Ai-based sensing in a wireless communication system

Abstract

Various aspects of the present disclosure relate to a network entity for wireless communication. A first network entity may be configured to, capable of, or operable to determine that retraining of an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is needed, transmit, to a second network entity, a training request relating to the AI / ML model, and receive, from the second network entity, a retrained AI / ML model retrained using at least one of pre-existing training data and / or simulated training data generated by a third network entity. A second network entity may be configured to receive the training request, transmit to a third network entity an indication that simulated training data is required, receive the simulated training data, retrain the AI / ML model using the pre-existing training data and / or the simulated training data, and transmit a retrained AI / ML model retrained using at least one of the pre-existing training data and / or the simulated training data.

Description

AI-BASED SENSING IN A WIRELESS COMMUNICATION SYSTEMTECHNICAL FIELD

[0001] The present disclosure relates generally to wireless communication, including AI- based sensing in a wireless communication system.BACKGROUND

[0002] A wireless communications system may include one or multiple network communication devices, which may be otherwise knowns as network equipment (NE) supporting wireless communications for one or multiple user communication devices, which may be otherwise known as user equipment (UE), or other suitable terminology. The wireless communications system may support wireless communications with one or multiple user communication devices by utilising resources of the wireless communication system (e.g., time resources (e.g., symbols, slots, subframes, frames, or the like) or frequency resources (e.g., subcarriers, carriers, or the like). Additionally, the wireless communications system may support wireless communications across various radio access technologies including third generation (3G) radio access technology, fourth generation (4G) radio access technology, fifth generation (5G) radio access technology, among other suitable radio access technologies beyond 5G (e.g., sixth generation (6G)).SUMMARY

[0003] An article “a” before an element is unrestricted and understood to refer to “at least one” of those elements or “one or more” of those elements. The terms “a,” “at least one,” “one or more,” and “at least one of one or more” may be interchangeable. As used herein, including in the claims, “or” as used in a list of items (e.g., a list of items prefaced by a phrase such as “at least one of’ or “one or more of’ or “one or both of’) indicates an inclusive list such that, for example, a list of at least one of A, B, or C means A or B or C or AB or AC or BC or ABC (i.e., A and B and C). Also, as used herein, the phrase “based on” shall not be construed as a reference to a closed set of conditions. For example, an example step that is described as “based on condition A” may be based on both a condition A and a condition B without departing from the scope of the present disclosure. In other words, as used herein,Docket No. SMM920250078-GR-NPthe phrase “based on” shall be construed in the same manner as the phrase “based at least in part on. Further, as used herein, including in the claims, a “set” may include one or more elements.

[0004] The following abbreviations are herewith defined, at least some of which are referred to within the following description: 5G- 5th Generation of Mobile Communications; 5GC - 5G Core; 5QI - 5G QoS Identifier; AF - Application Function; AI / ML - Artificial Intelligence / Machine Learning ; AISF - Al-based Sensing Function; AISLF - Al-based Sensing Logical Function; AISTF - Al-based Sensing Training Function; AMF - Access and Mobility Function; AS - Application Service; DL - Downlink; DNN - Data Network Name; GBR - Guaranteed Bit Rate; gNB - general Node B; IP - Internet Protocol; KPI - Key Performance Indicator; MDA - Management Data Analytics; NDT - Network Digital Twin; NEF - Network Exposure Function; NF - Network Function; NG-RAN - Next Generation- RAN; NR - New Radio; NWDAF - Network Data Analytics Function; 0AM - Operations, Administration and Maintenance; PCF - Policy Control Function; PDU - Packet Data Unit; QoE - Quality of Experience; QFI - QoS Flow Identifier; QoS - Quality of Service; RAN - Radio Access Network; RCF - RAN-core Converged Function; TA - Tracking Area; TRP - Transmission Point; SMF - Session Management Function; UAV - Unmanned Aerial Vehicle; UDM - User Data Manager ; UDR - User Data Repository; UE - User Equipment; UL - Uplink; UPF - User Plane Function; VNF - Virtual Network Function; V2X - Vehicle- to-Everything.

[0005] Network entities for wireless communication are described. Network entities may be configured to, capable of, or operable to perform one or more operations as described herein. For example, each network entity may include at least one memory; and at least one processor coupled with the at least one memory.

[0006] The network entity may be a first network entity and its at least one processor may be configured to cause it to: determine that retraining of an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is needed; transmit, to a second network entity, a training request relating to the AI / ML model; and receive, from the second network entity, a retrained AI / ML model retrained using at least oneDocket No. SMM920250078-GR-NPof pre-existing training data and / or simulated training data generated by a third network entity.

[0007] The network entity may be a second network entity and its at least one processor may be configured to cause it to: receive, from a first network entity, a training request relating to an artificial intelligence / machine learning, AI / ML, model associated with a sensing task; transmit, to a third network entity, an indication that simulated training data for the AI / ML model is required; receive, from the third network entity, the simulated training data for the AI / ML model; retrain the AI / ML model using pre-existing training data and / or the simulated training data; and transmit, to the first network entity, a retrained AI / ML model retrained using at least one of the pre-existing training data and / or the simulated training data.

[0008] The network entity may be a third network entity and its at least one processor may be configured to cause it to: receive, from a second network entity, an indication that simulated training data for an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is required; and transmit, to the second network entity, simulated training data for the AI / ML model.

[0009] A method performed by a first network entity is described. The method may comprise: determining whether retraining of an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is needed; transmitting, to a second network entity, a training request relating to the AI / ML model; and receiving, from the second network entity, a retrained AI / ML model retrained using at least one of pre-existing training data and / or simulated training data generated by a third network entity.

[0010] A method performed by a second network entity is described. The method may comprise: receiving, from a first network entity, a training request relating to an artificial intelligence / machine learning, AI / ML, model associated with a sensing task or transmitting, to a third network entity, an indication that simulated training data for the AI / ML model is required and receiving, from the third network entity, the simulated training data for the AI / ML model; retraining the AI / ML model using pre-existing training data and / or the simulated training data; and transmitting, to the first network entity, a retrained AI / ML model retrained using at least one of the pre-existing training data and / or the simulated training data.Docket No. SMM920250078-GR-NP

[0011] A method performed by a third network entity is described. The method may comprise: receiving, from a second network entity, an indication that simulated training data for an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is required; and transmitting, to the second network entity, simulated training data for the AI / ML model.

[0012] A processor for wireless communication is described. The processor may be configured to, capable of, or operable to perform one or more operations as described herein. For example, the processor may comprise at least one controller coupled with at least one memory and configured to cause the processor to execute the steps of any of the abovedescribed methods.BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 illustrates an example of a wireless communications system in accordance with aspects of the present disclosure.

[0014] Figures 2a and 2b illustrate example configurations of nodes for sensing in a wireless communications system in accordance with aspects of the present disclosure.

[0015] Figures 3a to 3e illustrate examples of architectures for sensing in a wireless communications system in accordance with aspects of the present disclosure.

[0016] Figure 4 illustrates an example of an Al-based sensing architecture in accordance with aspects of the present disclosure.

[0017] Figure 5 illustrates an example of an Al-based sensing architecture in accordance with aspects of the present disclosure.

[0018] Figures 6 to 9 illustrate a process flow 600 in accordance with aspects of the present disclosure.

[0019] Figure 10 illustrates an example of a processor 700 in accordance with aspects of the present disclosure.

[0020] Figure 11 illustrates an example of a network equipment (NE) 800 in accordance with aspects of the present disclosure.Docket No. SMM920250078-GR-NP

[0021] Figure 12 illustrates a flowchart of a method 900 performed by a NE in accordance with aspects of the present disclosure.

[0022] Figure 13 illustrates a flowchart of a method 1000 performed by a NE in accordance with aspects of the present disclosure.

[0023] Figure 14 illustrates a flowchart of a method 1100 performed by a NE in accordance with aspects of the present disclosure.DETAILED DESCRIPTION

[0024] A wireless communication system may comprise network entities including an Al-based Sensing Logical Function (AISLF), an Al-based Sensing Training Function (AISTF), and a Twin Platform that may interface with the AISLF and AISTF via a Twin Controller. In circumstances where inference yields an undesired outcome (such as improper or poor quality results)due to poor quality data or a change in a physical environment context takes place, the present disclosure tends to react to or quantify the undesired outcome and adapt or retrain an artificial intelligence / machine learning (AI / ML) model underpinning that inference. In some circumstances, the AI / / ML model is improved, possibly via retraining, through use of pre-existing training data and / or simulated training data that may be generated using the Twin Platform in a virtualisation of the real / physical world.

[0025] Aspects of the present disclosure are described in the context of a wireless communications system.

[0026] Figure 1 illustrates an example of a wireless communications system 100 in accordance with aspects of the present disclosure. The wireless communications system 100 may include one or more NE 102, one or more UE 104, and a core network (CN) 106. The wireless communications system 100 may support various radio access technologies. In some implementations, the wireless communications system 100 may be a 4G network, such as an LTE network or an LIE -Advanced (LTE-A) network. In some other implementations, the wireless communications system 100 may be a NR network, such as a 5G network, a 5G- Advanced (5G-A) network, or a 5G ultrawideband (5G-UWB) network. In other implementations, the wireless communications system 100 may be a combination of a 4G network and a 5G network, or other suitable radio access technology including Institute ofDocket No. SMM920250078-GR-NPElectrical and Electronics Engineers (IEEE) 802.11 (Wi-Fi), IEEE 802.16 (WiMAX), IEEE 802.20. The wireless communications system 100 may support radio access technologies beyond 5G, for example, 6G. Additionally, the wireless communications system 100 may support technologies, such as time division multiple access (TDMA), frequency division multiple access (FDMA), or code division multiple access (CDMA), etc.

[0027] The one or more NE 102 may be dispersed throughout a geographic region to form the wireless communications system 100. One or more of the NE 102 described herein may be or include or may be referred to as a network node, a base station, a network element, a network function, a network entity, a radio access network (RAN), a NodeB, an eNodeB (eNB), a next-generation NodeB (gNB), or other suitable terminology. An NE 102 and a UE 104 may communicate via a communication link, which may be a wireless or wired connection. For example, an NE 102 and a UE 104 may perform wireless communication (e.g., receive signalling, transmit signalling) over a Uu interface.

[0028] An NE 102 may provide a geographic coverage area for which the NE 102 may support services for one or more UEs 104 within the geographic coverage area. For example, an NE 102 and a UE 104 may support wireless communication of signals related to services (e.g., voice, video, packet data, messaging, broadcast, etc.) according to one or multiple radio access technologies. In some implementations, an NE 102 may be moveable, for example, a satellite associated with a non-terrestrial network (NTN). In some implementations, different geographic coverage areas associated with the same or different radio access technologies may overlap, but the different geographic coverage areas may be associated with different NE 102.

[0029] The one or more UE 104 may be dispersed throughout a geographic region of the wireless communications system 100. A UE 104 may include or may be referred to as a remote unit, a mobile device, a wireless device, a remote device, a subscriber device, a transmitter device, a receiver device, or some other suitable terminology. In some implementations, the UE 104 may be referred to as a unit, a station, a terminal, or a client, among other examples. Additionally, or alternatively, the UE 104 may be referred to as an Internet-of-Things (loT) device, an Internet-of-Everything (loE) device, or machine-type communication (MTC) device, among other examples.Docket No. SMM920250078-GR-NP

[0030] A UE 104 may be able to support wireless communication directly with other UEs 104 over a communication link. For example, a UE 104 may support wireless communication directly with another UE 104 over a device-to-device (D2D) communication link. In some implementations, such as vehicle-to-vehicle (V2V) deployments, vehicle-to-everything (V2X) deployments, or cellular-V2X deployments, the communication link may be referred to as a sidelink. For example, a UE 104 may support wireless communication directly with another UE 104 over a PC5 interface.

[0031] An NE 102 may support communications with the CN 106, or with another NE 102, or both. For example, an NE 102 may interface with other NE 102 or the CN 106 through one or more backhaul links (e.g., SI, N2, N2, or network interface). In some implementations, the NE 102 may communicate with each other directly. In some other implementations, the NE 102 may communicate with each other or indirectly (e.g., via the CN 106. In some implementations, one or more NE 102 may include subcomponents, such as an access network entity, which may be an example of an access node controller (ANC). An ANC may communicate with the one or more UEs 104 through one or more other access network transmission entities, which may be referred to as a radio heads, smart radio heads, or transmission-reception points (TRPs).

[0032] The CN 106 may support user authentication, access authorisation, tracking, connectivity, and other access, routing, or mobility functions. The CN 106 may be an evolved packet core (EPC), or a 5G core (5GC), which may include a control plane entity that manages access and mobility (e.g., a mobility management entity (MME), an access and mobility management functions (AMF)) and a user plane entity that routes packets or interconnects to external networks (e.g., a serving gateway (S-GW), a Packet Data Network (PDN) gateway (P-GW), or a user plane function (UPF)). In some implementations, the control plane entity may manage non-access stratum (NAS) functions, such as mobility, authentication, and bearer management (e.g., data bearers, signal bearers, etc.) for the one or more UEs 104 served by the one or more NE 102 associated with the CN 106.

[0033] The CN 106 may communicate with a packet data network over one or more backhaul links (e.g., via an SI, N2, N2, or another network interface). The packet data network may include an application server. In some implementations, one or more UEs 104Docket No. SMM920250078-GR-NPmay communicate with the application server. A UE 104 may establish a session (e.g., a protocol data unit (PDU) session, or the like) with the CN 106 via an NE 102. The CN 106 may route traffic (e.g., control information, data, and the like) between the UE 104 and the application server using the established session (e.g., the established PDU session). The PDU session may be an example of a logical connection between the UE 104 and the CN 106 (e.g., one or more network functions of the CN 106).

[0034] In the wireless communications system 100, the NEs 102 and the UEs 104 may use resources of the wireless communications system 100 (e.g., time resources (e.g., symbols, slots, subframes, frames, or the like) or frequency resources (e.g., subcarriers, carriers)) to perform various operations (e.g., wireless communications). In some implementations, the NEs 102 and the UEs 104 may support different resource structures. For example, the NEs 102 and the UEs 104 may support different frame structures. In some implementations, such as in 4G, the NEs 102 and the UEs 104 may support a single frame structure. In some other implementations, such as in 5G and among other suitable radio access technologies, the NEs 102 and the UEs 104 may support various frame structures (i.e., multiple frame structures). The NEs 102 and the UEs 104 may support various frame structures based on one or more numerologies.

[0035] One or more numerologies may be supported in the wireless communications system 100, and a numerology may include a subcarrier spacing and a cyclic prefix. A first numerology (e.g., / r=0) may be associated with a first subcarrier spacing (e.g., 15 kHz) and a normal cyclic prefix. In some implementations, the first numerology (e.g., / r=0) associated with the first subcarrier spacing (e.g., 15 kHz) may utilise one slot per subframe. A second numerology (e.g., / r=l) may be associated with a second subcarrier spacing (e.g., 30 kHz) and a normal cyclic prefix. A third numerology (e.g., / r=2) may be associated with a third subcarrier spacing (e.g., 60 kHz) and a normal cyclic prefix or an extended cyclic prefix. A fourth numerology (e.g., / r=3) may be associated with a fourth subcarrier spacing (e.g., 120 kHz) and a normal cyclic prefix. A fifth numerology (e.g., / r=4) may be associated with a fifth subcarrier spacing (e.g., 240 kHz) and a normal cyclic prefix.

[0036] A time interval of a resource (e.g., a communication resource) may be organised according to frames (also referred to as radio frames). Each frame may have a duration, forDocket No. SMM920250078-GR-NPexample, a 10 millisecond (ms) duration. In some implementations, each frame may include multiple subframes. For example, each frame may include 10 subframes, and each subframe may have a duration, for example, a 1 ms duration. In some implementations, each frame may have the same duration. In some implementations, each subframe of a frame may have the same duration.

[0037] Additionally, or alternatively, a time interval of a resource (e.g., a communication resource) may be organised according to slots. For example, a subframe may include a number (e.g., quantity) of slots. The number of slots in each subframe may also depend on the one or more numerologies supported in the wireless communications system 100. For instance, the first, second, third, fourth, and fifth numerologies (i.e., / r=0, jU=l, / r=2, jU=3, / r=4) associated with respective subcarrier spacings of 15 kHz, 30 kHz, 60 kHz, 120 kHz, and 240 kHz may utilise a single slot per subframe, two slots per subframe, four slots per subframe, eight slots per subframe, and 16 slots per subframe, respectively. Each slot may include a number (e.g., quantity) of symbols (e.g., OFDM symbols). In some implementations, the number (e.g., quantity) of slots for a subframe may depend on a numerology. For a normal cyclic prefix, a slot may include 14 symbols. For an extended cyclic prefix (e.g., applicable for 60 kHz subcarrier spacing), a slot may include 12 symbols. The relationship between the number of symbols per slot, the number of slots per subframe, and the number of slots per frame for a normal cyclic prefix and an extended cyclic prefix may depend on a numerology. It should be understood that reference to a first numerology (e.g., / i =0) associated with a first subcarrier spacing (e.g., 15 kHz) may be used interchangeably between subframes and slots.

[0038] In the wireless communications system 100, an electromagnetic (EM) spectrum may be split, based on frequency or wavelength, into various classes, frequency bands, frequency channels, etc. By way of example, the wireless communications system 100 may support one or multiple operating frequency bands, such as frequency range designations FR1 (410 MHz - 7.125 GHz), FR2 (24.25 GHz - 52.6 GHz), FR3 (7.125 GHz - 24.25 GHz), FR4 (52.6 GHz - 114.25 GHz), FR4a or FR4-1 (52.6 GHz - 71 GHz), and FR5 (114.25 GHz - 300 GHz). In some implementations, the NEs 102 and the UEs 104 may perform wireless communications over one or more of the operating frequency bands. In someDocket No. SMM920250078-GR-NPimplementations, FR1 may be used by the NEs 102 and the UEs 104, among other equipment or devices for cellular communications traffic (e.g., control information, data). In some implementations, FR2 may be used by the NEs 102 and the UEs 104, among other equipment or devices for short-range, high data rate capabilities.

[0039] FR1 may be associated with one or multiple numerologies (e.g., at least three numerologies). For example, FR1 may be associated with a first numerology (e.g., / r=0), which includes 15 kHz subcarrier spacing; a second numerology (e.g., / r=l), which includes 30 kHz subcarrier spacing; and a third numerology (e.g., / r=2), which includes 60 kHz subcarrier spacing. FR2 may be associated with one or multiple numerologies (e.g., at least 2 numerologies). For example, FR2 may be associated with a third numerology (e.g., / r=2), which includes 60 kHz subcarrier spacing; and a fourth numerology (e.g., / r=3), which includes 120 kHz subcarrier spacing.

[0040] Some terms used in this disclosure are provided example definitions as follows: Meta space: A subset of the metaverse, that can be implemented using an edge or a remote cloud, depending on the application; Digital Twin: A virtual representation of an object or system designed to accurately reflect a physical object; Network Digital Twin: A virtual model of a wired or wireless network that models its architecture and components while accurately mirroring its real-time state, configuration and behaviour; Twin Platform: A combination ofDTs and NDTs, formed by layering or nesting different DT / NDT components into a single platform to generate simulations based on configured parameters; Al-based Sensing Function (AISF): An NF at the core network that processes 3GPP sensing data sent by 3GPP sensing entities (RAN / UEs) to generate sensing results; Al-based Sensing Training Function (AISTF): A function part of the AISF that trains the Al model for Al-based sensing; Al-based Sensing Logical Function (AISLF): A function part of the AISF that performs inference for Al-based sensing; Physical environment context: The context of the physical world under which a set of sensing results are generated by AISLF; 3 GPP sensing data: Data derived from 3 GPP radio signals impacted (e.g., reflected, refracted, diffracted) by an object or environment of interest for sensing purposes, and optionally processed within the 5G system; Sensing result: processed 3GPP sensing data requested by a service consumer; Sensing service area: a service area where sensing services would solely rely onDocket No. SMM920250078-GR-NPinfrastructures and sensing technologies that can be assumed to be present anywhere where 5G is present. This includes both indoor and outdoor environments; Confidence level describes the percentage of all the possible measured sensing results that can be expected to include the true sensing result considering the accuracy; Sensing Resolution describes the minimum difference in the measured magnitude of target objects (e.g., range, velocity) to be allowed to detect objects in different magnitude. Simulated training data may refer to data generated from simulations by a Twin Platform based on configured parameters to be used for AI / ML model training associated with a sensing task.

[0041] Use cases and technologies, such as the metaverse and wireless sensing, have gained significant attention in the Fifth Generation (5G) of mobile communication systems. The metaverse comprises a persistent, shared virtual world where users can interact through avatars, often in 3D. The metaverse provides an interconnected digital space that users access through various devices and platforms.

[0042] The metaverse can be deployed at both the edge and a remote cloud. A meta space is a subset of the metaverse that can be implemented using an edge or a remote cloud, depending on the application.

[0043] A key enabler of the metaverse is Digital Twin (DT). A DT is a virtual representation of a real-world entity or system, serving as a digital counterpart to a physical object or process. Extending the notion of DT to wireless systems, the concept of Network Digital Twin (NDT) has been defined by ITU, IETF and 3 GPP. An NDT may be considered a replica of the physical components of the real- world mobile network, i.e., considering network elements, configurations, connectivity etc. The NDT may assist to determine how an actual mobile network would behave in certain scenarios considering “what-if’ scenarios. As the physical world is composed of both physical objects and the real network, both DT and NDT may be leveraged to accurately model the real world in the metaverse.

[0044] Integrated sensing and communication, also known as sensing, is another emerging technology. Sensing may be a 5G system feature that enables acquiring information about characteristics of the environment and / or objects within the physical environment using the impact of radio frequency signals upon the said physical objects. In the physical space, 3 GPP sensing entities such as UEs and / or gNBs may use 3 GPP based sensors to collectDocket No. SMM920250078-GR-NPsensing data e.g. data derived from the reflection, refraction, and / or diffraction of radio signals with objects in the environment. A 5G wireless sensing service may further process the transmissions, reflections and scattering of the wireless sensing signals and may provide the results of the sensing as a 5G wireless sensing service to an authorised consumer. In some applications, such as assisted automotive manoeuvring and navigation, 5G wireless sensing data may be combined with data from non-3GPP sensors such as camera, radar, LiDAR etc. to enrich or supplement a description of the sensing target. For example, in an Unmanned Aerial Vehicle (UAV) monitoring use case, in addition to 3 GPP sensing data such as UAV location, trajectory etc., the UAV’s colour, model and other information captured by cameras (i.e., non-3GPP data) may be used to enrich or supplement the monitoring results.

[0045] Scenarios of network-based and User Equipment (UE)-based radio sensing operations are depicted in Figures 2a and 2b and described in detail below. The depicted alternatives include scenarios of radio sensing where the network configures the participating sensing entities, i.e., network and UE nodes acting as sensing Tx nodes, network and UE nodes acting as sensing Rx nodes, as well as the configuration of sensing signal and necessary measurements and reporting procedures from the nodes. In this regard, the functional split between the network and the UE nodes for a specific sensing task (e.g., task of detecting presence of a pedestrian in a road) may take various forms, depending on the availability of sensing-capable devices and the requirements of the specific sensing task.

[0046] Figures 2a and 2b illustrate example configurations of nodes for sensing in a wireless communications system in accordance with aspects of the present disclosure. Figure 2a shows a first configuration 200 in which the network provides sensing Tx. Figure 2b shows a second configuration 205 in which a UE provides sensing Tx. The first configuration 200 and second configuration 205 may be different sensing modes with sensing signal transmission and sensing signal reception by the network, a UE, or a combination thereof.

[0047] The first configuration 200 comprises a gNB (sensing Rx) 222, a gNB (sensing Tx / Rx) 224 and a UE (sensing Rx) 210 arranged for sensing of an object 202. The second configuration 205 comprises a gNB (sensing Rx) 222, a UE (sensing Rx) 210 and a UE (sensing Tx / Rx) 212 arranged for sensing of the object 202.Docket No. SMM920250078-GR-NP

[0048] The first configuration 200 and second configuration 205 may be radio sensing scenarios. The first configuration 200 may comprise network-based radio sensing operations. The second configuration 205 may comprise UE-based radio sensing operations. In the first configuration 200 and second configuration 205, the network may configure the participating sensing entities, e.g., network and UE nodes acting as sensing Tx nodes, network and UE nodes acting as sensing Rx nodes. In the first configuration 200 and second configuration 205, the network may configure the sensing signal, measurements, and reporting procedures from the nodes. The functional split between the network and the UE nodes for a specific sensing task (e.g., task of detecting presence of a pedestrian in a road) may take various forms, depending on the availability of sensing-capable devices and the requirements of the specific sensing task. A node may comprise a participating sensing entity. A node may comprise a UE node. A node may comprise a network node. A node may comprise a sensing capable device.

[0049] In some examples described herein, a configuration of nodes for sensing in a wireless communication system comprises sensing Tx as a network node and sensing Rx as a separate network node: in this case, the sensing signal (sensing RS or another RS used for sensing, or the data / control channels known to the involved network / TRP nodes) is transmitted and received by different network entities. The involvement of UE nodes is limited to the aspects of interference management, when necessary. The network does not utilise UEs for sensing assistance in this scenario.

[0050] In some examples described herein, a configuration of nodes for sensing in a wireless communication system comprises sensing Tx as a network node and sensing Rx as the same network node: the sensing signal (e.g., sensing RS or another RS used for sensing, or the data / control channels known to the network / TRP nodes) is transmitted and received by the same network entity. The involvement of UE nodes is limited to the aspects of interference management, when necessary. The network does not utilise UEs for sensing assistance in this scenario.

[0051] In some examples described herein, a configuration of nodes for sensing in a wireless communication system comprises sensing Tx as network node and sensing Rx as a UE node: in this case, the sensing signal (e.g., sensing RS or other RS used for sensing or aDocket No. SMM920250078-GR-NPdata / control channel) is transmitted by a network entity and received by one or multiple UE nodes. The network configures the UEs to act as a sensing Rx node, according to the UE nodes capabilities for sensing, as well as the requirements of the desired sensing task.

[0052] In some examples described herein, a configuration of nodes for sensing in a wireless communication system comprises sensing Tx as a UE node and sensing Rx as a network node: in this case, the sensing signal (e.g., sensing RS or other RS used for sensing, or a data / control channel transmitted by the UE) is received by one or multiple network entities and transmitted by a UE node. The network configures the UE to act as a sensing Tx node, according to the UE nodes capabilities for sensing, as well as the nature of the desired sensing task.

[0053] In some examples described herein, a configuration of nodes for sensing in a wireless communication system comprises a sensing Tx as a UE node and sensing Rx as a separate UE node: in this case, the sensing signal (e.g., sensing RS or other RS used for sensing of a data / control channel) is received by one or multiple UE nodes and transmitted by a UE node. In this case, the network, or a UE node may potentially determine the configuration of the sensing scenario. The network may configure the UEs to act as a sensing Tx and / or sensing Rx nodes, according to the UE nodes capabilities for sensing, as well as the nature of the desired sensing task.

[0054] In some examples described herein, a configuration of nodes for sensing in a wireless communication system comprises a sensing Tx as a UE node and sensing Rx as the same UE node: in this case, the sensing signal (e.g., a sensing-dedicated RS or another RS used for sensing, or the data / control channels known to the UE) is transmitted by a UE node and received by the same UE node. The UE or the network may configure the sensing scenario, according to the UE nodes capabilities for sensing, as well as the nature of the desired sensing task.

[0055] Figures 3a to 3e illustrate examples of architectures for sensing in a wireless communications system in accordance with aspects of the present disclosure.

[0056] Network Architectures for supporting ISAC Sensing may relate to a target UE, or an object without network connectivity. For example, an object may comprise a UE with noDocket No. SMM920250078-GR-NPsim-card. IS AC sensing may be used for obtaining environment characteristics, e.g., sensing weather conditions, determining whether it is raining. ISAC sensing may use the radio signals from a plurality of nodes (e.g., one or more base stations); the plurality of node may be a sensing group. The locations of each node in the sensing group may be known. Sensing measurement data (e.g., sensing data) from each node of the sensing group may be collected synchronously. The collected sensing data is then provided to the mobile core network, which determines the sensing target and its corresponding characteristics. This may be configured before an actual sensing operation. Sensing data may initially be processed at the RAN level, e.g.., at the base station that acts as sensing Rx. This configuration tends to reduce the size of the sensing data and enable differentiation of target objects related to distinct sensing tasks.

[0057] ISAC may enhance 5G core architecture by introducing a Sensing Function (SF). Four proposals for enhancing the 5G core by introducing a SF as a dedicated or logical NF are considered in IMT-2020 as illustrated in Figures 3 a to 3 e.

[0058] Figure 3a illustrates an example of a first architecture 300 for sensing in a wireless communications system in accordance with aspects of the present disclosure. The first architecture 300 may represent a tight coupling ISAC network architecture.

[0059] The first architecture 300 comprises a UE 310, a (R)AN 320, UDM 330, NWDAF 335, LMF 340, PCF 345, AMF 350, SF 355, NEF 360, AF 365, and UPF 367. The first architecture 300 comprises the following reference points: an N1 reference point between the UE 310 and AMF 350, an N8 reference point between the AMF 350 and UDM 330, an NS3 reference point between the UDM 330 and SF 355, an NS1 reference point between the AMF 350 and SF 355, an NS4 reference point between the NWDAF 335 and SF 355, and NS6 reference point between the LMF 340 and SF 355, an NS5 reference point between the PCF 345 and SF 355, an NS2 reference point between the NEF 360 and SF 355, an NS7 reference point between the UPF 367 and SF 355, an N33 reference point between the NEF 360 and AF 365, an N8 reference point between the AMF 350 and UDM 330 and an N5 reference point between the PCF 345 and NEF 360.

[0060] The first architecture 300 may comprise a tight coupling ISAC network architecture. In the first architecture 300, the SF 355 appears as a dedicated NF handling both: (i) the sensing control plane aspects such as the interaction with the sensing consumerDocket No. SMM920250078-GR-NPvia NEF 360 and information exchange with other NFs, for gathering UE information, (e.g., from the AMF 350, UDM 330, and LMF 340, UE 310 related policies from the PCF 345, and analytics from the NWDAF 335 and (ii) the sensing radio signals for performing the analysis or prediction for determining the sensing target.

[0061] Figure 3 b illustrates an example of a second architecture 302 for sensing in a wireless communications system in accordance with aspects of the present disclosure. The second architecture 302 may represent an SB A of tight coupling IS AC network architecture.

[0062] The second architecture 302 comprises a UE 310, a (R)AN 320, UDM 330, NWDAF 335, LMF 340, PCF 345, AMF 350, SF 355, NEF 360, AF 365, and UPF 367. The second architecture 302 comprises the following reference points: an N1 reference point between the UE 310 and AMF 350, an N2 reference point between the AMF 350 and (R)AN 320, and an NS7 reference point between the UPF 367 and SF 355.

[0063] The second architecture 302 may be an alternative of the tight coupling IS AC network architecture of the first architecture 300. In the second architecture 302, the SF 355 directly interacts via NS7 with the UPF 367 to receive sensing data, while it interacts via the service-based interconnection medium with the remaining of the 5G core control plane NFs.

[0064] Figure 3 c illustrates an example of a third architecture 304 for sensing in a wireless communications system in accordance with aspects of the present disclosure. The third architecture 304 may represent a tight coupling IS AC network architecture with CP / UP split.

[0065] The third architecture 304 comprises a UE 310, a (R)AN 320, UDM 330, NWDAF 335, LMF 340, PCF 345, AMF 350, SF-Control plane (SF-C) 356, SF-User plane (SF-U) 357, NEF 360, AF 365, and UPF 367. The third architecture 304 comprises the following reference points: an N1 reference point between the UE 310 and AMF 350, an N8 reference point between the AMF 350 and UDM 330, an NS3 reference point between the UDM 330 and SF-C 356, an NS1 reference point between the AMF 350 and SF-C 356, an NS4 reference point between the NWDAF 335 and SF-C 356, and NS6 reference point between the LMF 340 and SF-C 356, an NS5 reference point between the PCF 345 and SF- C 356, an NS2 reference point between the NEF 360 and SF-C 356, an NS7 reference pointDocket No. SMM920250078-GR-NPbetween the UPF 367 and SF-U 357, an N33 reference point between the NEF 360 and AF 365, an N8 reference point between the AMF 350 and UDM 330 and an N5 reference point between the PCF 345 and NEF 360.

[0066] In third architecture 304 (e.g., a tight coupling ISAC network architecture with CP / UP split), the SF has two dedicated NF counter parts: (i) SF-C 356 that handles the control plane aspects as described above and (ii) SF-U 357 that is responsible for collecting the sensing radio signals via the user plane, e.g., via the (R)AN 320 and UPF 367. The third architecture 304 tends to enable splitting and offloading heavy data volumes associated with sensing radio signals to the user plane to ensure light traffic, e.g., only signalling, in the control plane.

[0067] Figure 3d illustrates an example of a fourth architecture 306 for sensing in a wireless communications system in accordance with aspects of the present disclosure. The fourth architecture 306 may represent a loose coupling ISAC network architecture.

[0068] The fourth architecture 306 comprises a UE 310, (R)AN 320, UDM 330, LMF 340, SF 355, AMF 350, GMLC 342, NEF 360 and AF 365. The fourth architecture 306 comprises the following reference points: an N1 reference point between the UE 310 and AMF 350, an NL1 reference point between the AMF 350 and the LMF 340 / SF 355, an NL7 reference point between the LMF 340 and SF 355, an N8 reference point between AMF 350 and the UDM 330, an NS 1 reference point between the AMF 350 and the NEF 360, an N2 reference point between the AMF 350 and the (R)AN 320, an NS2 reference point between the between the UDM 330 and the NEF 360, an NL5 reference point between the NEF 360 and GMLC 342, NL6 reference point between the UDM 330 and the GMLC 342 and an N33 reference point between the NEF 360 and the AF 365.

[0069] In the fourth architecture 306, the SF 355 is collocated with the LMF 340; for example, the SF 355 appears as a logical NF embedded in the LMF 340 to perform sensing taking advantage of the knowledge of a UE location.

[0070] Figure 3e illustrates an example of a fifth architecture 308 for sensing in a wireless communications system in accordance with aspects of the present disclosure.Docket No. SMM920250078-GR-NP

[0071] The fifth architecture 308 comprises a UE 310, a (R)AN 320, an AMF 350, an NWDAF 335, an SF 355, an NEF 360, and an AF 365. The fifth architecture 308 comprises the following reference points: an NS2 reference point between the AMF 350 and (R)AN 320, an NS2 reference point between the AMF 350 and SF 355, an NS1 reference point between the SF 355 and (R)AN 320, an NS4 reference point between the NWDAF 335 and SF 355, an NS 3 reference point between the SF 355 and NEF 360 and an N33 reference point between the NEF 360 and AF 365.

[0072] The fifth architecture 308 may represent a loose coupling ISAC network architecture where the SF 355 is independent of the 5G core, e.g., typically used for local field scenarios or private networks, and the interaction with the 5G core is minimal. The SF 355 may be close to the (R)AN 320. The (R)AN 320 may collect and process the sensing radio signals locally, and interact with 5G core for the purpose of exposure via NEF 360, for getting the UE location from the AMF 350 and for analytics (e.g., using an NWDAF).

[0073] For ISAC in 5G, ISAC operation may be based on the existing wireless infrastructure. The wireless infrastructure provides coverage to leverage the benefits of radio signal sensing as well as on the use of 5G core that can assist in collecting further information related to the UEs, policies, analytics and may facilitate sensing exposure towards external network consumers, e.g., AFs.

[0074] The processing of 3 GPP sensing data in order to derive a 3 GPP sensing result may be performed by an AI / ML model. In Al-based sensing, an AI / ML model may be trained using 3GPP sensing data collected from UEs and / or RANs (3GPP sensing entities). The present disclosure provides a 5G wireless sensing service that is Al -based, including an AI- based Sensing Function (AISF), that may comprise an inference part that may interact or be collocated with the SF and a training part that may be a separate entity, which can be shared by several inference counterparts.

[0075] Al model training for sensing may be deployed at the 3 GPP system - as a RAN NF, as a core network NF close to the RAN, or at the 0AM. The Al model training for sensing may be deployed at the core.Docket No. SMM920250078-GR-NP

[0076] Figure 4 illustrates a physical space or world 410, a 3GPP system 430, a simulation system 450 simulating the physical world 410 as a meta space 470, and a metaverse application 490. A UE 432 and a RAN 434 of the 3GPP system 430 are shown receiving sensing signals 436 from elements of the physical space 410. The 3 GPP system 430 is shown comprising an AISF 438. The metaverse application 490 is shown specifying to the 3 GPP system 430 sensing requirements 492. The simulation system 450 is shown comprising a Twin Controller (TC) 452 interfacing with the 3GPP system 430 for communication of sensing data 454 and aggregated sensing data 456. The TC 452 is also shown interfacing with a Twin Platform 458 that simulates the UE 432 as a simulated UE 460, the RAN 434 as a simulated RAN or RAN NDT 462, and generates the meta space 470. The Twin Platform 458 also simulates sensing signals 436 as simulated sensing signals 464. The TC 452 is shown communicating data 456 aggregated from the physical world and from the simulation and communicating it to the AISF 438. The AISF 438 is shown communicating an outcome / results 494 associated with the aggregated data 456 and / or sensing KPIs to the metaverse application 490.

[0077] Situational awareness in the smart transport metaverse may be a use case for the metaverse. In the smart transport scenario depicted in Figure 4, the physical world 410 may comprise traffic participants, such as vehicles 412, bicycle riders 414, pedestrians etc., road infrastructure such as sensors, cameras 416 and so on, environmental conditions such as weather and terrain, flora 420 such as trees, 3 GPP sensing entities such as the UEs 432 and RAN nodes 434. By means of sensing transmitters and receivers, the UEs 432 and RAN nodes 434 collect the raw 3GPP sensing data 436, which is sent to the AISF 438 at the core network.

[0078] The AI / ML model may be trained at the AISF 438 with the 3 GPP sensing data 436 and non-3GPP sensor data from the road infrastructure with the objective of training the AI / ML model to accurately derive the description and position of the physical elements sensed by the UEs 432 and RAN nodes 434. After training, the AISF 438 may use the trained AI / ML model, i.e., perform inference, to detect advanced object and perform localisation for the smart transport scenario. An outcome of the AISF 438 performing inference, which may include a determined accuracy of the description and position of an object at a given timeDocket No. SMM920250078-GR-NPinterval and may further include a confidence level, may be exposed or provided to a consumer such as the third party metaverse application 490 for further optimisation, e.g. for environment monitoring such as pedestrian intrusion on a highway.

[0079] Each of the physical elements or objects in the physical world 410 has a corresponding twin simulated at the meta space 470. The meta space 470 comprises Digital Twins (DTs) 472, 474, 476, 478, 480 of the traffic participants 414, 416, road infrastructure 416, environmental conditions 418, and flora 420, and corresponding DTs of UEs (or simulated UEs 460) and RANs (or RAN NDTs 462). The RAN NDTs 462 simulate the behaviour of a group of cells within a Tracking Area (TA). To realistically and accurately model the real world 410 in the meta space 470, the physical objects 412-420 and the sensing signals 436 are simulated by the DTs.

[0080] With the support of the 3GPP system 430, the meta space 470 and the physical world 410 may be periodically synchronised. In this synchronisation, the meta space 470 may collect one or more of the following data from the physical world: 3 GPP sensing data 436 and / or data collected from road infrastructure e.g. cameras and sensors. Several DTs can be combined or “nested” into a single Twin Platform 458, by layering or nesting different DT components together. The Twin Platform 458 generates simulation data 464 and uses this data to augment the data 436 of the physical world 410. This aggregated data 456 is consumed by the AISF 438 to support training. As individual components of the Twin Platform 458 may be hosted in different edge servers, the Twin Controller 452 may facilitate or control the interaction of the Twin Platform with the physical world 410.

[0081] Scenarios such as smart transport metaverse, and Al-based sensing, can benefit from each other. With reference to Figure 4, the simulation system 450, or metaverse, may synchronise with the sensing data 436 from the physical world 410 to update and maintain an accurate virtual representation of the real-world objects and the physical conditions. Additionally, the metaverse 450 may send the aggregated simulation data 456 to the 3 GPP system 430 for calibration or optimisation of the Al-based sensing operation.

[0082] The metaverse application 490 may initially provide a set of accuracy requirements for a given Al-based sensing task related to the smart transport scenario to the core network. The application 490 may also provide sensing assistance information that isDocket No. SMM920250078-GR-NPused to support the derivation of a sensing result. Examples of the sensing assistance information include map information, area information, UE position information, etc. One or more of these data may accompany or form part of sensing requirements 492 Based on operator policy, the AISF 438 may be configured to collect the 3 GPP sensing data 436 from the UEs 432 and / or RAN nodes 434. The AISF 438 may then perform inference and may determine an inference accuracy.

[0083] To fulfil any accuracy requirements that may be provided by the metaverse application 490, the AISF 438 may be trained using the Twin Platform 458 and / or based on data that is representative of the physical world 410. As the physical world 410 continuously changes, the physical environment context under which sensing results are generated by the AISF 438 also changes. For example, there may be a change in the weather conditions 418 (e.g., sunny to foggy), physical terrain 418 (e.g., lowland to highland), or even from outdoor to indoor i.e. change of context to automated guided vehicles in indoor factories. The updated physical context may also indicate an update in the physical environment of the UE associated to the sensing service area. To maximize the reuse of sensing data to train different AI / ML models and thus improve efficiency, the AISF 438 may be made aware of this change in physical environment context.

[0084] In addition, the Twin Platform 458 may generate simulations and corresponding simulation data. Since both the number and variety of issues occurring in an actual mobile network is limited, the training data generated by the Twin Platform may also be limited.

[0085] It may be difficult to determine if the degradation of inference accuracy of AISF 438 is due to a poorly trained model at the AISF 438, poor quality data, or a change in the physical environment context associated with the derived sensing results. Accommodating changes to the physical environment context of sensing tends to improve or ensure a level of inference accuracy related with sensing KPIs.

[0086] Figure 5 illustrates many features in common with Figure 4, where features in common are represented by numerals increased by 100. Differences are described below.Docket No. SMM920250078-GR-NP

[0087] As shown in Figure 5, the AISF 538 may comprise an Al-based Sensing Training Function (AISTF) 542 and an Al-based Sensing Learning Function (AISLF) 544. The AISF 538 may also comprise a Sensing Data Coordination and Storage Function (DCSF) 540.

[0088] A wireless sensing service that is Al-based may be provided. The service may be provided or partially provided by a AISF 538. The service may process 3GPP sensing data, such as sensing data 536, to generate 3GPP sensing results. The present disclosure relates to new mechanisms that assure or improve accuracy in Al -based sensing result extraction by, e.g., re-training an associated AI / ML model when needed.

[0089] The consumer of the sensing service, such as the metaverse application 590 (or a user thereof), upon identifying a drop in the accuracy of the sensing result, e.g., due to a change in physical environment context, may notify the AISF. This can then trigger the AISF to perform AI / ML model re-training, with the support of the Twin Platform, such as Twin Platform 558 of a simulation system 550, in the meta space. The Twin Platform may play two different roles. It can either be a part of a training entity that re-trains the respective AI / ML model for sensing, or it may provide additional data samples, e.g., 3 GPP sensing data, such as simulated sensing data 564, that may be used in the re-training process.

[0090] A metaverse (vertical) application, such as metaverse application 590, may send a request, such as request 592, to the core network of a 3GPP service, such as 3GPP service 530, related to a sensing service for a smart transport scenario. The application may provide a set of accuracy requirements related to sensing, that may include one or more of the following: required accuracy of positioning estimate; and confidence level, to the core network. 3GPP sensing entities, such as UEs 532 and / or Transmission Points (TRPs) such as RANs 534, collect 3GPP sensing data, such as sensing data 536, that is either pre-processed or sent as raw data to the AISF. The AISF may process the received 3GPP sensing data to extract the sensing result.

[0091] To handle inaccuracy in providing sensing results, the present disclosure may leverage the technical benefits of a Twin Platform such as Twin Platform 558. A Twin Platform at the meta space, that is a composition of several DTs and NDTs, such as simulated elements 572, 574, 576, 578, and 580, simulated UEs 560, simulated RANs 562, and / or simulated sensing data 564, representing the respective physical objects and networkDocket No. SMM920250078-GR-NPcomponents of the physical world, is used. Upon periodic data synchronisation between the physical world and meta space, the Twin Platform may simulate different networking scenarios for generating simulation data. The Twin Platform may also be used for sensing either as a part of a re-training tool or entity or to provision simulation data to complement physical world data for training when the AISF has insufficient data availability.

[0092] Considering that a physical environment context of sensing may change, the present disclosure tends to ensure a level of accuracy of sensing KPIs provided by the AI- based sensing function leveraging the benefits of the Twin Platform either as a part of training or for providing simulation data for training to the Al-based sensing function.

[0093] An AISF, such as AISF 538, may be split or divided into separate Al model training and inference capabilities.

[0094] The function or network entity responsible for Al model training may be an AISTF such as AISTF 542. The function or entity responsible for Al inference may be an AISLF such as AISLF 544. AISF 538 may be a single function or entity providing AISTF 542 and AISLF 544 services or may be composed of separate AISTF 542 and AISLF 544 functions. The AISF 538 can be deployed as one or several NFs.

[0095] The AISF 538 also may also include or may interact with one or more DCSFs 540. The DCSF 540 may store training data and corresponding metadata including the scenario and physical environment context under which the training data was or is to be generated or supplemented. The DCSF 540 may provide the functionality to collect and coordinate data related to sensing that is used for training and / or inference. The DCSF 540 may assure that the 3GPP sensing data 536 from TRPs 534 / UEs 532 for processing at the AISF 538 as well as data 556 from the Twin Platform 558 are commonly understood by the AISF 538 for training and inference. The AISF 538, that may include the AISLF 544, AISTF 542, and DCSF 540, may: be supplied with performance criteria relating to a given sensing task, target object(s), sensing requirements e.g. accuracy requirement of positioning estimate, confidence level and other sensing KPIs, and / or sensing assistance information, which may be provided by metaverse application 590 in a request 590 which may be referred to as an update request or as feedback; be informed from the sensing consumer, such as the metaverse application 590, that there is an accuracy related issue for a specific sensing result includingDocket No. SMM920250078-GR-NPalso details, e.g., sensing object was in a different position, or it was a different object, or something was detected by error, etc.; send a request, which may be referred to as a training request, to generate additional training data and information about the scenario, i.e., physical environment context and use case, under which the data generation is desired for the given sensing task; be supplied with additional training data and information about the scenario under which the data was generated for the given sensing task; be a part of or co-located with the Twin Platform 558; store 3GPP sensing data that can be used for training of an AI / ML model usable by the AISF 538 and information about the scenario under which the data was generated for the given sensing task; and / or translate 3GPP sensing data 536 from TRPs 534 / UEs 532 assuring common format and interpretation.

[0096] A Twin Platform may be a part of a sensing training entity or used to generate additional data, such as simulated sensing data 564, for retraining the AI / ML model at the AISTF 542. The Twin Platform 558 may comprise the following capabilities: providing a real-world and networking environment for training for AI / ML model using “what if scenarios” related to a specified sensing task; generating training data samples for the given sensing task; and performing corrective actions on the generated simulation data in one or more of the following ways: increasing the number of data samples; filling gaps in the dataset from or associated with the physical space 510; and reducing errors or noise in the existing dataset.

[0097] The present disclosure optimises a sensing service, e.g., for a smart transport metaverse use case, requested by the sensing consumer considering changes in the physical environment context. The consumer entity may be a core NF, application function (AF), another AISF or even a RAN NF. A RAN NF may consume the results of Al -based sensing for beam management.

[0098] The present disclosure may include a sensing function based on Al, e.g. an AISF, that consists of separate training (AISTF) and inference (AISLF) capabilities, as well as a function to store training data (DCSF). The Twin Platform may assist the training phase either by being a part of the AISTF or by providing simulation data for AI / ML model training.

[0099] The AISF may receive performance criteria related to a sensing task from a consumer request, such as request 592, and notify sensing results and KPIs, such as responseDocket No. SMM920250078-GR-NP594, for the given sensing task to the consumer. In case of an indicated inaccuracy from the consumer, the AISF 542 may send a request for training using the Twin Platform 558. The Twin Platform 558 is enhanced to participate in the training phase or to generate additional training data and / or perform corrective actions on the generated simulation data. The Twin Platform 558 may be a part of the 3 GPP core network, and may synchronise data with the physical space 510 periodically. The AISLF 544 and AISTF 542 may host multiple AI / ML models.

[0100] Based on operator policy, an 0AM may configure AISF 538 to collect 3GPP sensing data 536 for a given sensing task. The 0AM may configure the Twin Platform 558, via the Twin Controller 552, regarding the target sensing area and the corresponding DCSF 540 from which it shall collect data. The metaverse application or AF 590 (i.e. consumer) may subscribe to the sensing service and provide a set of accuracy requirements to the AISLF 544. The AISLF 544 may subscribe to or send a request to the AISTF 542 for AI / ML model provisioning, i.e. to get notifications when the AI / ML model has been re-trained by the AISTF 542 or to re-train an AI / ML model upon request.

[0101] For the given sensing task, 3 GPP sensing data and non-3GPP sensor data may be collected by the UEs 532 / TRPs 534 / sensors and pre-processed or sent to the AISLF 544. The AISF 538 may determine the AI / ML model for the sensing task and process the received data to generate 3GPP sensing results. The AISLF 544 may notify the sensing KPIs and 3GPP sensing results 536 to the consumer 590.

[0102] The consumer 590 may analyse the received 3GPP sensing results 536 and KPIs. If the consumer 590 identifies a drop in the sensing KPI(s) due to a change in the physical environment context, it may send feedback to the AISF 538, which may trigger AI / ML model replacement and / or re-training.

[0103] The AISLF may receive 592 the consumer request and decide if retraining the AI / ML model is required based on internal logic. Based on the observed values of the 3 GPP sensing data and their distribution, the AISF 538 may internally estimate the accuracy of the inference. For the instances where high-resolution non 3 GPP sensor data is available, e.g. from a camera, the AISF 538 may correlate the data to determine the accuracy. If thisDocket No. SMM920250078-GR-NPdeviation is below a pre-defined threshold, the AISLF 544 may trigger a re-training procedure.

[0104] In cases where re-training is required, the AISLF 544 may send a training request to the AISTF 542 which may include information related to the physical, i.e., real- world location of objects, and networking, e.g., base station configuration, environment conditions, sensing task ID (describing the sensing type, e.g., sensing for road safety or home security), description of target objects of interest, area of interest, and / or time schedule related to sensing where the AI / ML model was used. This information may either: (i) be used by the AISTF 542 to create “what if scenarios” using a Twin Platform 558 or (ii) be propagated by the AISTF 542 to the Twin Platform 558 to create a simulation environment for collecting and / or generating data samples for the given sensing task, AI / ML model. The generated data samples, which may be generated in the simulation environment or meta space 570, may be referred to as simulated training data 564.

[0105] The Twin Platform 558 may validate the received request and generate either updated weights of the AI / ML model if the Twin Platform 558 is co-located with the AISTF 542 and may be used to create “what if’ scenarios, or additional training data samples when the Twin Platform 558 is a separate entity that interacts with the AISTF 542.

[0106] The Twin Platform 558 may send 546 the generated AI / ML model weights and / or related information to the AISLF 544 or alternatively send the training data samples to the indicated AISTF 542. The AISTF 542 in turn may retrain and update the AI / ML model weights or related information using this additional data. The AISTF 542 may store this data in the DCSF 540 for future usage. The AISTF 542 may then provide the AISLF 544 with the updated AI / ML model parameters or notify that the updated parameters are available at the DCSF 540.

[0107] The AISLF 544 may perform inference based on the updated AI / ML model and provide 594 sensing KPIs to the consumer 590.

[0108] The above-described process may be run on-demand, i.e., when initiated by the consumer, or as a subscription, e.g., every few minutes, as determined by the AISLF 544.Docket No. SMM920250078-GR-NP

[0109] Figure 4 therefore optimises an Al-based sensing service, such as for a smart transport metaverse use case, considering changes in the physical environment context. The metaverse application may trigger sensing a re-training procedure for the AI / ML model used in the AISF 538 in the core network and the AISLF 544 may decide to request the AISTF 542 to provide retraining for the AI / ML model with respect to a specific sensing task, e.g., Sensing ID. The AISTF 542 may interact with the Twin Platform 558 of a simulation system 550 e.g., via the twin controller 552, providing information related to the physical and networking environment conditions, sensing task and target objects, area of interest, and / or time schedule related to sensing where the AI / ML model was used. The Twin Platform 558 may then generate and provide 556 AI / ML model weights, information, and / or additional 3 GPP sensing data 564 for training the AI / ML model used for sensing.

[0110] A request / subscription 554 sent to the Twin Controller 552 for training AI / ML model or generating training data samples for training an AI / ML model may include at least one of the following parameters: 1) Sensing task ID, 2) target object(s) to sense, 3) target sensing area, 4) networking conditions, e.g., type of base station involved, operating frequency, etc., 5) physical environment context information i.e. weather and terrain conditions, time of day, 6) requested time period for data generation, 7) minimum number of data samples to be generated, 7) AI / ML model ID for which additional data is requested, 8) format or type in which data is requested, and / or 9) AISTF or endpoint address.

[0111] A response 556 from the Twin Controller 552 to the AISTF 542 may include at least one of the following parameters: 1) Sensing task ID, 2) flag variable indicating the success or failure of training or for generating additional training data, 3) target sensing area, 3) data samples according to the requested data format or type, 4) time stamp of data generation, 5) physical environment context information, and / or 6) use case context for data storage purposes e.g. if the data generated can be trained for different AI / ML models in the same use case context (e.g. traffic monitoring and smart transport).

[0112] Figures 6 to 9 illustrate an example of a process flow 600 in accordance with aspects of the present disclosure. The process flow 600 may implement or be implemented by aspects of the wireless communication system 100. For example, the process flow 600 may include one or more objects 652 and one or more UEs, TRPs, and or sensors 654 in aDocket No. SMM920250078-GR-NPphysical space 650, a Twin Controller 658 of a Twin Platform 565, an AISLF 662, AISTF 664, and DCSF 666 of an AISF 660, an 0AM 668, and NEF 670, and an AF 672 (which may be a consumer), which may be one or more examples of devices described herein with reference to Figure 1.

[0113] The process flow 600 may be referred to as a procedure, including one or more operations performed by one or more of the objects 652, UEs, TRPs, and or sensors 654, the Twin Controller 658, the Twin Platform 565, the AISLF 662, the AISTF 664, the DCSF 666, the AISF 660, the 0AM 668, the NEF 670, and the AF 672.

[0114] Figures 6 to 9 illustrate both subscribe / notify and request / response models for optimising the Al-based sensing service described above. The request / response model involves the case when the consumer identifies a drop in inference accuracy and triggers a re-training procedure. The subscribe / notify model involves the case when the AISLF determines retraining of the AI / ML model is required for the sensing task. Figures 6 to 9 assume that the AF is untrusted, therefore an interaction with the NEF is shown, though this is to be understood as optional. Figures 6 to 9 assume that the Twin Platform is built and initialised in the meta space for the given Al-based sensing service, though this is also to be understood as optional. It is also assumed that there are an AISTF, AISLF and DCSF associated with a given sensing service or sensing task ID.

[0115] In the following description of the process flow 600, the operations or signalling performed between one or more of the objects 652, UEs, TRPs, and or sensors 654, the Twin Controller 658, the Twin Platform 565, the AISLF 662, the AISTF 664, the DCSF 666, the AISF 660, the 0AM 668, the NEF 670, and the AF 672 may be performed or signalled (e.g., transmitted, received) in a different order than the example order shown, or the operations or signalling performed by one or more of the objects 652, UEs, TRPs, and or sensors 654, the Twin Controller 658, the Twin Platform 565, the AISLF 662, the AISTF 664, the DCSF 666, the AISF 660, the 0AM 668, the NEF 670, and the AF 672 may be performed or signalled (e.g., transmitted, received) in different orders or at different times. Some operations or signalling may also be omitted from the process flow 600. Additionally, although some operations or signalling may be shown to occur at different times, these operations or signalling may occur at the same time or in overlapping time periods.Docket No. SMM920250078-GR-NP

[0116] The AISLF 662 may be referred to as a first network entity. The AISTF 664 may be referred to as a second network entity. The Twin Controller 658 may be referred to as a third network entity.

[0117] Figure 6 may be considered as illustrating preconditions associated with AI / ML model retraining described with reference to subsequent Figures. Referring to Figure 6, at 601, the 0AM 668 configures the Al-based sensing service and the respective AI / ML model at the AISLF 662, including the configuration of sensing transmitter(s) and receiver(s) as well as a respective sensing area of interest, for collecting 3GPP sensing data, processing it to create a 3 GPP sensing result. The Al-based sensing service carries out the sensing task, identified by a sensing task ID.

[0118] At 602, the 0AM 668 configures the Twin Platform 656, via the Twin Controller 658, regarding the network environment and target sensing area(s) and the corresponding DCSF 666, from which the Twin Platform 656 may collect data via the DCSF 666.

[0119] At 603, the consumer entity 672, which may be an AF, subscribes to the Al-based sensing service. The AF 672 may include in its subscription the application ID, target object(s) to sense, target sensing area, required accuracy of positioning estimate (in %) and confidence level (in %) of the corresponding target object(s), other sensing KPIs requirements, time period of reporting such KPIs and the sensing assistance information to support the derivation of the sensing results.

[0120] At 604, if the AF 672 is untrusted, the AF 672 interacts with the network via the NEF 670, which authorises and authenticates the request, controls contents of the request and selects the respective AISLF 662 to which forward the request. The NEF 670 maps an application ID from the AF request to a corresponding sensing task ID. The selection of AISLF 662 and mapping to sensing task ID may be based on pre-configured information in the NEF 670. The NEF 670 then forwards this request to the AISLF 662.

[0121] At 605 and 606, the Twin Platform 656 subscribes to the corresponding DCSF 666 for collecting 3 GPP sensing data and non-3GPP sensor data for a target sensing area. The DCSF 666 notifies the respective data to the Twin Controller 658. The notification includes 3GPP sensing data that is collected by the respective UEs / TRPs 654. This 3GPPDocket No. SMM920250078-GR-NPsensing data may include data derived from the reflections, refractions, and / or diffractions of 3 GPP radio signals by the target object(s) in the physical space. The non-3GPP data may include sensor data such as images, video feeds, and the like.

[0122] At 607, the AISLF 662 determines the AI / ML model that shall be used for inference of the specified sensing task ID, assuming that more than one AI / ML model is available for the same task with different requirements and / or for different sensing tasks.

[0123] At 608, the AISLF 662 requests or subscribes to AI / ML model provisioning services of the AISTF 664. As part of the model provisioning service, the AISTF 664 shall assist a request or notify the AISLF 662 (i.e. an authorised consumer) when an update of a specific AI / ML model is available. The AISLF 662 may include in its subscription the sensing task ID, the AI / ML model ID for which it requires the update and the Area of Interest (Aol) of the AISLF 662.

[0124] Figure 7 may be considered as illustrating sensing operations and KIP reporting preceding AI / ML model training described in subsequent Figures. With reference to Figure 7, at 609, the UEs / TRPs 654, by means of transmitting and receiving radio frequency signals, carry out the sensing operation on the target objects(s) 652. The estimation of parameters like signal strength, delay, doppler and angle spectrum information is obtained from scattered, reflected and / or refracted radio frequency signals used in the sensing transmitters and receivers. The non-3GPP sensors capture image, video feed etc. of the target object(s) in the physical space 650.

[0125] At 610, the 3 GPP sensing data and sensor data collected by the sensing entities 654 are sent to the AISF 660.

[0126] At 611, the AISF 660 processes the received 3GPP sensing data used as input to the AI / ML model for training and inference with specified Sensing task ID and AI / ML model ID. It executes training and inference using the AISTF 664 (when needed according to requests and / or subscriptions) and the AISLF 662 respectively and generates the 3GPP sensing results. The sensing results may also contain the inferred accuracy of the description and position of the target object(s) 650 and the confidence level(s) of such sensing results. For the derived sensing results, the AISLF 662 determines the sensing contextualDocket No. SMM920250078-GR-NPinformation, which includes the physical environment context under which the sensing results were derived. This sensing contextual information may include area information, time of capture, UE location and ID and / or physical environment context.

[0127] At 612, the AISLF 662 notifies the sensing results to the consumer 672. The AISLF 662 may include in the notification the sensing task ID, target object(s) sensed, the inferred accuracy of the description and position of the target object(s) (in %), the confidence level (in %), other sensing KPIs, and / or selected sensing contextual information.

[0128] Figures 8a and 8b may be considered as illustrating examples of the aforementioned retraining process and how it may be initiated or triggered. With reference to Figure 8a, at 613, the consumer 672 processes the exposed 3GPP sensing results from the AISLF 662 and may detect a drop or deviation in the expected performance, e.g., by a preconfigured threshold (e.g. below x%) due to a change in the physical environment context of sensing.

[0129] At 614 and 615, the consumer provides feedback from using the 3GPP sensing result (which may trigger AI / ML model re-training) to the AISLF 662. The consumer 672 may include in the feedback 614 the application ID, target object(s), target sensing area, a description of the change in the physical environment context of sensing, expected time validity of the new physical environment context, deviation of the observed and expected sensing results, time period of observation and / or an update in the service KPI requirements. If the consumer or AF 672 is untrusted, the feedback for Al-based sensing optimisation is sent via the NEF 670 as illustrated, though it is to be understood that if the consumer or AF 672 is trusted the feedback may be sent directly to the AISLF 662.

[0130] At 616, the AISLF 662 analyses the feedback 614 that it receives from the consumer 672 and decides whether to trigger retraining of the AI / ML model associated with the specified sensing task.

[0131] At 617, the AISLF 662 transmits what may be referred to as a training request to notify the AISTF 664 regarding the re-training of a specific AI / ML model. The AISLF 662 may include the sensing task ID, model ID, physical environment context information, sensing area of interest, description target objects, and / or time schedule, etc. This mayDocket No. SMM920250078-GR-NPinvolve the AISLF 662 and / or AISTF 664 communicating with the DCSF 666 to identify whether existing training data exists for retraining the model. Such existing training data may be referred to as pre-existing training data, i.e., training data already existing or available when retraining is determined to be needed or desired.

[0132] The AISTF 664 or DCSF 666 may notify, at 618, the AISLF 662 of existing training data.

[0133] At 619, the DCSF 666 may provide the existing training data to the AISTF 664 for retraining of the AI / ML model in accordance with the analysed feedback and notification.

[0134] With reference to Figure 8b, additionally or alternatively to steps 617 to 619, steps 620 to 623 may take place. Steps 620 to 623 may take place when it is determined, such as by the AISLF 662, that simulated training data is required. Simulated training data may be required where existing training data is not found, is incomplete, or otherwise determined to be inadequate for the retraining of the AI / ML model.

[0135] At 620, the AISTF 664 transmits an indication that simulated training data for the AI / ML model associated with the sensing task is required to utilise the Twin Platform 656 for re-training considering what-if scenarios or alternatively requests or subscribes to the Twin Platform 656, via the Twin Controller 658, for the generation of the simulated training data. The AISTF 664 may include in its indication or subscription information that facilitates the Twin Platform 656 generating appropriate data samples. This information may comprise the sensing task ID, target object(s), target sensing area, operating frequency, physical environment context information, requested time period for data generation, minimum number of data samples to be generated, an AI / ML model ID for which additional data is requested, and / or a format in which or type of data requested (that may indicate the expected compression or accuracy of the training data). The physical environment context information may include a description of the real-world (indoor or outdoor), weather and terrain conditions, and / or time of day. The AISTF 664 may also include LMF measurements in the physical environment context information.

[0136] At 621, the Twin Controller may validate the indication or subscription from the AISTF 664 and generates AI / ML model weights and / or the simulated training data based onDocket No. SMM920250078-GR-NPthe parameters specified by the AISTF 664. Insodoing, the Twin Platform 656 may create a meta space, virtual representation, or simulation of the real or physical world so that such simulated training data may be appropriately generated.

[0137] At 622, the Twin Controller 658 sends a response or notification to AISTF 664, possibly via DCSF 666, about the outcome of training either in term of updating the AI / ML model or providing data generation. The Twin Controller 658 may include in its response or notification the sensing task ID, model ID of target AI / ML model for which data is generated, and in case of error or unsuccessful outcome, a string variable indicating the cause of failure and the failure event.

[0138] At 623, the Twin Controller 658 sends a response to the AISTF 664, possibly via DCSF 666, including in its response or notification the AI / ML model updates or the simulated training data, and may include also the sensing task ID, target sensing area, time of generation, AI / ML model ID physical environment context information and / or use case context. The use case context refers to the context in which the generated data can be used for training an AI / ML model. The physical environment context information and use case context may be used as metadata for the generated data for storage and / or future training purposes.

[0139] At 624, the AISTF 664 uses existing training data and / or uses the simulated training data to retrain the AI / ML model, which may have been indicated by a model ID.

[0140] At 625, the AISTF 664 sends a notification to the AISLF 662 to indicate that a retrained AI / ML model for sensing is available. The AISTF 664 may include in its notification the sensing task ID and / or information about the retrained model. This information may include the model weight / parameters or other related information or an address from which the retrained model is available.

[0141] At 626, the AISLF 662 updates the AI / ML model, which may be indicated by a model ID, for inference, either directly upon receiving the updated model parameters from the AISTF 664 or by fetching the updated AI / ML model from the address.Docket No. SMM920250078-GR-NP

[0142] At 627, the sensing operation may repeat, wherein the AISLF 662 uses the updated AI / ML model to perform inference and reports the 3 GPP sensing results and KPIs according to steps 609 to 612 described above.

[0143] With reference to Figure 9, at 628, the AISLF 662 may analyse a distribution of sensing signals. Based on the observed values of transmissions, reflections and scattering of the wireless sensing signals and their distribution(s), the AISF 660 may internally estimate an accuracy of the inference. For instances when high-resolution non 3 GPP sensor data is available, e.g. from a high-resolution camera, the AISF 660 may remove this data at the instances of good estimates and measure the degradation of the accuracy. If this deviation is below a pre-defined threshold (e.g. y%), the AISLF 662 may decide to trigger Al-based sensing optimisation. This may take place before, after, and / or in addition to the retraining described above that is triggered by feedback from the consumer 672.

[0144] At 629, the AI / ML model re-training procedure may take place according to steps 617 to 627 described above.

[0145] Figure 10 illustrates an example of a processor 700 in accordance with aspects of the present disclosure. The processor 700 may be an example of a processor configured to perform various operations in accordance with examples as described herein. The processor 700 may include a controller 702 configured to perform various operations in accordance with examples as described herein. The processor 700 may optionally include at least one memory 704, which may be, for example, an L1 / L2 / L3 cache. Additionally, or alternatively, the processor 700 may optionally include one or more arithmetic-logic units (ALUs) 706. One or more of these components may be in electronic communication or otherwise coupled (e.g., operatively, communicatively, functionally, electronically, electrically) via one or more interfaces (e.g., buses).

[0146] The processor 700 may be a processor chipset and include a protocol stack (e.g., a software stack) executed by the processor chipset to perform various operations (e.g., receiving, obtaining, retrieving, transmitting, outputting, forwarding, storing, determining, identifying, accessing, writing, reading) in accordance with examples as described herein. The processor chipset may include one or more cores, one or more caches (e.g., memory local to or included in the processor chipset (e.g., the processor 700) or other memory (e.g.,Docket No. SMM920250078-GR-NPrandom access memory (RAM), read-only memory (ROM), dynamic RAM (DRAM), synchronous dynamic RAM (SDRAM), static RAM (SRAM), ferroelectric RAM (FeRAM), magnetic RAM (MRAM), resistive RAM (RRAM), flash memory, phase change memory (PCM), and others).

[0147] The controller 702 may be configured to manage and coordinate various operations (e.g., signalling, receiving, obtaining, retrieving, transmitting, outputting, forwarding, storing, determining, identifying, accessing, writing, reading) of the processor 700 to cause the processor 700 to support various operations in accordance with examples as described herein. For example, the controller 702 may operate as a control unit of the processor 700, generating control signals that manage the operation of various components of the processor 700. These control signals include enabling or disabling functional units, selecting data paths, initiating memory access, and coordinating timing of operations.

[0148] The controller 702 may be configured to fetch (e.g., obtain, retrieve, receive) instructions from the memory 704 and determine subsequent instruction(s) to be executed to cause the processor 700 to support various operations in accordance with examples as described herein. The controller 702 may be configured to track memory address of instructions associated with the memory 704. The controller 702 may be configured to decode instructions to determine the operation to be performed and the operands involved. For example, the controller 702 may be configured to interpret the instruction and determine control signals to be output to other components of the processor 700 to cause the processor 700 to support various operations in accordance with examples as described herein. Additionally, or alternatively, the controller 702 may be configured to manage flow of data within the processor 700. The controller 702 may be configured to control transfer of data between registers, arithmetic logic units (ALUs), and other functional units of the processor 700.

[0149] The memory 704 may include one or more caches (e.g., memory local to or included in the processor 700 or other memory, such RAM, ROM, DRAM, SDRAM, SRAM, MRAM, flash memory, etc. In some implementations, the memory 704 may reside within or on a processor chipset (e.g., local to the processor 700). In some other implementations, the memory 704 may reside external to the processor chipset (e.g., remote to the processor 700).Docket No. SMM920250078-GR-NP

[0150] The memory 704 may store computer-readable, computer-executable code including instructions that, when executed by the processor 700, cause the processor 700 to perform various functions described herein. The code may be stored in a non-transitory computer-readable medium such as system memory or another type of memory. The controller 702 and / or the processor 700 may be configured to execute computer-readable instructions stored in the memory 704 to cause the processor 700 to perform various functions. For example, the processor 700 and / or the controller 702 may be coupled with or to the memory 704, the processor 700, the controller 702, and the memory 704 may be configured to perform various functions described herein. In some examples, the processor 700 may include multiple processors and the memory 704 may include multiple memories. One or more of the multiple processors may be coupled with one or more of the multiple memories, which may, individually or collectively, be configured to perform various functions herein.

[0151] The one or more ALUs 706 may be configured to support various operations in accordance with examples as described herein. In some implementations, the one or more ALUs 706 may reside within or on a processor chipset (e.g., the processor 700). In some other implementations, the one or more ALUs 706 may reside external to the processor chipset (e.g., the processor 700). One or more ALUs 706 may perform one or more computations such as addition, subtraction, multiplication, and division on data. For example, one or more ALUs 706 may receive input operands and an operation code, which determines an operation to be executed. One or more ALUs 706 be configured with a variety of logical and arithmetic circuits, including adders, subtractors, shifters, and logic gates, to process and manipulate the data according to the operation. Additionally, or alternatively, the one or more ALUs 706 may support logical operations such as AND, OR, exclusive-OR (XOR), not-OR (NOR), and not-AND (NAND), enabling the one or more ALUs 706 to handle conditional operations, comparisons, and bitwise operations.

[0152] The processor 700 may support wireless communication in accordance with examples as disclosed herein. The processor 700 may be configured to or operable to support a means for determining whether retraining of an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is needed; transmitting, to a second networkDocket No. SMM920250078-GR-NPentity, a training request relating to the AI / ML model; and receiving, from the second network entity, a retrained AI / ML model retrained using at least one of pre-existing training data and / or simulated training data generated by a third network entity.

[0153] Alternatively, the processor 700 may be configured to or operable to support a means for receiving, from a first network entity, a training request relating to an artificial intelligence / machine learning, AI / ML, model associated with a sensing task or transmitting, to a third network entity, an indication that simulated training data for the AI / ML model is required and receiving, from the third network entity, the simulated training data for the AI / ML model; retraining the AI / ML model using pre-existing training data and / or the simulated training data; and transmitting, to the first network entity, a retrained AI / ML model retrained using at least one of the pre-existing training data and / or the simulated training data.

[0154] Alternatively, the processor 700 may be configured to or operable to support a means for receiving, from a second network entity, an indication that simulated training data for an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is required; and transmitting, to the second network entity, simulated training data for the AI / ML model.

[0155] Figure 11 illustrates an example of a NE 800 in accordance with aspects of the present disclosure. The NE 800 may include a processor 802, a memory 804, a controller 806, and a transceiver 808. The processor 802, the memory 804, the controller 806, or the transceiver 808, or various combinations thereof or various components thereof may be examples of means for performing various aspects of the present disclosure as described herein. These components may be coupled (e.g., operatively, communicatively, functionally, electronically, electrically) via one or more interfaces.

[0156] The processor 802, the memory 804, the controller 806, or the transceiver 808, or various combinations or components thereof may be implemented in hardware (e.g., circuitry). The hardware may include a processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), or other programmable logic device, or any combination thereof configured as or otherwise supporting a means for performing the functions described in the present disclosure.Docket No. SMM920250078-GR-NP

[0157] The processor 802 may include an intelligent hardware device (e.g., a general- purpose processor, a DSP, a CPU, an ASIC, an FPGA, or any combination thereof). In some implementations, the processor 802 may be configured to operate the memory 804. In some other implementations, the memory 804 may be integrated into the processor 802. The processor 802 may be configured to execute computer-readable instructions stored in the memory 804 to cause the NE 800 to perform various functions of the present disclosure.

[0158] The memory 804 may include volatile or non-volatile memory. The memory 804 may store computer-readable, computer-executable code including instructions when executed by the processor 802 cause the NE 800 to perform various functions described herein. The code may be stored in a non-transitory computer-readable medium such the memory 804 or another type of memory. Computer-readable media includes both non- transitory computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A non-transitory storage medium may be any available medium that may be accessed by a general-purpose or specialpurpose computer.

[0159] In some implementations, the processor 802 and the memory 804 coupled with the processor 802 may be configured to cause the NE 800 to perform one or more of the functions described herein (e.g., executing, by the processor 802, instructions stored in the memory 804). For example, the processor 802 may support wireless communication at the NE 800 in accordance with examples as disclosed herein. The NE 800 may be configured to support a means for determining whether retraining of an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is needed; transmitting, to a second network entity, a training request relating to the AI / ML model; and receiving, from the second network entity, a retrained AI / ML model retrained using at least one of pre-existing training data and / or simulated training data generated by the third network entity.

[0160] Alternatively, the NE 800 may be configured to or operable to support a means for receiving, from a first network entity, a training request relating to an artificial intelligence / machine learning, AI / ML, model associated with a sensing task or transmitting, to a third network entity, an indication that simulated training data for the AI / ML model is required and receiving, from the third network entity, the simulated training data for theDocket No. SMM920250078-GR-NPAI / ML model; retraining the AI / ML model using pre-existing training data and / or the simulated training data; and transmitting, to the first network entity, a retrained AI / ML model retrained using at least one of the pre-existing training data and / or the simulated training data.

[0161] Alternatively, the NE 800 may be configured to or operable to support a means for receiving, from a second network entity, an indication that simulated training data for an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is required; and transmitting, to the second network entity, simulated training data for the AI / ML model.

[0162] The controller 806 may manage input and output signals for the NE 800. The controller 806 may also manage peripherals not integrated into the NE 800. In some implementations, the controller 806 may utilise an operating system such as iOS®, ANDROID®, WINDOWS®, or other operating systems. In some implementations, the controller 806 may be implemented as part of the processor 802.

[0163] In some implementations, the NE 800 may include at least one transceiver 808. In some other implementations, the NE 800 may have more than one transceiver 808. The transceiver 808 may represent a wireless transceiver. The transceiver 808 may include one or more receiver chains 810, one or more transmitter chains 812, or a combination thereof.

[0164] A receiver chain 810 may be configured to receive signals (e.g., control information, data, packets) over a wireless medium. For example, the receiver chain 810 may include one or more antennas for receive the signal over the air or wireless medium. The receiver chain 810 may include at least one amplifier (e.g., a low-noise amplifier (LNA)) configured to amplify the received signal. The receiver chain 810 may include at least one demodulator configured to demodulate the receive signal and obtain the transmitted data by reversing the modulation technique applied during transmission of the signal. The receiver chain 810 may include at least one decoder for decoding the processing the demodulated signal to receive the transmitted data.

[0165] A transmitter chain 812 may be configured to generate and transmit signals (e.g., control information, data, packets). The transmitter chain 812 may include at least one modulator for modulating data onto a carrier signal, preparing the signal for transmissionDocket No. SMM920250078-GR-NPover a wireless medium. The at least one modulator may be configured to support one or more techniques such as amplitude modulation (AM), frequency modulation (FM), or digital modulation schemes like phase-shift keying (PSK) or quadrature amplitude modulation (QAM). The transmitter chain 812 may also include at least one power amplifier configured to amplify the modulated signal to an appropriate power level suitable for transmission over the wireless medium. The transmitter chain 812 may also include one or more antennas for transmitting the amplified signal into the air or wireless medium.

[0166] Figure 12 illustrates a flowchart of a method 900 in accordance with aspects of the present disclosure. The operations of the method 900 may be implemented by a NE as described herein. In some implementations, the NE may execute a set of instructions to control the function elements of the NE to perform the described functions.

[0167] At 902, the method 900 may include determining whether retraining of an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is needed. The operations of 902 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of 902 may be performed by a NE as described with reference to Figure 11.

[0168] At 904, the method 900 may include transmitting, to a second network entity, a training request relating to the AI / ML model. The operations of 904 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of 904 may be performed by a NE as described with reference to Figure 11.

[0169] At 906, the method 900 may include receiving, from the second network entity, a retrained AI / ML model retrained using at least one of pre-existing training data and / or simulated training data generated by a third network entity. The operations of 906 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of 906 may be performed by a NE as described with reference to Figure 11.

[0170] It should be noted that the method 900 described herein describes a possible implementation, and that the operations and the steps may be rearranged or otherwise modified and that other implementations are possible.Docket No. SMM920250078-GR-NP

[0171] Figure 13 illustrates a flowchart of a method 1000 in accordance with aspects of the present disclosure. The operations of the method 1000 may be implemented by a NE as described herein. In some implementations, the NE may execute a set of instructions to control the function elements of the NE to perform the described functions.

[0172] At 1002, the method 1000 may include receiving, from a first network entity, a training request relating to an artificial intelligence / machine learning, AI / ML, model associated with a sensing task or transmitting, to a third network entity, an indication that simulated training data for the AI / ML model is required and receiving, from the third network entity, the simulated training data for the AI / ML model. The operations of 1002 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of 1002 may be performed by a NE as described with reference to Figure 11.

[0173] At 1004, the method 1000 may include retraining the AI / ML model using preexisting training data and / or the simulated training data. The operations of 1004 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of 1004 may be performed by a NE as described with reference to Figure 11.

[0174] At 1006, the method 1000 may include transmitting, to the first network entity, a retrained AI / ML model retrained using at least one of the pre-existing training data and / or the simulated training data. The operations of 1006 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of 1006 may be performed a NE as described with reference to Figure 11.

[0175] It should be noted that the method 1000 described herein describes a possible implementation, and that the operations and the steps may be rearranged or otherwise modified and that other implementations are possible.

[0176] Figure 14 illustrates a flowchart of a method 1100 in accordance with aspects of the present disclosure. The operations of the method 1100 may be implemented by a NE as described herein. In some implementations, the NE may execute a set of instructions to control the function elements of the NE to perform the described functions.Docket No. SMM920250078-GR-NP

[0177] At 1102, the method 1100 may include receiving, from a second network entity, an indication that simulated training data for an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is required. The operations of 1102 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of 1102 may be performed by a NE as described with reference to Figure 11.

[0178] At 1104, the method 1100 may include transmitting, to the second network entity, simulated training data for the sensing task. The operations of 1104 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of 1104 may be performed by a NE as described with reference to Figure 11.

[0179] It should be noted that the method 1100 described herein describes a possible implementation, and that the operations and the steps may be rearranged or otherwise modified and that other implementations are possible.

[0180] In accordance with the present disclosure, there is provided a first network entity for wireless communication, comprising: at least one memory; and at least one processor coupled with the at least one memory and configured to cause the first network entity to: determine that retraining of an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is needed; transmit, to a second network entity, a training request relating to the AI / ML model; and receive, from the second network entity, a retrained AI / ML model retrained using at least one of pre-existing training data and / or simulated training data generated by a third network entity.

[0181] The first network entity so configured tends to provide a mechanism for reviewing and acting to improve accuracy in Al-based sensing result extraction by, for example, triggering a re-training of an associated AI / ML model when deemed required or desirable. Where training data is not available, a further advantage is provided by the obtaining of simulated training data, that is, training data that is generated in a virtual or simulation environment / by a simulation of real-world circumstances, so that the model can be retrained by appropriate data previously unavailable enabling different circumstances to be accommodated and accuracy of the model concordantly improved.Docket No. SMM920250078-GR-NP

[0182] Sensing may involve and / or refer to acquiring information about characteristics of the environment and / or objects within the environment by examining and / or analysing the effects of the said physical objects on electromagnetic signals such as radio signals.

[0183] A sensing task is a task in which sensing of one or more objects or quantities is required or desired and is context-dependent. For sensing, contextual information is definable as information which provides context to the conditions under which the sensing results were derived. For example, in an automotive context, a sensing task may involve detecting a presence of a pedestrian in a road. An AI / ML model for such a task may therefore be understood to employ inference to aid in detecting the pedestrian given input data from appropriate sources, e.g., from electromagnetic and / or acoustic sensors of a vehicle.

[0184] The at least one processor may be configured to cause the first network entity to determine that retraining is needed in response to receiving feedback relating to a drop in or deviation from an expected performance.

[0185] This allows a consumer (of the related sensing service) that identifies a drop in the accuracy of the sensing result, e.g., due to a change in physical environment context, to notify the first network device, thereby enabling consumer triggering of a model update. Further, this provides a quantitative basis for controlling when the model is to be updated.

[0186] The feedback may be received from the consumer via a fourth network entity. This may be configured to occur if the consumer is not trusted.

[0187] The at least one processor may be configured to cause the first network entity to analyse the feedback and to determine that the retraining is needed based on an outcome of the analysis.

[0188] The at least one processor may be configured to cause the first network entity to determine that the retraining is needed when the outcome of the analysis indicates that the drop in or deviation from the expected performance has passed a pre-determined threshold.

[0189] This provides a further or alternative quantitative basis for controlling when the model is to be updated.Docket No. SMM920250078-GR-NP

[0190] The training request may be transmitted in response to determining that a parameter associated with the AI / ML model is less than a threshold. The parameter may be received from the second network entity. The parameter may comprise, represent, or indicate a performance of the AI / ML model. The parameter may comprise, represent, or indicate the drop in or deviation from the expected performance

[0191] The at least one processor may be configured to cause the first network entity to estimate an accuracy of AI / ML model inference and to determine that the retraining is needed based on an outcome of the estimation.

[0192] The estimation may comprise analysing at least one of: a distribution of sensing signals; values of transmissions, reflections and / or scattering of wireless sensing signals; and / or and a distribution of values of transmissions, reflections and / or scattering of wireless sensing signals, or correlating sensing signals.

[0193] The at least one processor may be configured to cause the first network entity to update the retrained AI / ML model for inference.

[0194] At least one of the feedback and training request may comprise at least one of: a sensing task identifier that identifies the sensing task; a model identifier; an application identifier; one or more target object(s); a target sensing area; a description of a change in a physical environment context of sensing; an expected time validity of a new physical environment context; a deviation between an observed and an expected sensing result; a time period of observation; and / or an update in service KPI requirements.

[0195] The training request may comprise information related to at least one of: physical / real-world locations of objects; a base station configuration; one or more environment conditions; a description of a sensing type; a description of one or more target objects of interest; an area of interest; and / or a time schedule related to sensing in relation to which the AI / ML model was used.

[0196] The description of a sensing type may relate to sensing for road safety and / or home security.Docket No. SMM920250078-GR-NP

[0197] The receiving the retrained AI / ML model may comprise receiving, from the second network entity, a notification that the retrained AI / ML model is available.

[0198] The receiving the retrained AI / ML model may comprise transmitting a model request, to the second network entity, for the retrained AI / ML model.

[0199] The at least one processor is configured to cause the first network entity to transmit the training request via, or additionally to, a storage function configured to store training data.

[0200] The storage function may comprise a Data Coordination and Storage Function, DCSF.

[0201] In accordance with the present disclosure, there is provided a second network entity for wireless communication, comprising: at least one memory; and at least one processor coupled with the at least one memory and configured to cause the second network entity to: receive, from a first network entity, a training request relating to an artificial intelligence / machine learning, AI / ML, model associated with a sensing task; transmit, to a third network entity, an indication that simulated training data for the AI / ML model is required; receive, from the third network entity, the simulated training data for the AI / ML model; retrain the AI / ML model using pre-existing training data and / or the simulated training data; and transmit, to the first network entity, a retrained AI / ML model retrained using at least one of the pre-existing training data and / or the simulated training data.

[0202] A second network entity so configured provides support for the first entity in improving the accuracy of a retrained AI / ML model by obtaining simulated training data where required for retraining and performing the retraining according to the reception of training data and / or simulated training data.

[0203] The training request may include at least one of: a sensing task identifier that identifies the sensing task; a model identifier; an application identifier; one or more target object(s); a target sensing area; a description of a change in a physical environment context of sensing; an expected time validity of a new physical environment context; a deviation between an observed and an expected sensing result; a time period of observation; and / or an update in service key performance indicator, KPI, requirements.Docket No. SMM920250078-GR-NP

[0204] The training request may comprise information related to at least one of: physical / real-world locations of objects; a base station configuration; one or more environment conditions; a description of a sensing type, such as sensing for road safety or home security; a description of target objects of interest; an area of interest; and / or a time schedule related to sensing where the AI / ML model was used.

[0205] The training request may comprise information the third network entity is to use to generate data samples relating to the simulated training data. This may comprise generating data samples considering the updated physical environment context and / or other parameters included in the training request.

[0206] The information the third network entity is to use to generate data samples relating to the simulated training data may comprise one or more target object(s), a target sensing area, an operating frequency, physical environment context information, a requested time period for data generation, a minimum number of data samples to be generated, an AI / ML model identifier, ID, for which additional data is requested, a format or a type in which data is requested (which may indicate the expected compression or accuracy of the simulated training data).

[0207] The physical environment context information may include a description of the real- world (indoor or outdoor), weather and terrain conditions, and / or a time of day. The second request may include Location Management Function, LMF, measurements in the physical environment context information.

[0208] The transmitting of the retrained AI / ML model may comprise transmitting, to the first network entity, a notification that the retrained AI / ML model is available.

[0209] The transmitting of the retrained AI / ML model may comprise receiving, from the first network entity, a model request for the retrained AI / ML model.

[0210] The second network entity may receive the training request via, or additionally from, a storage function configured to store training data. The storage function may comprise a DCSF.Docket No. SMM920250078-GR-NP

[0211] In accordance with the present disclosure, there is provided a third network entity for wireless communication, comprising: at least one memory; and at least one processor coupled with the at least one memory and configured to cause the third network entity to: receive, from a second network entity, an indication that simulated training data for an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is required; and transmit, to the second network entity, simulated training data for the AI / ML model.

[0212] A third network entity so configured provides support for the first and second entities in improving the accuracy of an AI / ML model by providing simulated training data for retraining the model in circumstances where training data is not available.

[0213] Each of the first, second, and third network entities may contribute to the updating and / or retraining of the model and associated improvement of the model’s accuracy in circumstances where training data is or is not available. A system comprising all three network entities may provide an improved sensing service via the provision of improvable models for one or more sensing tasks.

[0214] The indication may include at least one of: a sensing task identifier that identifies the sensing task; a model identifier; an application identifier; one or more target object(s); a target sensing area; a description of a change in a physical environment context of sensing; an expected time validity of a new physical environment context; a deviation between an observed and an expected sensing result; a time period of observation; and / or an update in service KPI requirements.

[0215] The third network entity may be configured to simulate an environment corresponding to the sensing task and generate simulated training data based on the simulation.

[0216] The third network entity may be configured to simulate the environment based on the indication.

[0217] In accordance with the present disclosure, there is provided a method for wireless communication performed by a first network entity, the method comprising: determining whether retraining of an artificial intelligence / machine learning, AI / ML, model associatedDocket No. SMM920250078-GR-NPwith a sensing task is needed; transmitting, to a second network entity, a training request relating to the AI / ML model; and receiving, from the second network entity, a retrained AI / ML model retrained using at least one of pre-existing training data and / or simulated training data generated by a third network entity.

[0218] In accordance with the present disclosure, there is provided a method for wireless communication performed by a second network entity, the method comprising: receiving, from a first network entity, a training request relating to an artificial intelligence / machine learning, AI / ML, model associated with a sensing task or transmitting, to a third network entity, an indication that simulated training data for the AI / ML model is required and receiving, from the third network entity, the simulated training data for the AI / ML model; retraining the AI / ML model using pre-existing training data and / or the simulated training data; and transmitting, to the first network entity, a retrained AI / ML model retrained using at least one of the pre-existing training data and / or the simulated training data.

[0219] In accordance with the present disclosure, there is provided a method for wireless communication performed by a third network entity, the method comprising: receiving, from a second network entity, an indication that simulated training data for an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is required; and transmitting, to the second network entity, simulated training data for the AI / ML model.

[0220] The first network entity may be in a 3rd Generation Partnership Project, 3 GPP, network. The first network entity may comprise an Al-based Sensing Logical Function, AISLF. The second network entity may be a part of the 3GPP network or may be part of a twin platform. The second network entity may comprise an Al-based Sensing Training Function, AISTF. The third network entity may comprise a twin controller, TC. The TC may be a part of the twin platform. The fourth network entity may comprise a Network Exposure Function, NEF. The consumer may comprise an Application Function, AF. The storage function, which may comprise a Data Coordination and Storage Function, DCSF, may be a part of the 3 GPP network.

[0221] A Twin Platform simulates and provides AI / ML model parameters and / or training data to an Al-based sensing function (AISF). It may be desirable to determine if the degradation of sensing results is due to a poorly trained model at the AISF, caused by eitherDocket No. SMM920250078-GR-NPpoor quality data or due to a change in the physical environment context. A problem may be how to ensure a level of sensing result accuracy and associated KPIs by the AISF, given that the physical environment context of sensing may change.

[0222] The present disclosure tends to enable a consumer of sensing to trigger re-training based on the drop in sensing result accuracy due to a change in physical environment context. The present disclosure further tends to enable the AISF to perform or trigger AI / ML model retraining based upon the change in physical environment context of sensing, potentially with the support of the Twin Platform in the meta space that provides training for the AI / ML model or additional data samples for the retraining, thereby optimising the Al -based sensing service.

[0223] The present disclosure follows both subscribe / notify and request / response models: to perform re-training using the Twin Platform for creating “what if’ scenarios; and to provide missing sensing data to a sensing training entity that is responsible to perform retraining.

[0224] In accordance with the present disclosure, there is provided a first network entity (such as an AISLF) for wireless communication, comprising: at least one memory; and at least one processor coupled with the at least one memory and configured to cause the network entity to: receive a trigger from a consumer indicating the need for optimising the AI / ML model used for specific sensing task identifier, transmit a request for AI / ML training related to a given sensing task identifier, receive an outcome of AI / ML update for the given sensing task identifier.

[0225] In accordance with the present disclosure, there is provided a second network entity (such as an AISTF) for wireless communication, comprising: at least one memory; and at least one processor coupled with the at least one memory and configured to cause the network entity to: receive a request for AI / ML training related to a given sensing task identifier and / or AI / ML model ID, transmit information related to a sensing task identifier and / or trained AI / ML model ID to a first network entity.

[0226] In accordance with the present disclosure, there is provided a third network entity (such as a Twin Controller) for wireless communication, comprising: at least one memory;Docket No. SMM920250078-GR-NPand at least one processor coupled with the at least one memory and configured to cause the network entity to: receive a request for training an AI / ML model ID from a second network entity generating training data for a given AI / ML model ID related to a sensing task ID, considering the updated physical context information and request parameters, from a second network entity, transmit an outcome of AI / ML model training information to a second network entity.

[0227] The trigger for optimising the AI / ML model related to a sensing service may include at least one of the following parameters: an identifier related to the given sensing task ID, an information related to a list of target objects to be sensed; an information related to the target sensing area; a time information indicating when a result was used; an information indicating the deviation of observed and expected sensing results; a time information during which the said deviation is observed; and / or an information indicating an update in the service KPI requirements.

[0228] The information related to the physical context of sensing may include at least one of the following parameters: an information related to an update in the terrain density or statistics within the target sensing area; an information related to an update in the physical environment of the UE associated to the sensing service area; an information related to an update in the weather conditions in the target sensing area; and / or an information indicating the time of day when the said update in the physical context of sensing is observed.

[0229] The request for training an AI / ML model or generating AI / ML training data related to a given sensing task ID and AI / ML model ID may include at least one of the following parameters: a sensing task identifier related to the sensing service; an information related to a list of target objects to be sensed; an information related to the target sensing area; an information related to the physical context of sensing, an information related to the operating frequency of the sensing service; a time information indicating the period for which the AI / ML training data samples are to be generated; an information related to the minimum number of data samples to be generated; an information indicating the format or type in which the data samples are to be generated; an information indicating the address of a second network entity to which the data samples are to be transmitted; and / or an identifier of the target AI / ML model for which the data samples are to be generated.Docket No. SMM920250078-GR-NP

[0230] The outcome of AI / ML training data generation related to a given sensing service may include at least one of the following parameters: a sensing task identifier related to the sensing service; an identifier of the target AI / ML model for which the data samples are generated; an identifier indicating the success or failure of the training data generation; an information indicating the cause of failure of the training data generation, in case a failure event occurs; an information related to the data samples that have been generated; an information related to the KPI (refresh rate) of the training data generated; a time information indicating the time during which the said data samples were generated; an information related to the physical context of sensing; and / or an information indicating the use case context where the generated data samples are applicable.

[0231] In accordance with the present disclosure, there is provided a consumer entity (such as an AF) for wireless communication, comprising: at least one memory; and at least one processor coupled with the at least one memory and configured to cause the network entity to: transmit a trigger for sensing optimisation related to a sensing service towards a network entity.

[0232] In accordance with the present disclosure, there is provided a method in where the process for optimising a sensing task ID involves the interaction among a consumer and a sensing inference function, a sensing inference function and a sensing training function, and a sensing training function and a Digital Twin Platform.

[0233] It should be noted that the method described herein describes a possible implementation, and that the operations and the steps may be rearranged or otherwise modified and that other implementations are possible.

[0234] The description herein is provided to enable a person having ordinary skill in the art to make or use the disclosure. Various modifications to the disclosure will be apparent to a person having ordinary skill in the art, and the generic principles defined herein may be applied to other variations without departing from the scope of the disclosure. Thus, the disclosure is not limited to the examples and designs described herein but is to be accorded the broadest scope consistent with the principles and novel features disclosed herein.Docket No. SMM920250078-GR-NP

Claims

CLAIMS1. A first network entity for wireless communication, comprising: at least one memory; and at least one processor coupled with the at least one memory and configured to cause the first network entity to: determine that retraining of an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is needed; transmit, to a second network entity, a training request relating to the AI / ML model; receive, from the second network entity, a retrained AI / ML model retrained using at least one of pre-existing training data and / or simulated training data generated by a third network entity.

2. The first network entity according to claim 1, wherein the at least one processor is configured to cause the first network entity to determine that the retraining is needed in response to receiving feedback relating to a drop in or deviation from an expected performance.

3. The first network entity according to claim 2, wherein the at least one processor is configured to cause the first network entity to analyse the feedback and to determine that the retraining is needed based on an outcome of the analysis.

4. The first network entity according to claim 3, wherein the at least one processor is configured to cause the first network entity to determine that the retraining is needed when the outcome of the analysis indicates that the drop in or deviation from the expected performance has passed a pre-determined threshold.

5. The first network entity according to claim 1, wherein the at least one processor is configured to cause the first network entity to estimate an accuracy of AI / ML model inference and to determine that the retraining is needed based on an outcome of the estimation.Docket No. SMM920250078-GR-NP6. The first network entity according to claim 5, wherein the estimation comprises analysing at least one of: a distribution of sensing signals; values of transmissions, reflections and / or scattering of wireless sensing signals; and / or and a distribution of values of transmissions, reflections and / or scattering of wireless sensing signals, or correlating sensing signals.

7. The first network entity according to any preceding claim, wherein the at least one processor is configured to cause the first network entity to update the retrained AI / ML model for inference.

8. The first network entity according to any preceding claim, wherein at least one of the feedback and training request comprises at least one of: a sensing task identifier that identifies the sensing task; a model identifier; an application identifier; one or more target object(s); a target sensing area; a description of a change in a physical environment context of sensing; an expected time validity of a new physical environment context; a deviation between an observed and an expected sensing result; a time period of observation; and / or an update in service KPI requirements.

9. The first network entity according to any preceding claim, wherein the training request comprises information related to at least one of: physical / real-world locations of objects; a base station configuration; one or more environment conditions; a description of a sensing type; a description of one or more target objects of interest; an area of interest; and / or a time schedule related to sensing in relation to which the AI / ML model was used.

10. The first network entity according to any preceding claim, wherein the at least one processor is configured to cause the first network entity to transmit the training request via, or additionally to, a storage function configured to store training data.

11. A second network entity for wireless communication, comprising: at least one memory; andDocket No. SMM920250078-GR-NPat least one processor coupled with the at least one memory and configured to cause the second network entity to: receive, from a first network entity, a training request relating to an artificial intelligence / machine learning, AI / ML, model associated with a sensing task; transmit, to a third network entity, an indication that simulated training data for the AI / ML model is required; receive, from the third network entity, the simulated training data for the AI / ML model; retrain the AI / ML model using pre-existing training data and / or the simulated training data; and transmit, to the first network entity, a retrained AI / ML model retrained using at least one of the pre-existing training data and / or the simulated training data.

12. A third network entity for wireless communication, comprising: at least one memory; and at least one processor coupled with the at least one memory and configured to cause the third network entity to: receive, from a second network entity, an indication that simulated training data for an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is required; and transmit, to the second network entity, simulated training data for the AI / ML model.

13. A method for wireless communication performed by a first network entity, the method comprising: determining that retraining of an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is needed; transmitting, to a second network entity, a training request relating to the AI / ML model; receiving, from the second network entity, a retrained AI / ML model retrained using at least one of pre-existing training data and / or simulated training data generated by the third network entity.Docket No. SMM920250078-GR-NP14. A method for wireless communication performed by a second network entity, the method comprising: receiving, from a first network entity, a training request relating to an artificial intelligence / machine learning, AI / ML, model associated with a sensing task or transmitting, to a third network entity, an indication that simulated training data for the AI / ML model is required and receiving, from the third network entity, the simulated training data for the AI / ML model; retraining the AI / ML model using pre-existing training data and / or the simulated training data; and transmitting, to the first network entity, a retrained AI / ML model retrained using at least one of the pre-existing training data and / or the simulated training data.

15. A method for wireless communication performed by a third network entity, the method comprising: receiving, from a second network entity, an indication that simulated training data for an artificial intelligence / machine learning, AI / ML, model associated with a sensing task is required; and transmitting, to the second network entity, simulated training data for the AI / ML model.Docket No. SMM920250078-GR-NP

Images