Method and apparatus of supporting artificial intelligence (AI) applications in wireless communications
A service-based interface facilitates direct communication between RAN and core network functions for AI/ML model training and deployment, addressing the integration challenge and enhancing RAN performance through optimized AI/ML use cases.
Patent Information
- Application Number
- PCT/CN2024/135510
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2024-11-29
- Publication Date
- 2025-10-02
AI Technical Summary
Current wireless communication systems lack efficient mechanisms for integrating artificial intelligence (AI) and machine learning (ML) models at the radio access network (RAN) level, limiting the optimization of network performance and functionality.
A service-based interface is introduced to enable direct communication between RAN entities and core network functions (NFs) for training, deploying, and managing AI/ML models, allowing RAN entities to request and deploy AI/ML models trained by NFs, such as a Network Data Analytics Function (NWDAF), for various use cases including CSI compression, beam prediction, and resource management.
Enhances RAN performance by enabling optimized AI/ML model deployment and management, improving CSI prediction, beam management, and resource allocation, thereby enhancing network efficiency and user experience.
Smart Images

Figure CN2024135510_02102025_PF_FP_ABST
Abstract
Description
METHOD AND APPARATUS OF SUPPORTING ARTIFICIAL INTELLIGENCE (AI) APPLICATIONS IN WIRELESS COMMUNICATIONSTECHNICAL FIELD
[0001] The present disclosure relates to wireless communications, and more specifically to techniques of supporting artificial intelligence (AI) applications in wireless communications.BACKGROUND
[0002] A wireless communications system may include one or multiple network communication devices, such as base stations, which may support 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 utilizing 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, the 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] Some implementations of the methods and apparatuses described herein may further include a radio access network (RAN) entity for wireless communication, which may include: at least one memory; and at least one processor coupled with the at least one memory and configured to cause the RAN entity to: send, to a first network function (NF) , a model request related message of requesting at least one AI model trained by the first NF; receive, from the first NF, a model related message including a file of the at least one AI model or information related to fetching the file of the at least one AI model; and deploy the at least one AI model for RAN side use cases based on the model related message.
[0005] In some implementations of the methods and apparatuses described herein, the model request related message indicates one or multiple of the following: model information; model identifier (ID) ; model accuracy check flag; available data requirement for model training for RAN side use case; model use case context; input data information to be used for training models; or output data information to be used for training models.
[0006] In some implementations of the methods and apparatuses described herein, the model related message further indicates one or multiple of the following: model accuracy; input data information to be used for training models; or output data information to be used for training models.
[0007] In some implementations of the methods and apparatuses described herein, in the case of receiving the model related message including information related to fetching the file of the at least one AI model, the at least one processor is further configured to cause the RAN entity to: retrieve the file of the at least one AI model based on the information related to fetching the file of the at least one AI model.
[0008] In some implementations of the methods and apparatuses described herein, the at least one processor is further configured to cause the RAN entity to: receive, from the first NF, a data request related message of requesting RAN side data, including information related to needed data; and send, to the first NF, the RAN side data or information related to fetching the RAN side data based on the data request related message.
[0009] In some implementations of the methods and apparatuses described herein, in the case of sending the information related to fetching the RAN side data, the at least one processor is further configured to cause the RAN entity to: receive, from the first NF, a data fetching related message of fetching RAN side data; and send, to the first NF, the RAN side data based on the data fetching related message.
[0010] In some implementations of the methods and apparatuses described herein, the at least one processor is further configured to cause the RAN entity to: receive, from a second NF, a data request related message of requesting RAN side data, including information related to needed data; and send, to the second NF, the RAN side data or information related to fetching the RAN side data based on the data request related message.
[0011] In some implementations of the methods and apparatuses described herein, in the case of sending the information related to fetching the RAN side data, the at least one processor is further configured to cause the RAN entity to: receive, from the second NF, a data fetching related message of fetching RAN side data; and send, to the second NF, the RAN side data based on the data fetching related message.
[0012] In some implementations of the methods and apparatuses described herein, the at least one processor is further configured to cause the RAN entity to: request storing RAN side data in a second NF by invoking a data storage request.
[0013] In some implementations of the methods and apparatuses described herein, the at least one processor is further configured to cause the RAN entity to: monitor the at least one AI model deployed at the RAN side; and report monitoring results of the at least one AI model to the first NF.
[0014] In some implementations of the methods and apparatuses described herein, before reporting the monitoring results of the at least one AI model, the at least one processor is further configured to cause the RAN entity to: send, to the first NF, a monitoring register related message of indicating that monitoring the at least one AI model deployed at the RAN side has started; and receive, from the first NF, a result request related message of requesting the monitoring results of the at least one AI model.
[0015] In some implementations of the methods and apparatuses described herein, the monitoring register related message indicates an ID of the at least one AI model being monitored at the RAN side.
[0016] In some implementations of the methods and apparatuses described herein, the result request related message indicates one or multiple of the following: an ID of the at least one AI model being monitored, analytics IDs, accuracy metrics to be monitored, reporting thresholds or reporting periods.
[0017] In some implementations of the methods and apparatuses described herein, the at least one processor is further configured to cause the RAN entity to: determine whether to report the monitoring results of the at least one AI model to the first NF based on the reporting thresholds or reporting periods.
[0018] Some implementations of the methods and apparatuses described herein may further include a NF for wireless communication, which may include: at least one memory; and at least one processor coupled with the at least one memory and configured to cause the NF to: receive, from a RAN entity, a model request related message of requesting at least one AI model trained by the NF; and send, to the RAN entity, a model related message including a file of at least one AI model or information related to fetching the file of the at least one AI model.
[0019] In some implementations of the methods and apparatuses described herein, the NF is a core network (CN) NF containing model training logical function (MTLF) , or an entity or node dedicated for RAN domain computing and analytics tasks.
[0020] In some implementations of the methods and apparatuses described herein, before sending the model related message, the at least one processor is configured to further cause the NF to: train the at least one AI model based on RAN side data received from the RAN entity or a different NF.
[0021] In some implementations of the methods and apparatuses described herein, before sending the model related message, the at least one processor is configured to further cause the NF to: send, to the RAN entity or the different NF, a data request related message of requesting RAN side data, including information related to needed data; and receive, from the RAN entity or the different NF, the RAN side data or information related to fetching the RAN side data.
[0022] In some implementations of the methods and apparatuses described herein, in the case of receiving the information related to fetching the RAN side data, the at least one processor is further configured to cause the NF to: send, to the RAN entity or the different NF, a data fetching related message of fetching RAN side data; and retrieve, from the RAN entity or the different NF, the RAN side data.
[0023] Some implementations of the methods and apparatuses described herein may further include a method performed by a RAN entity, which may include: sending, to a first NF, a model request related message of requesting at least one AI model trained by the first NF;receiving, from the first NF, a model related message including a file of at least one AI model or information related to fetching the file of the at least one AI model; and deploying the at least one AI model for RAN side use cases based on the model related message.
[0024] Some implementations of the methods and apparatuses described herein may further include a method performed by a NF, which may include: receiving, from a RAN entity, a model request related message of requesting at least one AI model trained by the NF; and sending, to the RAN entity, a model related message including a file of the at least one AI model or information related to fetching the file of the at least one AI model.BRIEF DESCRIPTION OF THE DRAWINGS
[0025] Figure 1 illustrates an example of a wireless communications system in accordance with aspects of the present disclosure.
[0026] Figure 2 illustrates an example of a service based interface that offering RAN services to other NFs under 6G or higher generation in accordance aspects of the present disclosure.
[0027] Figure 3 depicts the different communication models for NF / NF service interaction.
[0028] Figure 4 illustrates an example of a procedure for a RAN entity to request AI / ML models from a first NF via the service based interface in accordance aspects of the present disclosure.
[0029] Figure 5 illustrates an example of a wireless communication apparatus in accordance with aspects of the present disclosure.
[0030] Figure 6 illustrates a flowchart of method performed by a NE in accordance with aspects of the present disclosure.
[0031] Figure 7 illustrates a flowchart of method performed by a NF in accordance with aspects of the present disclosure.DETAILED DESCRIPTION
[0032] AI, at least including machine learning (ML) is used to learn and perform certain tasks via training neural networks (NNs) with vast amounts of data, which is successfully applied in computer vison (CV) and nature language processing (NLP) areas. Deep learning, which is a subordinate concept of ML, utilizes multi-layered NNs as an “AI / ML model” (or referred to as AI / ML model or the like) or "AI-based model" (or referred to as AI / ML based model or the like) to learn how to solve problems and / or optimize performance from vast amounts of data. If AI / ML models used on AI-based methods are well trained, the AI-based methods can obtain better performance than the traditional methods. Thus, 3rd generation partnership program (3GPP) has been considering to introduce AI / ML into 3GPP since 2016. For example, 3GPP expects that AI / ML models used by or located at the network side (also referred to as network sided models or RAN side models or the like) may be used to realize AI / ML based beam prediction or channel state information (CSI) prediction or CSI compression to optimize the network performance etc.
[0033] RAN side AI / ML models may be trained by various manners. For example, in 5G, RAN side AI / ML models may be trained by a NE or RAN node (e.g., gNB) itself, or by an operations administration and maintenance (OAM) system. Besides those, various aspects of the present disclosure propose that RAN side AI / ML models may be trained by a NF via a service based interface, e.g., in the case of a RAN node or entity lacking computing resources. The service based interface allows end-to-end communications (which means end-to-end from protocol point of view) between a RAN entity and another NF (hereinafter first NF) , which may be a CN NF or the like, or another RAN domain entity or node or the like.
[0034] For example, in accordance with various aspects of the present disclosure, a RAN entity, e.g., a gNB or a center unit (CU) of a gNB or the like may send a model request related message to a first NF via a service based interface. The model request related message requests AI / ML model (s) trained by the first NF, and may indicate one or multiple of the following parameters or information: model information, model ID; model accuracy check flag; available data requirement for model training for RAN side use case; model use case context; input data information to be used for training models; or output data information to be used for training models etc. An example of the first NF may be a CN NF containing MTLF, e.g., a CN network data analytics function (NWDAF) , or an entity or node dedicated for RAN domain computing and analytics tasks, e.g., a RAN domain NWDAF.
[0035] Based on the received model request related message, the first NF may send a model related message to the RAN entity via the service based interface, including file (s) of AI / ML model (s) (each file corresponding to an AI / ML model) or information related to fetching the file (s) of the AI / ML model (s) . In some cases, before sending the model related message, the first NF may need to collect data, e.g., directly or indirectly from RAN side to train the AI / ML model (s) as requested.
[0036] At the RAN entity side, after receiving the model related message, which indicates file (s) of AI / ML model (s) or information related to fetching the file (s) of the AI / ML model (s) , the RAN entity may obtain or determine the AI models. For example, the RAN entity may direct obtain the file (s) of the requested AI / ML model (s) as indicated in the model related message; or retrieve or fetch the file (s) of the requested AI model (s) based on the information related to fetching the files of the AI / ML models, e.g., addresses of the files of the AI / ML models. Then, the RAN entity may deploy the obtained AI / ML models at RAN side for RAN side use cases, e.g., using the AI / ML functionalities (including one or more models) or models for predictions or network decisions for RAN side use cases.
[0037] For example, the RAN entity may use the RAN side AI / ML functionalities or models to optimize the RAN side performance considering but not limited to the following use cases: - CSI compression: ■ A pair of AI / ML model can be used by UE and next generation (NG) -RAN node to compress the CSI report. The UE part model works as an encoder to compress the generated CSI report. The NG-RAN part model works as a decoder to decompress the received CSI report after compression. - CSI prediction in temporal domain: ■ NG-RAN node can predict the future channel state based on historical CSI measurement result. - Beam prediction in spatial domain: ■ NG-RAN node can predict the layer (L) 1 quality of set A beams (or a set of beams to be predicted) based on L1 measurement result of set B beams (or a set of beams to be measured) . Wherein the set B beams can be a subset of set A beams or a different set than set A beams. - Beam prediction in temporal domain: ■ NG-RAN node can predict the L1 quality of set A beams in the future based on historical L1 measurement result of set B beams. Wherein the set B beams can be a subset of set A beams or a different set than set A beams. - radio resource management (RRM) measurement prediction in temporal domain: ■ NG-RAN node can predict the L3 cell / beam quality of a serving cell based on the historical L1 / L3 measurement result of the same serving cell. - RRM measurement prediction in spatial domain: ■ predict the L3 measurement result for set A cell / beam using L1 / L3 measurement result of set B cell / beam at the same time instance (set A can be different set as set B, or a superset of set B) , wherein set A cell / beam and set B cell / beam are of the same frequency. The associated cells could be collocated. The associated cells could be collocated. - RRM measurement prediction in frequency domain: ■ predict the L3 measurement result for set A cell / beam using L1 / L3 measurement result of set B cell / beam at the same time instance (set A can be different set as set B, or a superset of set B) , wherein set A cell / beam and set B cell / beam are of different frequencies. - Load balancing related prediction: ■ NG-RAN can predict the radio resource status of its cells and decide the load balancing strategy taking the prediction result into account. - Mobility enhancement prediction: ■ NG-RAN can predict the cells UE will visit and decide the mobility strategy taking the prediction result into account. - Energy consumption prediction: ■ NG-RAN can predict the energy consumption in the future and decide on the energy saving strategy taking the prediction result into account. - Coverage and capacity optimization (CCO) issue prediction: ■ NG-RAN can predict the coverage or capacity issue that may happen in the future and adjust the cell or beam coverage configuration taking the prediction result into account. - Quality of experience (QoE) prediction: ■ NG-RAN can predict the QoE of a UE in the future and adjust the resource allocation strategy taking the prediction result into account. - RRM event prediction: ■ Predict if any of the RRM event (e.g., A1 / A2…A5 event, recovery link failure (RLF) ) in the future. - line of sight (LOS) / non-line of sight (NLOS) estimation: ■ In AI / ML assisted gNB assisted positioning, gNB may use AI / ML model to estimate if the path between UE and gNB is NLOS or LOS.
[0038] Aspects of the present disclosure are described in the context of a wireless communications system.
[0039] 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 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 LTE-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 of Electrical 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.
[0040] 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 signaling, transmit signaling) over a Uu interface.
[0041] 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 112 associated with the same or different radio access technologies may overlap, but the different geographic coverage areas may be associated with different NE 102.
[0042] 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 (IoT) device, an Internet-of-Everything (IoE) device, or machine-type communication (MTC) device, among other examples.
[0043] 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 114 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.
[0044] 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., S1, N2, N3, 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) .
[0045] The CN 106 may support user authentication, access authorization, 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.
[0046] The CN 106 may communicate with a packet data network over one or more backhaul links (e.g., via an S1, N2, N3, or another network interface) . The packet data network may include an application server. In some implementations, one or more UEs 104 may 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) .
[0047] 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.
[0048] 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., μ=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., μ=0) associated with the first subcarrier spacing (e.g., 15 kHz) may utilize one slot per subframe. A second numerology (e.g., μ=1) may be associated with a second subcarrier spacing (e.g., 30 kHz) and a normal cyclic prefix. A third numerology (e.g., μ=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., μ=3) may be associated with a fourth subcarrier spacing (e.g., 120 kHz) and a normal cyclic prefix. A fifth numerology (e.g., μ=4) may be associated with a fifth subcarrier spacing (e.g., 240 kHz) and a normal cyclic prefix.
[0049] A time interval of a resource (e.g., a communication resource) may be organized according to frames (also referred to as radio frames) . Each frame may have a duration, for example, 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.
[0050] Additionally or alternatively, a time interval of a resource (e.g., a communication resource) may be organized 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., μ=0, μ=1, μ=2, μ=3, μ=4) associated with respective subcarrier spacings of 15 kHz, 30 kHz, 60 kHz, 120 kHz, and 240 kHz may utilize 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., μ=0) associated with a first subcarrier spacing (e.g., 15 kHz) may be used interchangeably between subframes and slots.
[0051] 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 some implementations, 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.
[0052] 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., μ=0) , which includes 15 kHz subcarrier spacing; a second numerology (e.g., μ=1) , which includes 30 kHz subcarrier spacing; and a third numerology (e.g., μ=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., μ=2) , which includes 60 kHz subcarrier spacing; and a fourth numerology (e.g., μ=3) , which includes 120 kHz subcarrier spacing.
[0053] Under 5G, end-to-end communications between RAN, e.g., NG-RAN and a CN node or entity, e.g., network data analytics function (NWDAF) is not supported because the only interface that NG-RAN maintains towards 5GC is the point to point NG interface towards 5G AMF. Any communication between NG-RAN node or entity and other 5GC node or entity has to be relayed by 5G AMF, which is not efficient. Thus, the industry desires that the RAN in the future, e.g., NG-RAN under 6G (6G-RAN) may support a service based interface, e.g., Nran towards the CN, e.g., 6GC.
[0054] Figure 2 illustrates an example of service based interface that offering RAN services to other NFs under 6G or higher generation in accordance aspects of the present disclosure.
[0055] Referring to Figure 2, 6G-RAN may support a service based interface, e.g., Nran towards 6GC. The service based interface allows the end-to-end communications between 6G-RAN node or entity and a CN node or entity under 6G (6GC node or entity) . That is, the 6GC node or entity can invoke the services provided by the 6G-RAN node if it knows the corresponding uniform resource identifier (URI) , vice versa. In some cases, the service based interface may allow NF (s) , e.g., NWDAF or location management function (LMF) etc., positioned in RAN side rather than CN side as legacy.
[0056] In accordance with various aspects of the present disclosure, an exemplary service based interface, e.g., Nran is a service based interface specified in TS 23.501 or the like. In the context of this specification, an NF service is offering a capability to authorized consumers.
[0057] Specifically, NFs may offer different capabilities and thus, different NF services to distinct consumers. Each of the NF services offered by Nran shall be self-contained, reusable and use management schemes independently of other NF services offered by the same NF (e.g. for scaling, healing, etc. ) .
[0058] The discovery of the NF instance and NF service instance is specified in clause 6.3.1, TS 23.502. There can be dependencies between NF services within the same NF due to sharing some common resources, e.g. context data. This does not preclude that NF services offered by a single NF are managed independently of each other.
[0059] Each NF service shall be accessible by means of an interface. An interface may consist of one or several operations.
[0060] System procedures, as specified in TS 23.502 can be built by invocation of a number of NF services. Figure 7.2.1-3 as specified in TS 23.502 shows an illustrative example on how a procedure can be built, wherein it is not expected that system procedures depict the details of the NF Services within each NF.
[0061] The following clauses provide for each NF the NF services it exposes through its service based interfaces. This annex provides a high level description of the different communication models that NF and NF services can use to interact which each other. Table 1 shows Table E. 1-1 as specified in TS 23.502, which summarizes the communication models, their usage and how they relate to the usage of a service communication proxy (SCP) . The SCP can be used for indirect communications between NF / NF service instances. End-to-end communications between consumer and producer, which means end-to-end from protocol point of view, can be realized in direct or indirect way according to Table 1. In addition, Figure 3 depicts the different communication models for NF / NF service interaction, which reproduces Figure E. 1-1 as specified in TS 23.502. Table 1: Communication models for NF / NF services interaction summary Model A -Direct communication without network repository function (NRF) interaction: Neither NRF nor SCP are used. Consumers are configured with producers' "NF profiles" and directly communicate with a producer of their choice. Model B -Direct communication with NRF interaction: Consumers do discovery by querying the NRF. Based on the discovery result, the consumer does the selection. The consumer sends the request to the selected producer. Model C -Indirect communication without delegated discovery: Consumers do discovery by querying the NRF. Based on discovery result, the consumer does the selection of an NF Set or a specific NF instance of NF set. The consumer sends the request to the SCP containing the address of the selected service producer pointing to a NF service instance or a set of NF service instances. In the latter case, the SCP selects an NF Service instance. If possible, the SCP interacts with NRF to get selection parameters such as location, capacity, etc. The SCP routes the request to the selected NF service producer instance. Model D -Indirect communication with delegated discovery: Consumers do not do any discovery or selection. The consumer adds any necessary discovery and selection parameters required to find a suitable producer to the service request. The SCP uses the request address and the discovery and selection parameters in the request message to route the request to a suitable producer instance. The SCP can perform discovery with an NRF and obtain a discovery result.
[0062] Persons skilled in the art would well know that the service based interface illustrated above is only an example, which may evolve as 3GPP evolves.
[0063] Figure 4 illustrates an example of a procedure for a RAN entity to request AI / ML models from a first NF via the service based interface in accordance aspects of the present disclosure.
[0064] An exemplary RAN entity may be a gNB or an entity of a split gNB, e.g., CU of gNB. An example of the RAN entity hosts one or multiple of the following functions, wherein some or all of the RAN functionalities may be supported in a single RAN node or entity: - Functions for Radio Resource Management: Radio Bearer Control, Radio Admission Control, Connection Mobility Control, Dynamic allocation of resources to UEs in both uplink and downlink (scheduling) ; - IP and Ethernet header compression, uplink data decompression, encryption and integrity protection of data; - Connection setup and release; - Scheduling and transmission of paging messages; - Scheduling and transmission of system broadcast information (originated from the AMF or OAM) ; - Measurement and measurement reporting configuration for mobility and scheduling.
[0065] The first NF is at least responsible for or capable of training AI / ML models. An example of the first NF is a NWDAF, e.g., a 6GC NWDAF (or more evolved one) , or an enhanced 5GC NWDAF, or a NWDAF-like entity or node dedicated for RAN domain computing and analytics tasks. An exemplary NWDAF may include one or more of the following functionalities: - Support data collection from NFs and AFs; - Support data collection from OAM; - Support retrieval of information from data repositories (e.g. unified data repository (UDR) via unified data management (UDM) for subscriber-related information or via network exposure function (NEF) (packet flow description function (PFDF) ) for packet flow description (PFD) information) ; - Support data collection of location information from location calculation service (LCS) system; - NWDAF service registration and metadata exposure to NFs and AFs; - Support analytics information provisioning to NFs and AFs; - Support analytics collection from MDAF; - Support AI / ML model training and provisioning to NWDAF containing analytics logical function (AnLF) or NWDAF containing MTLF; - Support AI / ML Model storage to and retrieval from analytics data repository function (ADRF) ; - Support bulked data related to Analytics ID (s) provisioning for NFs; - Support accuracy information about Analytics IDs provisioning for NFs; - Support accuracy information or accuracy degradation about AI / ML model provisioning for NFs; - Support roaming exchange capability to exchange data and analytics between public land mobile networks (PLMNs) ; - Support Federated Learning (FL) to train an AI / ML model among multiple NWDAFs (containing MTLF) .
[0066] Herein, it is assumed that the RAN entity has finished the discovery procedure and has identified the associated NWDAF at least for the purpose of AI / ML model training.
[0067] Referring to Figure 4, in step 401, the RAN entity, which is a NWDAF service consumer may send a model request related message to the NWDAF via the service based interface. The service based interface is as illustrated above or the like. The model request related message may be used to invoke the model training service or model provision service supported by the NWDAF, and request the NWDAF to provide a required AI / ML model (s) trained for concerned RAN side use case (s) .
[0068] For example, in some cases, the RAN entity may be not aware of or not configured with information, e.g., an ID that represents an AI / ML model of available AI / ML models at the NWDAF. The RAN entity may subscribe or modify a subscription for training an AI / ML model (s) for the concerned RAN side use case (s) by invoking the model training service supported by the NWDAF, e.g., Nnwdaf_MLModelTraining_Subscribe service operation or the like. In some cases, the RAN entity may be aware of or configured with information that represents an AI / ML model of available AI / ML models at the NWDAF. The RAN entity may subscribe or modify a subscription for providing a trained AI / ML model (s) for the concerned RAN side use case (s) by invoking the model provision service supported by the NWDAF, e.g., Nnwdaf_MLModelProvision_Subscribe service operation or the like.
[0069] The RAN entity may provide one or multiple of the following exemplary parameters or information in the model request related message: - Analytics ID: identifies the analytics (or inferences) for the provided AI / ML model is used. - Model Interoperability Information. This is vendor-specific information that conveys, e.g., requested model file format, model execution environment, etc. The encoding, format, and value of Model Interoperable Information is not specified since it is vendor specific information, and is agreed between vendors, if necessary for sharing purposes. - A Notification Target Address (and Notification Correlation ID) allowing the Event Receiving NF to correlate notifications received from the Event provider with this subscription. A subscription is associated with a unique Notification Target Address (and Notification Correlation ID) . In the case that the NF consumer subscribes to the NF producer on behalf of other NF, the NF consumer includes the Notification Target Address (and Notification Correlation ID) of other NF for the Event ID which is to be notified to other NF directly and the Notification Target Address (and Notification Correlation ID) of itself for the Subscription change related event notification. Each Notification Target Address (and Notification Correlation ID) is associated with related (set of) Event ID (s) . - Model Information (e.g., address (e.g. universal resource locator (URL) or fully qualified domain name (FQDN) of model files, which belongs to information related to fetching files of AI / ML models) . In an example, a RAN entity may provide the information related to the AI / ML model that is currently used and requests the NWDAF to update the AI / ML model. - Model ID: identifies the provided AI / ML model. In an example, a RAN entity may provide the information related to the AI / ML model that is currently used and requests the NWDAF to update the AI / ML model. - Model Accuracy Check Flag: identifies that the request is for using the local training data as the testing dataset to calculate the Model Accuracy of the global AI / ML model provided by the service consumer NWDAF acting as the FL Server NWDAF. The same or another similar flag may be used for the RAN entity to request using NWDAF’s local data as the testing dataset to calculate the Model Accuracy of the trained AI / ML model. - Correlation ID: identifies the Federated Learning procedure for training the AI / ML model. This parameter is included when the service is used for Federated Learning. - Available data requirement. This is for informing the requirement on available data for the AI / ML model training. For example FL Server NWDAF sends the requirement in preparation request to a FL Client NWDAF for selecting the FL Client NWDAF which can meet the available data requirement. The following available data requirements can be included: a) Event ID list to be collected for local model training. b) Dataset statistical properties (e.g., UNIFORM_DIST_DATA, NO_OUTLIERS) c) Time window of the data samples. d) Minimum number of data samples. In an example, this parameter can be used for RAN to inform NWDAF about the requirement on available data for AI / ML model training for RAN side use case, and the following requirements can be further provided: e) Prediction window information, e.g., the minimum and maximum time in advance the model can predict. f) Observation window information, e.g., the prediction is made based on the historical measurement observed in a time window in the past. g) Frequency list the data is collected from. h) Cell list the data is collected from. i) The set A and set B beam association, e.g., for beam prediction use case, where ■ Set A related information also refers to information related to the beams to be predicted, e.g., number of beams in Set A, and distance and / periodicity of SS / PBCH block (SSB) / CSI-RS; and ■ Set B related information also refers to information related to the beams to be measured, e.g., number of beams in Set B, type of SSB or CSI-RS, distance and / or periodicity of SSB / CSI-RS, and mapping from Set B beam (s) to Set A beams. j) Quantity of the data, e.g., reference signal receiving power (RSRP) or reference signal receiving quality (RSRQ) , L1 measurement or L3 measurement. k) Resolution of the data. l) Network side condition represented by an associated ID. - Availability time requirement. This is for informing the requirement on availability time for the AI / ML model training, e.g. FL Server NWDAF sends the requirement in preparation request to FL Client NWDAF for selecting the FL Client NWDAF which is available in the required time for training AI / ML model. - Training Filter Information: enables to select which data for the AI / ML model training is requested, e.g. single network slice selection assistance information (S-NSSAI) , Area of Interest. - Target of Training Reporting: indicates the object (s) for which data for AI / ML model training is requested, i.e. a group of UEs or any UE (i.e. all UEs) . - Use case context: indicates the context of use of AI / ML model. For example, it can be any of the RAN side use case considered in the present disclosure, CSI compression, CSI prediction in temporal domain, beam prediction in spatial domain, beam prediction in temporal domain, RRM prediction in temporal domain, RRM prediction in spatial domain, RRM prediction in frequency domain, load balancing related prediction, mobility enhancement prediction, energy consumption prediction, CCO issue prediction, or QoE prediction etc. - Training Reporting Information with the following parameters: ■ Maximum response time: indicates maximum time for waiting notifications (i.e. training results) . - Iteration round ID: indicates the iteration round number of current AI / ML model training. - Expiry time. - Input Data Information: include the list of parameters that will be used by the AI / ML model as input. Examples are given in the table below. - Output Data Information: include the list of parameters that will be generated by the model as output.
[0070] Examples of input data information and output data information are shown in Table 2 below. Table 2
[0071] The NWDAF may check whether there is the AI / ML model (s) trained as required by the RAN entity. If there is no AI / ML trained as required by the RAN entity, the NWDAF may train the AI / ML model (s) based on the received model request related message at step 403, e.g., by collecting new data or using the data that it owns. For example, to train a RAN side AI / ML model, the NWDAF may collect the relevant training data from RAN side at step 405, e.g., utilizing an application programming interface (API) , e.g., Nran_DataCollection or the like supported by the RAN side for exposing RAN side data. Herein, it is assumed that the RAN side also refers to the same RAN entity for simplification, while it may be a different RAN entity or node in some cases. Some exemplary procedures or schemes of collecting or retrieving training data are illustrated in the following in detail.
[0072] In accordance with some aspects of the present disclosure, the NWDAF may directly collect or retrieve the data for model training from RAN side via the service based interface, e.g., by invoking a data exposure service, e.g., Nran_DataExposure service offered by the RAN side to request the needed training data.
[0073] For example, at step 407, the NWDAF may invoke data exposure service, e.g., the Nran_DataExposure service supported in the RAN side by sending a data request related message over the service based interface, e.g., sending a data exposure subscription message (e.g., a Nran_DataExposure_Subscribe message or the like) . The data request related information or parameters related to the RAN data that the NWDAF needs. An example of data request related message may indicate one or multiple of the following: a) The time period that data should be collected continuously from the same UE (the UE is selected by the RAN) . b) Frequency list the data is collected from. c) Cell list the data is collected from. d) The set A and set B beam association, e.g., for beam prediction use case, where ■ Set A related information also refers to information related to the beams to be predicted, e.g., number of beams in Set A, and distance and / periodicity of SS / PBCH block (SSB) / CSI-RS; and ■ Set B related information also refers to information related to the beams to be measured, e.g., number of beams in Set B, type of SSB or CSI-RS, distance and / or periodicity of SSB / CSI-RS, and mapping from Set B beam (s) to Set A beams. e) Quantity of the data, e.g., RSRP or RSRQ, L1 measurement or L3 measurement. f) Resolution of the data. g) Network side condition represented by an associated ID. h) Type of the data, which can be any of the following: ■ Raw CSI-RS measurement result before compression ■ Associated ID representing the network deployment ■ Consecutive CSIs (historical CSIs) where the type of CSIs includes raw channel matrices, or PMIs, or eigen vectors, etc. ■ Associated ID representing the network deployment ■ L1-RSRP measurement based on Set B or a subset of Set B of beams ■ Partial L1-RSRP measurement based on Set B or a subset of Set B + beam indicator ■ Top-k beam indicator, k is an integer configured by NWDAF. ■ Cell level L3-RSRP measurement ■ Beam level L3-RSRP measurement ■ Historical RRM event (e.g., A1 / A2…A5, RLF event) ■ Historical Resource status (e.g., available PRB) of a cell ■ Number of UEs ■ Number of RRC connections ■ UE historical visited cells ■ Energy consumption ■ Radio resource status ■ CCO state ■ CCO issue ■ QoE.
[0074] In some cases, similar to an event exposure subscription message, e.g., Nran_EventExposure_Subscribe message, the data request related message may also indicate one or multiple of the following: - Notification Target Address (and Notification Correlation IDs) - UE (s) ID (subscription permanent identifier (SUPI) or Internal Group Identifier or indication that any UE is targeted) - Data Collection ID - Filter (s) associated with each Data Collection ID - Subscription Correlation ID (in the case of modification of the event subscription) - Expiry time - A list of group member UE (s) whose subscription to event notification (s) are removed or added for a group-based event notification subscription, operation indication (cancellation or addition) .
[0075] After receiving the data request related message, the RAN entity may collect the RAN side data as required. In step 409, the RAN entity may send the RAN side data or information related to fetching the RAN side data to the NWDAF via the service based interface, e.g., by a data exposure notify message (e.g., Nran_DataExposure_Notify message or the like) . The RAN side data sent to the NWDAF may be in a format of string, array, or object etc. The information related to fetching the RAN side data (or fetch instruction information or the like) may indicate the address information (e.g., Uri) that the NWDAF could fetch the file or document containing the required training data.
[0076] If the RAN side notifies the NWDAF with the fetch instruction information, then the NWDAF may fetch the needed data according to the fetch instruction information. For example, at step 411, the NWDAF may send a data fetching related message, e.g., a data exposure fetch message (e.g., Nran_DataExposure_Fetch message or the like) to RAN side via the service based interface to fetch the file or document containing the training data. Accordingly, the RAN side may send the file or document containing the training data to the NWDAF via the service based interface at step 413, e.g., by a further data exposure notify message as used in step 409.
[0077] In accordance with some aspects of the present disclosure, the NWDAF may not directly collect or retrieve the data for model training from the RAN side. Instead, the NWDAF may collect or retrieve the data for model training from the RAN side indirectly via another NF (hereinafter, second NF) . The second NF may be an ADRF or a data collection coordination function (DCCF) or the like.
[0078] In some implementations of the present disclosure, the NWDAF may invoke a data management service, e.g., Ndccf_DataManagement service offered by the second NF via the service based interface. Then, the second NF may invoke a data exposure service, e.g., Nran_DataExposure service offered by the RAN side to request the training data. It is upon the second NF to select and collect the required training data from RAN side.
[0079] For example, at step 415, the NWDAF may send a data management subscription message to the second NF, e.g., Ndccf_DataManagement_Subcribe message or the like in the case of DCCF or Nadrf_DataManagement_Subcribe message or the like in the case of ADRF.
[0080] Then, the DCCF or ADRF may obtain the training data from the RAN side by a data exposure service operation at step 417, e.g., Nran_DataExposure service operation, which is similar to that recited at steps 407 to 413 between the RAN side and NWDAF, and thus will not repeated herein.
[0081] At step 419, the second NF, e.g., the DCCF or ADRF may send the RAN side data or information related to fetching the RAN side data to the NWDAF, e.g., by a data management notify message. An exemplary data management notify message may be Ndccf_DataExposure_Notify message in the case of DCCF or the like, or Nadrf_DataExposure_Notify message in the case of ADRF or the like. Similarly, if the second NF notifies the NWDAF with the fetch instruction information (information related to fetching the RAN side data) , the NWDAF may fetch the data according to the information related to fetching the RAN side data. For example, at step 421, the NWDAF may send a data fetching related message to the second NF, e.g., a data management fetch message. An exemplary data management fetch message may be Ndccf_DataManagement_Fetch message in the case of DCCF or the like, or Nadrf_DataManagement_Fetch message in the case of ADRF or the like. The second NF may send the file or document containing the training data to the NWDAF at step 423, e.g., by a further data management notify message as used in step 419.
[0082] In some implementations of the present disclosure, the NWDAF may retrieve the training data stored in the second NF, e.g., the ADRF. For example, before the NWDAF may invoke a data management retrieve service, e.g., Ndccf_DataManagement_Retrieve service offered by the second NF, at step 425, the RAN side may proactively store the training data in the second NF, e.g., ADRF via a data management storage request message (e.g., Nadrf_DataManagement_StorageRequest or the like) .
[0083] When the ADRF receives a data request related message from the NWDAF at step 427, e.g., by a data management retrieve request message (e.g., Nadrff_DataManagement_RetrieveRequest message or the like) , the ADRF may send the RAN side data or information related to fetching the RAN side data to the NWDAF at step 429, e.g., by a data management retrieve request response message (e.g., Nadrff_DataManagement_RetrieveRequestResponse message or the like) . Similarly, if the ADRF notifies the NWDAF with the fetch instruction information (information related to fetching the RAN side data, the NWDAF may fetch the data according to the information related to fetching the RAN side data. For example, at step 431, the NWDAF may send a further data management retrieve request message as that used at step 427. The ADRF may send the file or document containing the training data to the NWDAF at step 433 by a further data management retrieve request response message as that used in step 429.
[0084] When there is the required AI / ML model (s) or the NWDAF completes the AI / ML model training, the NWDAF may send or notify the RAN entity with model related information in step 435. For example, in the case that the RAN entity requests the trained AI / ML model (s) by invoking a model training service, the NWDAF may send the model file (s) or information related to fetching the model file (s) by invoking a model training notification service, e.g., Nnwdaf_MLModelTraining_Notify service operation. In the case that the RAN entity requests the trained AI / ML model (s) by invoking a model provision service, the NWDAF may send the model file (s) or information related to fetching model file (s) by invoking a model provision notification service, e.g., Nnwdaf_MLModelProvision_Notify service operation.
[0085] Specifically, the NWDAF may provide or notify one or more of the following information or parameters in the model related information: - The Notification Correlation Information. - Model Information which includes: ■ the AI / ML model file address (e.g. URL or FQDN) or analytics data repository function (ADRF) (Set) ID and Model Identifier or Model Storage Transaction Identifier if available, or the Model file (e.g., in String) - Model ID: identifies the provisioned AI / ML model. - Model Accuracy: The model accuracy of the trained RAN side model, which is calculate by the FL Client NWDAF using the local training data as the testing dataset. In one example, the model accuracy of the trained RAN side model representing the accuracy of the prediction result inferred by the AI / ML model. - Status report of FL training: Accuracy of local model and Training Input Data Information (e.g. areas covered by the data set, sampling ratio, maximum / minimum of value of each dimension, etc. ) , which are generated by the FL Client NWDAF during FL procedure. The parameters in Training Input Data Information are up to the implementation. - Correlation ID. This parameter may be included when the service is used for Federated Learning. - Iteration round ID: indicates the iteration round number of AI / ML model training indicated by the FL Server NWDAF. - Delay Event Notification with the following parameters: ■ delay event indication: this parameter indicates that FL Client NWDAF is not able to complete the training of the interim local AI / ML model within the maximum response time provided by the FL Server NWDAF. ■ cause code (e.g. local AI / ML model training failure, more time necessary for local AI / ML model training, etc. ) . ■ Expected time to complete the training: Indicates to the FL Server NWDAF that expected remaining training time, and may be provided with Delay Event Notification. - Input Data Information: include the list of parameters that will be used by the AI / ML model as input. Examples are similar or identical to those illustrated in Table 2. - Output Data Information: include the list of parameters that will be generated by the model as output. Examples are similar or identical to those illustrated in Table 2.
[0086] In the case that the model related information provides the model file (s) , the RAN entity may directly obtain the AI / ML model (s) . In the case that the model related information provides only the information related to fetching the model file (s) , the RAN entity may need to retrieve or fetch the model file (s) based on the information related to fetching the model file (s) . For example, the RAN entity may retrieve the model files based on a URL of model files notified by the NWDAF.
[0087] After receiving and deploying an RAN sided model trained by the NWDAF, the RAN may monitor the performance of the AI / ML model when the inference starts. For example, the RAN entity may compare the predicted results (e.g., L1 / L3 beam qualities) with the actual measurement results to monitor the accuracy of the AI / ML model. In some cases, the RAN entity may report the monitoring results to the NWDAF.
[0088] For example, the RAN may send a monitoring register related message to the NWDAF via the service based interface at step 437, e.g., by a model monitoring register request message (e.g., a Nnwdaf_MLModelMonitor_Register request message or the like) . The model monitoring register request message may indicate that monitoring at least one AI model deployed at the RAN side has started. An exemplary model monitoring register request message may indicate an ID of the at least one AI model being monitored at the RAN side. In some cases, an exemplary model monitoring register request message may further indicate one or multiple of the following: RAN ID (e.g., RAN node ID) , a subscription endpoint of a model monitoring subscription service (e.g., Nran_MLModelMonitor_Subscribe service operation at the RAN or the like) , e.g., RAN URI address.
[0089] Then, the NWDAF may be aware of the AI / ML model ID and the ID of the RAN entity that is monitoring the accuracy of that AI / ML model. The NWDAF may send result request related message to the RAN entity at step 439, e.g., by a model monitoring subscription request message (e.g., Nran_MLModelMonitor_Subscribe request or the like) . An exemplary result request related message may indicate one or multiple of the following: an ID of the AI model (s) being monitored, analytics ID (s) , accuracy metrics to be monitored, reporting threshold (s) or reporting period (s) etc.
[0090] The RAN entity may determine whether to report the monitoring results of the concerned AI / ML model (part or all of the deployed RAN side AI / ML models) to the NWDAF, e.g., based on the reporting threshold (s) or reporting period (s) etc. The reporting threshold (s) may be locally configured or indicated in the result request related message. For example, the RAN entity may determine whether the analytics accuracy of the AI / ML model is insufficient (or satisfying or the like) , e.g., whether the deviation of the output analytics using the trained AI / ML model from ground truth data is greater than the reporting threshold (s) , or whether the reporting period indicated in the result request related message is reached.
[0091] In the case that the RAN entity determines to report the monitoring results of an AI / ML model, the RAN entity may send the monitoring results of the AI / ML model to the NWDAF at step 441, e.g., by a model monitoring notification message (e.g., Nnwdaf_MLModelMonitor_Notify or the like) .
[0092] The monitoring results may include one or multiple of the following: analytics feedback information, or analytics accuracy information of the related AI / ML model, or the number of inferences that were performed during a time interval, or an indication that the analytics accuracy of the AI / ML model does not meet the accuracy requirement for the AI / ML model (or an indication that the analytics accuracy of the AI / ML model meets the accuracy requirement for the AI / ML model) . An example of analytics accuracy information of the related AI / ML model may be a deviation value which indicates the deviation of the predictions generated using the AI / ML model from the ground truth data, and / or the network data when the deviation occurs. The NWDAF may use the network data when the deviation occurs to further train the related AI / ML model in some cases. Regarding the time interval where the inferences were performed, it may be between transmission of the model monitoring register request message and the model monitoring notification message, or between the time of the last model monitoring notification message and the time of the current model monitoring notification message.
[0093] Since end-to-end communications between the first NF or second NF and the RAN entity can be performed via the service based interface in various scenarios as illustrated above, the communication efficiency will be increased.
[0094] Figure 5 illustrates an example of a wireless communication apparatus 500 in accordance with aspects of the present disclosure, which may be a NE (or RAN entity) or a NF. The wireless communication apparatus 500 may include a processor 502, a memory 504, a controller 506, and a transceiver 508. The processor 502, the memory 504, the controller 506, or the transceiver 508, 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.
[0095] The processor 502, the memory 504, the controller 506, or the transceiver 508, 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.
[0096] The processor 502 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 502 may be configured to operate the memory 504. In some other implementations, the memory 504 may be integrated into the processor 502. The processor 502 may be configured to execute computer-readable instructions stored in the memory 504 to cause the wireless communication apparatus 500 to perform various functions of the present disclosure.
[0097] The memory 504 may include volatile or non-volatile memory. The memory 504 may store computer-readable, computer-executable code including instructions when executed by the processor 502 cause the wireless communication apparatus 500 to perform various functions described herein. The code may be stored in a non-transitory computer-readable medium such the memory 504 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 special-purpose computer.
[0098] In some implementations, the processor 502 and the memory 504 coupled with the processor 502 may be configured to cause the wireless communication apparatus 500 to perform one or more of the functions described herein (e.g., executing, by the processor 502, instructions stored in the memory 504) . For example, the processor 502 may support wireless communication at the wireless communication apparatus 500 in accordance with examples as disclosed herein. In the case that the wireless communication apparatus 500 is a Nor RAN entity, the RAN entity may be configured to support a means for sending, to a first NF, a model request related message of requesting at least one AI model trained by the first NF; a means for receiving, from the first NF, a model related message including a file of the at least one AI model or information related to fetching the file of the at least one AI model; and a means for deploying the at least one AI model for RAN side use cases based on the model related message. In the case that the wireless communication apparatus 500 is a NF, the NF may be configured to support a means for receiving, from a RAN entity, a model request related message of requesting at least one AI model trained by the NF; and a means for sending, to the RAN entity, a model related message including a file of the at least one AI model or information related to fetching the file of the at least one AI model.
[0099] The controller 506 may manage input and output signals for the wireless communication apparatus 500. The controller 506 may also manage peripherals not integrated into the wireless communication apparatus 500. In some implementations, the controller 506 may utilize an operating system such as or other operating systems. In some implementations, the controller 506 may be implemented as part of the processor 502.
[0100] In some implementations, the wireless communication apparatus 500 may include at least one transceiver 508. In some other implementations, the wireless communication apparatus 500 may have more than one transceiver 508. The transceiver 508 may represent a wireless transceiver. The transceiver 508 may include one or more receiver chains 510, one or more transmitter chains 512, or a combination thereof.
[0101] A receiver chain 510 may be configured to receive signals (e.g., control information, data, packets) over a wireless medium. For example, the receiver chain 510 may include one or more antennas for receive the signal over the air or wireless medium. The receiver chain 510 may include at least one amplifier (e.g., a low-noise amplifier (LNA) ) configured to amplify the received signal. The receiver chain 510 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 510 may include at least one decoder for decoding the processing the demodulated signal to receive the transmitted data.
[0102] A transmitter chain 512 may be configured to generate and transmit signals (e.g., control information, data, packets) . The transmitter chain 512 may include at least one modulator for modulating data onto a carrier signal, preparing the signal for transmission over 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 512 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 512 may also include one or more antennas for transmitting the amplified signal into the air or wireless medium.
[0103] Figure 6 illustrates a flowchart of a method in accordance with aspects of the present disclosure. The operations of the method may be implemented by a NE or RAN entity as described herein. In some implementations, the NE or RAN entity may execute a set of instructions to control the function elements of the NE or RAN entity to perform the described functions.
[0104] At step 601, the method may include sending, to a first NF, a model request related message of requesting at least one AI model trained by the first NF. The operations of step 601 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of 601 may be performed by a RAN entity as described with reference to Figure 5.
[0105] At step 603, the method may include receiving, from the first NF, a model related message including a file of at least one AI model or information related to fetching the file of the at least one AI model; and deploying the at least one AI model for RAN side use cases based on the model related message. The operations of step 603 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of step 603 may be performed by a RAN entity as described with reference to Figure 5.
[0106] At step 605, the method may include deploying the at least one AI model for RAN side use cases based on the model related message. The operations of step 605 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of step 605 may be performed by a RAN entity as described with reference to Figure 5.
[0107] 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.
[0108] Figure 7 illustrates a flowchart of a method in accordance with aspects of the present disclosure. The operations of the method may be implemented by a NF as described herein. In some implementations, the NF may execute a set of instructions to control the function elements of the NF to perform the described functions.
[0109] At step 701, the method may include receiving, from a RAN entity, a model request related message of requesting at least one AI model trained by the NF. The operations of step 701 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of step 701 may be performed by a NF as described with reference to Figure 5.
[0110] At step 703, the method may include sending, to the RAN entity, a model related message including a file of the at least one AI model or information related to fetching the file of the at least one AI model. The operations of step 703 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of step 703 may be performed by a NF as described with reference to Figure 5.
[0111] 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.
[0112] 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.
Claims
1.A radio access network (RAN) entity for wireless communication, comprising:at least one memory; andat least one processor coupled with the at least one memory and configured to cause the RAN entity to:send, to a first network function (NF) , a model request related message of requesting at least one artificial intelligence (AI) model trained by the first NF;receive, from the first NF, a model related message including a file of the at least one AI model or information related to fetching the file of the at least one AI model; anddeploy the at least one AI model for RAN side use cases based on the model related message.2.The RAN entity of claim 1, wherein the model request related message indicates one or multiple of the following:model information;model identifier (ID) ;model accuracy check flag;available data requirement for model training for RAN side use case;model use case context;input data information to be used for training models; oroutput data information to be used for training models.3.The RAN entity of claim 1, wherein the model related message further indicates one or multiple of the following:model accuracy;input data information to be used for training models; oroutput data information to be used for training models.4.The RAN entity of claim 1, wherein in the case of receiving the model related message including information related to fetching the file of the at least one AI model, the at least one processor is further configured to cause the RAN entity to:retrieve the file of the at least one AI model based on the information related to fetching the file of the at least one AI model.5.The RAN entity of claim 1, wherein the at least one processor is further configured to cause the RAN entity to:receive, from the first NF, a data request related message of requesting RAN side data, including information related to needed data; andsend, to the first NF, the RAN side data or information related to fetching the RAN side data based on the data request related message.6.The RAN entity of claim 5, wherein in the case of sending the information related to fetching the RAN side data, the at least one processor is further configured to cause the RAN entity to:receive, from the first NF, a data fetching related message of fetching RAN side data; andsend, to the first NF, the RAN side data based on the data fetching related message.7.The RAN entity of claim 1, wherein the at least one processor is further configured to cause the RAN entity to:receive, from a second NF, a data request related message of requesting RAN side data, including information related to needed data; andsend, to the second NF, the RAN side data or information related to fetching the RAN side data based on the data request related message.8.The RAN entity of claim 7, wherein in the case of sending the information related to fetching the RAN side data, the at least one processor is further configured to cause the RAN entity to:receive, from the second NF, a data fetching related message of fetching RAN side data; andsend, to the second NF, the RAN side data based on the data fetching related message.9.The RAN entity of claim 1, wherein the at least one processor is further configured to cause the RAN entity to:request storing RAN side data in a second NF by invoking a data storage request.10.The RAN entity of claim 1, wherein the at least one processor is further configured to cause the RAN entity to:monitor the at least one AI model deployed at the RAN side; andreport monitoring results of the at least one AI model to the first NF.11.The RAN entity of claim 10, wherein before reporting the monitoring results of the at least one AI model, the at least one processor is further configured to cause the RAN entity to:send, to the first NF, a monitoring register related message of indicating that monitoring the at least one AI model deployed at the RAN side has started; andreceive, from the first NF, a result request related message of requesting the monitoring results of the at least one AI model.12.The RAN entity of claim 11, wherein the monitoring register related message indicates an identifier (ID) of the at least one AI model being monitored at the RAN side.13.The RAN entity of claim 11, wherein the result request related message indicates one or multiple of the following: an identifier (ID) of the at least one AI model being monitored, analytics IDs, accuracy metrics to be monitored, reporting thresholds or reporting periods.14.The RAN entity of claim 13, wherein the at least one processor is further configured to cause the RAN entity to:determine whether to report the monitoring results of the at least one AI model to the first NF based on the reporting thresholds or reporting periods.15.A network function (NF) for wireless communication, comprising:at least one memory; andat least one processor coupled with the at least one memory and configured to cause the NF to:receive, from a radio access network (RAN) entity, a model request related message of requesting at least one artificial intelligence (AI) model trained by the NF; andsend, to the RAN entity, a model related message including a file of at least one AI model or information related to fetching the file of the at least one AI model.16.The NF of claim 15, wherein the NF is a core network (CN) NF containing model training logical function (MTLF) , or an entity or node dedicated for RAN domain computing and analytics tasks.17.The NF of claim 16, wherein before sending the model related message, the at least one processor is configured to further cause the NF to:train the at least one AI model based on RAN side data received from the RAN entity or a different NF.18.The NF of claim 17, wherein before sending the model related message, the at least one processor is configured to further cause the NF to:send, to the RAN entity or the different NF, a data request related message of requesting RAN side data, including information related to needed data; andreceive, from the RAN entity or the different NF, the RAN side data or information related to fetching the RAN side data.19.A method performed by a radio access network (RAN) entity, comprising:sending, to a first network function (NF) , a model request related message of requesting at least one artificial intelligence (AI) model trained by the first NF;receiving, from the first NF, a model related message including a file of at least one AI model or information related to fetching the file of the at least one AI model; anddeploying the at least one AI model for RAN side use cases based on the model related message.20.A method performed by a network function (NF) , comprising:receiving, from a radio access network (RAN) entity, a model request related message of requesting at least one artificial intelligence (AI) model trained by the NF; andsending, to the RAN entity, a model related message including a file of the at least one AI model or information related to fetching the file of the at least one AI model.
Citation Information
Patent Citations
Model training method and device, terminal, functional entity and network equipment
CN118827405A
Data processing method and apparatus
US20230127558A1
Methods and systems for artificial intelligence based architecture in wireless network
US20230319585A1
Methods and system for serviced-based ai / ML model training, verification, registration and deployment in ran intelligent controllers
WO2023198799A1