or at least one AI / ML function for performing one or more tasks associated with data transmission on the communication link between the UE and one or more network entities of the wireless communication network. in, The UE is used to monitor the communication link in response to one or more specific events; and The UE is configured to: in response to detecting one or more specific events in a specific event on the communication link, or in response to specific signaling from a network entity of the wireless communication network, trigger one or more actions regarding the use of one or more AI / ML models or AI / ML functions.

2. The user equipment (UE) according to claim 1, wherein, The UE is used to trigger one or more actions after the nth occurrence of the specific event or the nth occurrence of the specific signaling, where n is an integer greater than 0.

3. The user equipment (UE) according to claim 2, wherein, The UE is used to trigger one or more actions only when n specific events or n specific signaling events occur within a predefined time window.

4. The user equipment (UE) according to any one of claims 1 to 3, wherein the one or more specific events include one or more of the following: - The communication link between the UE and one or more entities of the wireless communication network is impaired; - Impairment of data transmission on the communication link between the UE and one or more entities of the wireless communication network.

5. The user equipment (UE) according to claim 4, wherein, The UE is configured to determine if the communication link is impaired if one or more of the following apply: - Experience one or more communication link failures; - The quality of the communication link degrades below the configured or pre-configured threshold, or degrades by more than the configured or pre-configured amount during a predefined time period, such as Reference Received Power (RSRP), Reference Received Quality (RSRQ), Radio Signal Strength Indicator (RSSI), Signal-to-Interference-plus-Noise Ratio (SINR), or Signal-to-Noise Ratio (SNR).

6. The user equipment (UE) according to claim 5, wherein, The one or more faults include one or more of the following: - The number of consecutive failures; - The number of failures during a predefined time period, such as the number of failures within a configured or pre-configured time interval; - The percentage of failures during a predefined time period, such as the percentage of failures within a configured or pre-configured time interval.

7. The user equipment (UE) according to any one of claims 4 to 6, wherein, The UE is configured to determine if data transmission on the communication link is impaired if one or more of the following apply: - One or more data transmissions failed; - The ratio of successful data transmission to unsuccessful data transmission from the UE to one or more entities exceeds a configured or pre-configured threshold, such as the ratio of HARQ-ACK to HARQ-NACK; - The signal strength of the radio signal including the data transmission drops below a configured or pre-configured threshold, or drops more than a configured or pre-configured amount during a predefined time period, such as Reference Received Power (RSRP), Reference Received Quality (RSRQ), Radio Signal Strength Indicator (RSSI), Signal-to-Interference-plus-Noise Ratio (SINR), or Signal-to-Noise Ratio (SNR). - The number or proportion of data transmissions that conflict with other UEs during a predefined time period exceeds a configured or pre-configured threshold; - The level of interference detected on the communication link exceeds the configured or pre-configured threshold; - Network congestion exceeds the configured or pre-configured threshold; - The number of failed RACH attempts exceeds the configured or pre-configured threshold; - The number of failed SIB decoding attempts exceeds the configured or pre-configured threshold; - The number of failed cell reselection attempts exceeds the configured or pre-configured threshold; - Geographic region / location; - Scene type, such as urban area, suburbs or rural area.

8. The user equipment (UE) according to any one of the preceding claims, wherein, Specific signaling from the network entity, such as from a neighboring UE or from a base station, is provided by the network entity when the network entity detects one or more specific events on other communication links between the network entity and the UE, or between the network entity and one or more other entities of the wireless communication network, such as damage to the other communication links or damage to data transmission on the other communication links.

9. The user equipment (UE) according to claim 8, wherein, The UE is used to receive the specific signaling via unicast, multicast, or broadcast transmission.

10. The user equipment (UE) according to claim 9, wherein, The specific signaling from the network entity, such as from a neighboring UE or from a base station, is multicast, such as in a group common PDCCH (GC-PDCCH), or broadcast, such as in an SIB, instructing all UEs or a group of UEs to perform one or more actions regarding the use of one or more AI / ML models or AI / ML functions.

11. The user equipment (UE) according to claim 10, wherein, One or more actions regarding the use of one or more AI / ML models or AI / ML functions include: deactivating all AI / ML models or functions and / or some AI / ML models or functions in all UEs or specific UEs, for example, based on emergency trigger signaling.

12. The user equipment (UE) according to any one of the preceding claims, wherein, The UE is used to trigger one or more actions regarding the use of one or more AI / ML models or AI / ML functions only when one or more other configuration or pre-configuration conditions are met.

13. The user equipment (UE) according to claim 12, wherein, One or more configuration or pre-configuration conditions include one or more of the following: - The specific location or area where the UE is located; - The signal strength of the radio signal including the data transmission drops below a configured or pre-configured threshold, or drops more than a configured or pre-configured amount during a predefined time period, such as Reference Received Power (RSRP), Reference Received Quality (RSRQ), Radio Signal Strength Indicator (RSSI), Signal-to-Interference-plus-Noise Ratio (SINR), or Signal-to-Noise Ratio (SNR). - The percentage or number of LBT failures caused by communication from the base station serving the UE exceeds the configured or pre-configured threshold; - There is no evacuation signal on one or more frequency bands where data transmission is performed.

14. The user equipment (UE) according to claim 13, wherein, The threshold changes dynamically based on time-series conditions.

15. The user equipment (UE) according to any one of the preceding claims, wherein, The one or more actions include one or more of the following: - Modifications to one or more AI / ML models or AI / ML functions currently in use; - Modify the communication link and continue to use one or more currently used AI / ML models or AI / ML functions for the modified communication link; - Modify one or more AI / ML models or AI / ML functions and communication links currently in use, and use the modified one or more AI / ML models or AI / ML functions for the modified communication links.

16. The user equipment (UE) according to claim 15, wherein, Modifications to one or more AI / ML models or AI / ML functions include one or more of the following: - Deactivate some or all AI / ML models or functions currently in use; - Switch to a different AI / ML model or AI / ML function to perform the one or more tasks, such as configuring or pre-configuring a fallback AI / ML model or fallback AI / ML function; - Adapt some or all of the AI / ML models or functions currently in use, such as increasing quantization granularity, resetting parameters to initial or default values, performing retraining, or performing fine-tuning. - Transition to a disconnected state, such as the RRC_IDLE state or the RRC_INACTIVE state, which allows the use of disconnected AI / ML models and functions; - Switch to rollback to a non-AI process.

17. The user equipment (UE) of claim 16, wherein, In the non-connected state, the UE is used to use a more robust non-connected mode AI / ML model or AI / ML function compared to the connected mode AI / ML model or AI / ML function.

18. The user equipment (UE) according to claim 16 or 17, wherein, To activate some or all of the currently used AI / ML models or AI / ML functions, including one or more of the following other actions: - Turn off AI / ML models or AI / ML features; - Stop executing the one or more of the tasks mentioned above; - Switch to a different task; - Perform the one or more tasks using conventional computing techniques.

19. The user equipment (UE) according to any one of claims 15 to 18, wherein, The modifications to the communication link include one or more of the following: - Perform carrier aggregation, for example, if the UE has already performed carrier aggregation, aggregate carriers in a lower frequency band; - Switch to a carrier in another frequency band, such as a carrier in FR1, and deactivate a carrier in a higher frequency band; - Switch to a wider beam, for example by temporarily deactivating AI beamforming; - Switch to a different beam pattern / codebook that is more suitable for the current network scenario; - Perform a handover to a new cell with a better communication link; - Redirect the UE to a less congested cell or sector to balance network load and reduce the likelihood of RLF.

20. The user equipment (UE) according to any one of the preceding claims, wherein, The UE is used to report one or more actions to the wireless communication network, such as to the base station of the wireless communication network, for example via RRC, MAC-CE, or PHY layer signaling.

21. The user equipment (UE) according to any one of the preceding claims, wherein, The one or more tasks include one or more of the following: - RAN access based on AI / ML models; - Network energy saving based on AI / ML models; - Load balancing based on AI / ML models; - Mobility optimization based on AI / ML models; - Use cases based on AI / ML models, such as: ○ Channel State Information (CSI) feedback, such as CSI compression and / or CSI prediction; or ○ Beam management; or ○Location, such as direct AI / ML location (e.g., fingerprint location) and / or AI / ML assisted location; - Mobility management based on AI / ML models, such as switching HO prediction and / or conditional switching CHO prediction; - Selection of modulation and encoding / decoding schemes (MCS) based on AI / ML models; - Synchronization based on AI / ML models; - Encoding and / or decoding and / or precoding based on AI / ML models; - Modulation and / or demodulation based on AI / ML models; - Localization or ranging based on AI / ML models; - Joint Communication and Perception JSAC based on AI / ML models; - Feedback computation based on AI / ML models, such as CSI / CQI / PMI / RI feedback; - Interference management based on AI / ML models; - Experience Quality of Experience (QoE) and / or Service Quality (QoS) prediction based on AI / ML models; - Network traffic forecasting based on AI / ML models.

22. The user equipment (UE) according to any one of the preceding claims, wherein, The UE includes one or more of the following: a power-limited UE, or a handheld UE (such as a UE used by pedestrians, referred to as a Vulnerable Road User UE or P-UE), or a personal or handheld UE used by public safety personnel and emergency responders (referred to as a Public Safety UE, PS-UE), or an IoT UE or an environmental IoT UE (such as a sensor, actuator, or UE deployed in a campus network to perform repetitive tasks and requiring input from a gateway node at periodic intervals), or a mobile terminal, or a fixed terminal, or a cellular IoT UE, or an industrial IoT UE (IIoT), or a SL UE, or a vehicle-mounted UE, or a vehicle-mounted team leader UE (GL-UE), or a dispatch UE (S-UE), or an IoT or narrowband IoT (NB-IoT) device, NTN UE, or WiFi device or WiFi site STA, or ground vehicle, or aircraft, or drone, or mobile base station, or roadside unit RSU, or building, or any other item / device (e.g., sensor or actuator) equipped with network connectivity to enable the item / device to communicate using the wireless communication network, or any other item / device (e.g., sensor or actuator) equipped with network connectivity to enable the item / device to communicate using the sidelink of the wireless communication network, or any network entity with sidelink capability.

23. A wireless communication network, such as a 3GPP system, comprising one or more user equipment (UE) according to any one of the preceding claims, and one or more base stations (BS).

24. The wireless communication network of claim 23, wherein the BS comprises one or more of the following: a macro cell base station, or a small cell base station, or a centralized unit of a base station, or a distributed unit of a base station, or an integrated access backhaul (IAB) node, or a roadside unit (RSU), or a WiFi access point (AP), or a UE, or an SL UE, or a group leader UE (GL-UE), or a relay or remote radio head, or an AMF, or an SMF, or a core network entity, or a mobile edge computing (MEC) entity, or a network slice in an NR or 5G core context, or any transmit / receive point (TRP) that enables an item or device to communicate using the wireless communication network, the item or device being equipped with a network connection to communicate using the wireless communication network.

25. A method for operating a user equipment (UE) for a wireless communication network, wherein, The UE is configured or pre-configured with at least one artificial intelligence / machine learning model (AI / ML model) or at least one AI / ML function for performing one or more tasks associated with data transmission on a communication link between the UE and one or more network entities of the wireless communication network, the method comprising: Monitor the communication link for one or more specific events; and In response to the detection of one or more specific events on the communication link, or in response to specific signaling from a network entity of the wireless communication network, one or more actions are triggered regarding the use of one or more AI / ML models or AI / ML functions.

26. A non-transitory computer program product comprising a computer-readable medium storing instructions that, when executed on a computer, perform the method of claim 25.

Images