User equipment-requested data transfers over the user plane in wireless systems
UE-requested data transfers over the user plane in wireless systems address network congestion and latency issues by optimizing the delivery of large machine learning models through the 5G Core Network architecture, improving AI/ML operations.
Patent Information
- Application Number
- PCT/CN2024/103303
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2024-07-03
- Publication Date
- 2025-10-30
AI Technical Summary
Existing wireless communication systems face network congestion and high delivery latency in the control plane when transferring large machine learning models, particularly large language models, which overload network functions and increase Service Based Interface signaling costs.
Implementing UE-requested data transfers over the user plane (UP) to mitigate network congestion and latency, allowing for efficient delivery of machine learning models and related data through the user plane, utilizing the 5G Core Network architecture and its components like AMF, SMF, UPF, and NWDAF to facilitate AI/ML operations.
Reduces network congestion in the control plane, decreases delivery latency, and optimizes the transfer of large machine learning models, enhancing the efficiency and reliability of AI/ML operations in wireless communication systems.
Smart Images

Figure CN2024103303_30102025_PF_FP_ABST
Abstract
Description
USER EQUIPMENT-REQUESTED DATA TRANSFERS OVER THE USER PLANE IN WIRELESS SYSTEMSTECHNICAL FIELD
[0001] This disclosure is directed generally to digital wireless communications.BACKGROUND
[0002] Mobile telecommunication technologies are moving the world toward an increasingly connected and networked society. In comparison with the existing wireless networks, next generation systems and wireless communication techniques will need to support a much wider range of use-case characteristics and provide a more complex and sophisticated range of access requirements and flexibilities.
[0003] Long-Term Evolution (LTE) is a standard for wireless communication for mobile devices and data terminals developed by 3rd Generation Partnership Project (3GPP) . LTE Advanced (LTE-A) is a wireless communication standard that enhances the LTE standard. The 5th generation of wireless system, known as 5G, advances the LTE and LTE-Awireless standards and is committed to supporting higher data-rates, large number of connections, ultra-low latency, high reliability and other emerging business needs.SUMMARY
[0004] Techniques are disclosed for user equipment (UE) -requested data transfers over the user plane (UP) , e.g., to support machine learning (ML) models, in current and emerging cellular networks. The described embodiments advantageously reduce network congestion in the control plane, which is currently being used for data transfers for artificial intelligence (AI) / ML-related applications.
[0005] In an example aspect, a wireless communication method includes transmitting, by a wireless device to a control plane network function, a registration message comprising artificial intelligence (AI) -related information for the wireless device, and receiving, from the control plane network function, a registration response message.
[0006] In another example aspect, a wireless communication method includes transmitting, by a network function to a network storage, a network function profile comprising a capability for establishing a user plane between the network function and a wireless device for different artificial intelligence (AI) -related purposes, and receiving, from the network storage, a response message comprising a success or failure indication associated with the network storage performing a storage operation on the network function profile.
[0007] In yet another example aspect, a wireless communication method includes transmitting, by a wireless device to a network function, a first request message to obtain at least one of an information for a machine learning (ML) model, an updated information for the ML model, or training data for the ML model over a network plane specified by the wireless device, receiving, in response to the first request message, a response message via the network plane specified by the wireless device, and performing, based on the response message, an ML model training operation or an ML model inference operation. In this example method, the response message comprises the information for the ML model, the updated information for the ML model, or the training data for the ML model.
[0008] In yet another example aspect, the above-described methods are embodied in the form of processor-executable code and stored in a non-transitory computer-readable storage medium. The code included in the computer readable storage medium when executed by a processor, causes the processor to implement the methods described in this patent document.
[0009] In yet another example aspect, a device that is configured or operable to perform the above-described methods is disclosed.
[0010] The above and other aspects and their implementations are described in greater detail in the drawings, the descriptions, and the claims.
[0011] BRIEF DESCRIPTION OF THE DRAWING
[0012] FIG. 1 shows an example architecture for the Fifth Generation Core (5GC) network.
[0013] FIG. 2 shows an example data collection architecture of the network data analytics function (NWDAF) from any 5GC network function (NF) .
[0014] FIG. 3 shows an example network data analytics exposure architecture.
[0015] FIG. 4 is a timing diagram of an example method for a UE reporting and / or updating its AI-related information.
[0016] FIG. 5 is a timing diagram of an example method for a network-initiated ML model update via the user plane (UP) .
[0017] FIG. 6 is a timing diagram of an example method for a UE-initiated request for an ML model or update, and / or a request for training data.
[0018] FIGS. 7 to 9 show flowchart for example wireless communication methods.
[0019] FIG. 10 shows a block diagram of an example hardware platform that may be a part of a network device or a communication device.
[0020] FIG. 11 shows an example of wireless communication including a base station (BS) and user equipment (UE) based on some implementations of the disclosed technology.DETAILED DESCRIPTION
[0021] In existing wireless communication standards (e.g., 3GPP Rel-18) , various artificial intelligence and machine learning (AI / ML) systems are being deployed to manage complex networks intelligently, address system optimization problems, and improve user experience in Fifth Generation (5G) New Radio (NR) system and beyond. AI / ML models can be readily deployed in many applications in mobile devices in existing 5G NR systems and upcoming 6G system deployments, e.g., image recognition, natural language recognition, speech recognition, and video processing.
[0022] The ML model size, especially for large language models (LLMs) , is generally large, e.g., one billion parameters typically require 2 GB, and the largest Llama-2 model has 70 billion parameters. Existing systems are deliver the ML models to a User Equipment (UE) via the control plane (CP) , which can overload one or more network functions (e.g., AMF) and increase Service Based Interface (SBI) signaling costs in 5GC. Furthermore, the delivery latency via the control plane is relatively large compared to that of the user plane (UP) .
[0023] Embodiments of the disclosed technology provide methods and systems for UE-requested (or UE-initiated) data transfers over the UP between the UE and any 5GC network function (NF) , which mitigates the drawbacks described above.
[0024] The example headings for the various sections below are used to facilitate the understanding of the disclosed subject matter and do not limit the scope of the claimed subject matter in any way. Accordingly, one or more features of one example section can be combined with one or more features of another example section. Furthermore, 5G terminology is used for the sake of clarity of explanation, but the techniques disclosed in the present document are not limited to 5G technology only, and may be used in wireless systems that implemented other protocols, such as 6G technology and systems.
[0025] 1 Introduction
[0026] The Fifth Generation Core (5GC) Network architecture relies on a "Service-Based Architecture" (SBA) framework, where the architecture elements are defined in terms of "Network Functions" (NFs) rather than by "traditional" Network Entities. Via interfaces of a common framework, any given NF offers its services to all the other authorized NFs and / or to any "consumers" that are permitted to make use of these provided services. Such an SBA approach offers modularity and reusability.
[0027] 1.1 5GC Architecture
[0028] FIG. 1 shows an example architecture for the 5GC network. As shown therein, the 5GC network includes the following components:
[0029] – User Equipment (UE) .
[0030] – Radio Access Network (RAN) . The RAN manages the radio resource and delivers the user data received over the N3 interface to UE and delivers the user data from UE over the N3 interface. The RAN performs mapping between DRBs (Dedicated Radio Bearers) and the quality of service (QoS) flows in the Packet Data Unit (PDU) session.
[0031] – Access and Mobility Management Function (AMF) . The AMF supports the following functionalities: registration management, connection management, and reachability management and mobility management. This function also performs access authentication and access authorization. The AMF is the non-access stratum (NAS) security termination and relays the Session Management (SM) NAS between UE and Session Management Function (SMF) , etc.
[0032] – Session Management Function (SMF) . The SMF supports the following functionalities: session establishment, modification and release, UE internet protocol (IP) address allocation &management (including optional authorization functions) , selection and control of user plane function (UPF) , downlink data notification, etc. The SMF controls the UPF via N4 interface association. The SMF provides PDR (Packet Detection Rule) to UPF to instruct how to detect user data traffic, FAR (Forwarding Action Rule) , QER (QoS Enforcement Rule) and URR (Usage Reporting Rule) to instruct the UPF how to perform the user data traffic forwarding, QoS handling and usage reporting for the user data traffic detected by using the PDR.
[0033] – User Plane Function (UPF) . The UPF supports the following functionalities: serving as an anchor point for intra- / inter-radio access technology (RAT) mobility, packet routing &forwarding, traffic usage reporting, QoS handling for the user plane, downlink packet buffering and downlink data notification triggering, etc. General Packet Radio Services (GPRS) tunneling protocol user plane (GTP-U) tunnel is used over N3 interface between the RAN and UPF, and is per PDU session. For downlink traffic the UPF binds the downlink traffic to QoS flows within the GTP-U tunnel of the PDU session by using the FARs received from SMF. For uplink traffic, the RAN transfer the user plane traffic to QoS flows identified by the UE.
[0034] – Policy Control Function (PCF) . The PCF provides QoS policy rules to control plane functions (CPFs) to enforce the rules. The PCF transforms the AF requests into Policy and Charging Control (PCC) rules that apply to PDU Sessions.
[0035] – Unified Data Management (UDM) . The UDM performs the generation of the 3rd Generation Partnership Project (3GPP) Authentication and Key Agreement (AKA) Authentication Credential, access authorization based on subscription data, UE's serving network function (NF) registration management (e.g. storing serving AMF for UE, storing serving SMF for UE's PDU session) , subscription management, etc. The UDM accesses the Unified Data Repository (UDR) to retrieve UE subscription data and store the UE context in the UDR. In some examples, the UDM and UDR can be deployed together.
[0036] 1.2 Network Data Analytics Function (NWDAF)
[0037] The Network Data Analytics Function (NWDAF) is a 5GC NF located on the control plane and performs statistical data analysis (e.g., the distribution information of the datasets) and machine learning (ML) -related tasks in 5GS. The NWDAF may interact with different entities for different purposes, e.g., data collection (as shown in FIG. 2) and data analytics exposure (as shown in FIG. 3) . Some of the various functions supported by the NWDAF include the following:
[0038] – Data collection based on subscription to events provided by AMF, SMF, UPF, PCF, UDM, Network Slice Admission Control Function (NSACF) , AF (directly or via Network Exposure Function (NEF) ) and OAM;
[0039] – Analytics and data collection using the Data Collection Coordination Function;
[0040] – Retrieval of information from data repositories (e.g. UDR via UDM for subscriber-related information or via NEF Packet Flow Detection function (PFDF) for Packet Flow Detection (PFD) information) ;
[0041] – Data collection of location information from the Location Services system;
[0042] – Storage and retrieval of information from Analytics Data Repository Function;
[0043] – Analytics and data collection from Messaging Framework Adaptor Function;
[0044] – Retrieval of information about NFs (e.g., from the Network Repository Function (NRF) for NF-related information) ;
[0045] – On-demand provision of analytics to consumers;
[0046] – Provision of bulked data related to analytics identifiers (IDs) ;
[0047] – Provision of accuracy information about analytics IDs; and / or
[0048] – Provision of ML model accuracy information or ML model accuracy degradation related to an ML model.
[0049] In existing implementations, a single instance or multiple instances of the NWDAF may be deployed in a Public Land Mobile Network (PLMN) . An NWDAF can contain the following logical functions:
[0050] – Analytics logical function (AnLF) : A logical function in NWDAF, which performs inference, derives analytics information (i.e., derives statistics and / or predictions based on analytics consumer request) and exposes analytics service i.e., Nnwdaf_AnalyticsSubscription or Nnwdaf_AnalyticsInfo.
[0051] – Model training logical function (MTLF) : A logical function in NWDAF, which trains machine learning (ML) models and exposes new training services (e.g., providing trained ML model) .
[0052] An NWDAF can contain an MTLF or an AnLF, or both logical functions.
[0053] 1.3 Location Management Function (LMF)
[0054] The LMF manages the overall coordination and scheduling of resources required for the location of a UE that is registered with or accessing 5GCN. It also calculates or verifies a final location and any velocity estimate and may estimate the achieved accuracy. The LMF receives location requests for a target UE from the serving AMF using the Nlmf interface. The LMF interacts with the UE in order to exchange location information applicable to UE assisted and UE based position methods and interacts with the NG-RAN, N3IWF or TNAN in order to obtain location information.
[0055] The LMF shall determine the result of the positioning in geographical coordinates and / or in local coordinates. If requested and if available, the positioning result may also include the velocity of the UE. The coordinate type (s) is determined by LMF when receiving a location request, based on LCS Client type and supported GAD shapes. If the location request indicates regulatory LCS Client type the LMF shall determine a geographical location and optionally a location in local coordinates. For location request indicates a value added LCS Client type, the LMF may determine the UE location in local coordinates or geographical coordinates or both. If the supported GAD shapes is not received or local coordinates is not included in the supported GAD shapes, the LMF shall determine a geographical location.
[0056] Additional functions which may be performed by an LMF to support location services include the following:
[0057] – Support a request for a single location, a periodic location, and / or a triggered location, received from a serving AMF for a target UE.
[0058] – Determine type and number of position methods and procedures based on UE and PLMN capabilities, QoS, UE connectivity state per access type, LCS Client type, coordinate type and optionally service type and indication of requiring reliable UE location information.
[0059] – Report UE location estimates directly to a Gateway Mobile Location Centre (GMLC) for periodic or triggered location of a target UE.
[0060] – Support cancelation of periodic or triggered location for a target UE.
[0061] – Support the provision of broadcast assistance data to UEs via NG-RAN in ciphered or un-ciphered form and forward any ciphering keys to subscribed UEs via the AMF.
[0062] – Support change of a serving LMF for periodic location reporting or triggered location reporting for a target UE.
[0063] – Support of receiving stored UE Positioning Capability from AMF and support of providing updated UE Positioning Capability to AMF.
[0064] – Map the UE location to a geographical area where the PLMN is or is not allowed to operate based on the request from AMF.
[0065] – Support determination of a UE location at a scheduled location time.
[0066] – Support determination of indoor or outdoor for a location estimate.
[0067] – Determine whether to use user plane or control plane for positioning.
[0068] – Support handling of 5GC-MT-LR, 5GC-NI-LR, 5GC-MO-LR and deferred 5GC-MT-LR for periodic or triggered location over a user plane connection between UE and LMF over TLS.
[0069] – Support collection of GNSS assistance data from AFs.
[0070] – Support service level PRU Association, PRU Association update, and / or PRU Disassociation.
[0071] – LMF supports verification of a PRU initiated Association or Disassociation by checking whether there is an PRU verified indication from AMF.
[0072] – LMF stores the received PRU information contained in service level PRU Association message and removes the PRU information after PRU Disassociation.
[0073] – LMF keeps PRU information for PRUs which are in OFF state.
[0074] – LMF may indicate support of PRU function to NRF via NF profile and may further send the PRU indication via NF profile update if PRU is stationary PRU.
[0075] – LMF may request a PRU to associate to a new LMF by returning a Routing ID of the new LMF.
[0076] Herein, PRU ON / OFF states indicate temporarily availability of the PRU functionality of a UE at the serving LMF (e.g., PRU OFF due to other high priority tasks / energy saving at the UE, or the UE temporarily loses network coverage) .
[0077] – Support selection of a PRU based on stored PRU information if the LMF needs to obtain the location measurements from the PRU to assist positioning of a target UE.
[0078] – Support to obtain PRU location measurements.
[0079] – Support to obtain PRU location measurements from other PRU serving LMF (s) .
[0080] – As a serving LMF of target UE (s) , support discovery and selection of other PRU serving LMF (s) by querying the NRF and support to request PRU location measurements from the selected LMF (s) .
[0081] – As a serving LMF of PRU (s) , support to provide PRU location measurements to other LMF (s) after receiving a request from other LMF (s) .
[0082] – Support to determine UE location using obtained PRU location measurements.
[0083] – Support a request for user plane reporting from a UE to an LCS Client or AF for a periodic or triggered 5GC-MT-LR. Subsequently, support the transfer of cumulative event reports from the target UE via control plane back to the H-GMLC and LCS Client or AF. Also support any request for assistance data received in a cumulative event report.
[0084] – Determine UE location for a UE connecting to a MBSR based on location and velocity of the MBSR and the timing of the location estimations for the target UE and MBSR.
[0085] – For a regulatory location service, support reporting of multiple INTERMEDIATE location estimates to GMLC.
[0086] 1.4 Operation, Administration, and Maintenance (OAM) function
[0087] The Operation, Administration, and Maintenance (OAM) function includes a set of processes and tools that is essential for ensuring the efficient and reliable functioning of telecommunication networks. OAM encompasses various activities aimed at managing network performance, diagnosing issues, and maintaining overall network health.
[0088] Operation. Day-to-day management of the network, including monitoring network performance, configuring network devices, and ensuring smooth functionality.
[0089] Administration. Involves tasks such as setting up user accounts, managing network resources, and enforcing security policies to ensure the network operates within defined parameters.
[0090] Maintenance. Entails regular and preventive activities to keep the network in optimal condition, such as software updates, hardware checks, and troubleshooting faults.
[0091] 1.5 Over-the-Top (OTT) Server
[0092] An OTT (Over-the-Top) server is a critical component in the ecosystem of delivering media and streaming services over the internet. OTT refers to the delivery of audio, video, and other media content directly to users over the internet, bypassing traditional cable or broadcast television platforms. Popular OTT services include Netflix, Amazon Prime Video, and Disney+. In some embodiments, the OTT sever may be an AF (Application Function) in 5GC.
[0093] 1.6 Examples of AI / ML positioning
[0094] In existing implementations, AI / ML-positioning includes:
[0095] Direct AI / ML positioning. Herein, the output of an AI / ML model (e.g., the UE location) is used in the positioning process. In other examples, direct AI / ML positioning includes fingerprinting based on channel observation as the input of AI / ML model.
[0096] AI / ML assisted positioning. Herein, the output of an AI / ML model (e.g., new measurements and / or enhancement of existing measurements) is used in the positioning process. In other examples, AI / ML assisted positioning includes line-of-sight (LOS) and non-line-of-sight (NLOS) LOS / NLOS identification, timing and / or angle of measurement, likelihood of measurement, etc.
[0097] It is noted that, as used in this patent document, data includes training data, which is the input for an ML model training operation.
[0098] 2 Example embodiments of the disclosed technology
[0099] Embodiments of the disclosed technology provide methods and procedures for a UE-requested (or UE-initiated) reporting or updating of its AI-related information (e.g., as described in Section 2.1 and the timing diagram in FIG. 4) , a network-initiated ML model update via the UP (e.g., as described in Section 2.2 and the timing diagram in FIG. 5) , and a UE-initiated ML model request / update or a request for training data (e.g., as described in Section 2.3 and the timing diagram in FIG. 6) .
[0100] 2.1 UE-initiated reporting and updating of AI-related information
[0101] FIG. 4 is a timing diagram of an example method for UE reporting or updating of AI-related information. As shown therein, this example method includes the following operations:
[0102] 1. The UE sends or updates its AI-related information in the registration message to RAN. The AI-related information includes UE ID, UE AI configuration information, an acceptable ML model type (e.g., neural network, support vector machine (SVM) , random forest (RF) ) , maximum size of ML model file, ML model parameter dimension, a maximum number of ML model parameters per model, UE location information, UE computing power information, ML model complexity information (e.g., maximum restriction of model complexity) , ML model running environment information, an inference input information of the ML model, the energy consumption information of the ML model (e.g., the maximum energy (or cost) to train or perform inference using the ML model) .
[0103] 2. RAN forward the UE AI-related information to AMF.
[0104] 3. AMF may store the UE AI-related information. To store UE AI-related information, AMF first sends UE AI-related information to UDM, and then UDM stores the AI-related information to UDR.
[0105] 4. If AMF did not store the UE AI-related information, it may forward the information to UDM or UDR.
[0106] 5. The UDM or UDR may store the UE-AI related information.
[0107] 6. The UDM or UDR sends a response (storage success / failure) to the AMF
[0108] 7. AMF sends the registration accept message to RAN.
[0109] 8. RAN forwards the registration accept message to UE.
[0110] 2.2 Network-initiated ML model update via UP
[0111] FIG. 5 is a timing diagram of an example method for a network-initiated ML model update via the UP. Herein, the purpose of LMF is to report its capability of building UP for different purposes, support user plane (UP) transfer of AI-related information, support selection / re-selection of appropriate LMF to obtain appropriate ML model / data for a UE-initiated ML model request or network-initiated ML Model delivery / update, UE (training) data collection, or data delivery. As shown therein, this example method includes the following operations:
[0112] 1. The LMF registers / updates its NF profile to NRF. The NF profile includes the capability of building UP for different purposes (e.g., ML model, ML model delivery / update, UE (training) data collection, or data delivery) and / or support for user plane (UP) transfer of AI-related information (e.g. ML model information, (training) data) .
[0113] 2. The NRF stores the NF profile of LMF.
[0114] 3. The NRF sends a response message, which includes a success / failure indication of the storage operation, to the LMF.
[0115] 2.3 UE-initiated ML model transfer, data transfer, UE data collection
[0116] FIG. 6 is a timing diagram of an example method for a UE-initiated request for an ML model or update, and / or a request for training data. As shown therein, this example method includes the following operations:
[0117] 1. The UE sends an ML model request, ML model update request, or (training) data transfer and / or collection request to AMF via UL NAS MM (mobility management) message. The UE may specify the UE AI related information (e.g., UE ID, UE AI configuration information, required ML Model type (e.g., neural network, support vector machine (SVM) , random forest (RF) ) , maximum size of ML model file, ML model parameter dimension, maximum number of ML model parameter per model, time-related information (e.g., when the ML model is needed, when the data can be delivered to UE, when the (training) data collection can be performed, etc. ) , ML model filter information (e.g., Area of Interest, S-NSSAI) , UE location information, UE computing power information, the ML model complexity information (e.g., maximum restriction of model complexity) , ML model accuracy information, ML model metrics requirement, ML model running environment information, Analytics ID, or the use of ML model (e.g., for direct AI / ML positioning or AI / ML assisted positioning) , the inference input information of the ML model, the energy consumption information of the ML model (e.g., the maximum energy (or cost) to train or perform inference using the ML model) ) in the message. The UE may also specify whether to use the user plane (UP) or control plane (CP) for ML model delivery and / or (training) data delivery.
[0118] 2. The AMF selects an appropriate LMF to perform the task, e.g., based on a local configuration or querying the NRF.
[0119] 3. The AMF forwards the message to the selected LMF.
[0120] 4. The LMF sends user plane information and the confirmation of requested purpose of using UP (e.g., (training) data transfer to UE, ML model update / transfer) . The user plane information includes the user plane address of the LMF, and security information related to the ML model transfer or data transfer. If the LMF intends to deliver ML model based on UE request, it may include the use of ML model, e.g., the ML Model is to be used for AI / ML positioning (e.g., direct AI / ML positioning or AI / ML assisted positioning) and / or the ML model related Analytics ID. In some examples, there is a one-to-one mapping between the ML model and the use of the ML model.
[0121] 5. The AMF forwards the message to UE via DL NAS MM message.
[0122] 6. The UE may send an acknowledgement to LMF through AMF, to acknowledge the use of user plane connection for (training) data transfer / collection, ML model transfer, or perform (training) data collection from UE. The UE may send a rejection message that includes the reason (s) for rejecting (e.g., the rejection cause value) the building of the UP (e.g., the UE does not support UP building, the UE is temporarily unavailable, overloaded state (e.g., system overload or device overload) . In this message, UE may also specify its AI-related information if it does not send this information in step 1.
[0123] 7. The AMF forwards the message to LMF.
[0124] 8. The UE establishes a secured user plane connection with LMF. In some embodiments, a single UP can be used for more than one purpose (e.g., ML model transfer and UE (training) data collection or data delivery to UE) . The UE may send acknowledgement described in step 6 after establishing the user plane connection.
[0125] 9. If the purpose of building user plane is ML model delivery, the LMF prepares (or configures) the ML Model.
[0126] 10. The LMF may train the ML Model itself, or request an ML model from NWDAF, OAM, OTT server, or another LMF. If the purpose of building user plane is data transfer, LMF may collect data from other 5GC NF, OAM, or RAN, before it sends the collected data to UE via UP. LMF may select appropriate ML model based on UE AI-related information on its own, or discover another LMF or NWDAF from NRF with UE AI-related information, and then request the ML model from NWDAF or another LMF.
[0127] 11. The LMF delivers data or ML model information (e.g., the ML model file, ML model file size, ML model execution environment, inference input data information, a URL / FQDN link containing ML model or ML model parameters, number of parameters in the ML model, the ML model complexity, the estimated energy consumption information related to the ML Model, Area of Interest (Tracking Area, cell or below cell) information related to the ML Model) to UE via UP.
[0128] 2.4 Example methods and implementations
[0129] In some embodiments, and in the context of the embodiments described in Section 2.1 through 2.3, the User Equipment (UE) is configured to perform the following operations:
[0130] – Request an ML model, ML model update, and / or (training) data from LMF
[0131] – Send UP building purpose acknowledgement / rejection to LMF
[0132] – Report UE AI-related information to AMF, UDM, and / or UDR in the registration message In some embodiments, and in the context of the embodiments described in Section 2.1 through 2.3, the AMF, UDM, and / or UDR is configured to receive, from the UE, and store the AI-related information.
[0133] In some embodiments, and in the context of the embodiments described in Section 2.1 through 2.3, the Location Management Function (LMF) , or another similarly-configured network function, is configured to perform the following operations:
[0134] – Register or update its capability of building user plane for different purposes
[0135] – Prepare ML model information and (training) data
[0136] – Send (training) data or ML model information to UE upon receiving a request
[0137] In some embodiments, and in the context of the embodiments described in Section 2.1 through 2.3, the NRF is configured to receive, from the LMF, and store the capability of building the user plane for different purposes.
[0138] FIG. 7 shows a flowchart for an example wireless communication method 700. The method 700 includes, at operation 710, transmitting, by a wireless device to a control plane network function, a registration message comprising artificial intelligence (AI) -related information for the wireless device.
[0139] The method 700 includes, at operation 720, receiving, from the control plane network function, a registration response message.
[0140] FIG. 8 shows a flowchart for an example wireless communication method 800. The method 800 includes, at operation 810, transmitting, by a network function to a network storage, a network function profile comprising a capability for establishing a user plane between the network function and a wireless device for different artificial intelligence (AI) -related purposes.
[0141] The method 800 includes, at operation 820, receiving, from the network storage, a response message comprising a success or failure indication associated with the network storage performing a storage operation on the network function profile.
[0142] FIG. 9 shows a flowchart for an example wireless communication method 900. The method 900 includes, at operation 910, transmitting, by a wireless device to a network function, a first request message to obtain at least one of an information for a machine learning (ML) model, an updated information for the ML model, or training data for the ML model over a network plane specified by the wireless device.
[0143] The method 900 includes, at operation 920, receiving, in response to the first request message, a response message via the network plane specified by the wireless device. In this example, the response message comprises the information for the ML model, the updated information for the ML model, or the training data for the ML model.
[0144] The method 900 includes, at operation 930, performing, based on the response message, an ML model training operation or an ML model inference operation.
[0145] The described features can be implemented to further provide one or more of the following technical solutions:
[0146] 1. A wireless communication method, comprising: transmitting, by a wireless device to a control plane network function, a registration message comprising artificial intelligence (AI) -related information for the wireless device; and receiving, from the control plane network function, a registration response message.
[0147] 2. The method of solution 1, wherein the wireless device and the control plane network function are communicatively coupled through a radio access network.
[0148] 3. The method of solution 1, wherein the AI-related information comprises at least one of an identifier (ID) of the wireless device, AI configuration information, a model type of a machine learning (ML) model, a maximum size of the ML model, a parameter dimension of the ML model, a maximum number of parameters of the ML model, time-related information, filter information for the ML model, location information for the wireless device, computing power information of the wireless device, complexity information of the ML model, accuracy information for the ML model, ML model metrics information, running environment information for the ML model, a use of the ML model by the wireless device, an inference input information of the ML model, an energy consumption information of the ML model, a use for the ML model, or an analytics ID of the ML model.
[0149] 4. The method of any of solutions 1 to 3, wherein the control plane network function is configured to store the AI-related information.
[0150] 5. The method of any of solutions 1 to 3, wherein the control plane network function is configured to: receive the registration message comprising the AI-related information, and transmit, upon determining that the control plane network function will not store the AI-related information, the AI-related information to a data management network function or a data repository function, and wherein the data management network function or the data repository function is configured to: perform a storage operation on the AI-related information, and transmit, to the control plane network function, a success or failure indication for the storage operation.
[0151] 6. The method of solution 5, wherein the control plane network function is an Access and Mobility Management Function (AMF) , the data management network function is a Unified Data Management (UDM) , and the data repository function is a Unified Data Repository (UDR) .
[0152] 7. A wireless communication method, comprising: transmitting, by a network function to a network storage, a network function profile comprising a capability for establishing a user plane between the network function and a wireless device for different artificial intelligence (AI) -related purposes; and receiving, from the network storage, a response message comprising a success or failure indication associated with the network storage performing a storage operation on the network function profile.
[0153] 8. The method of solution 7, wherein the different AI-related purposes for establishing the user plane comprise delivering an information for a machine learning (ML) model, delivering an updated information for the ML model, and / or transferring training data for the ML model.
[0154] 9. The method of solution 8, wherein the user plane is configured to be used for one or more purposes of the different AI-related purposes.
[0155] 10. The method of any of solutions 7 to 9, wherein the network function is a Location Management Function (LMF) and the network storage is a Network Repository Function (NRF) .
[0156] 11. A wireless communication method, comprising: transmitting, by a wireless device to a network function, a first request message to obtain at least one of an information for a machine learning (ML) model, an updated information for the ML model, or training data for the ML model over a network plane specified by the wireless device; receiving, in response to the first request message, a response message via the network plane specified by the wireless device, wherein the response message comprises the information for the ML model, the updated information for the ML model, or the training data for the ML model; and performing, based on the response message, an ML model training operation or an ML model inference operation.
[0157] 12. The method of solution 11, wherein the first request message comprises a use of the ML model that includes (a) direct AI / ML positioning or (b) AI / ML assisted positioning.
[0158] 13. The method of solution 11, wherein the network plane specified by the wireless device comprises a user plane, a data plane, or a control plane.
[0159] 14. The method of solution 13, wherein, when the first request message specifies the network plane being the user plane, the network function is configured to transmit, to the wireless device, a second request message for establishing the user plane between the wireless device and the network function.
[0160] 15. The method of solution 14, wherein the wireless device is configured to transmit, in response to the second request message, an acknowledgement message.
[0161] 16. The method of solution 15, wherein the acknowledgement message or the first request message comprises artificial intelligence (AI) -related information for the wireless device.
[0162] 17. The method of solution 14, wherein the wireless device is configured to transmit, in response to the second request message, a rejection message including a reason for rejecting an establishment of the user plane.
[0163] 18. The method of solution 17, wherein the reason comprises the wireless device not supporting the establishment of the user plane, the user plane being temporarily unavailable, the wireless device being overloaded, a system of the wireless device being overloaded, or a system of the wireless device lacking a requisite computing power.
[0164] 19. The method of solution 13, wherein wireless device and the network function are communicatively coupled through a control plane network function configured to: receive, from the wireless device, the first request message; select the network function from a plurality of network functions; and transmit, to the network function, the first request message.
[0165] 20. The method of solution 19, wherein the control plane network function is an Access and Mobility Management Function (AMF) that is configured to query a Network Repository Function (NRF) to determine the plurality of network functions.
[0166] 21. The method of solution 13, wherein, when the first request message comprises a request for the information for the ML model, the network function is configured to generate an information for a trained ML model.
[0167] 22. The method of solution 21, wherein the network function is configured to generate the information for the trained ML model.
[0168] 23. The method of solution 21, wherein the network function is configured to request, based on information from a network storage, the information for the trained ML model from an alternative network function.
[0169] 24. The method of solution 23, wherein the network storage is a Network Repository Function (NRF) and the alternative network function is a network data analytics function (NWDAF) , an operation, administration, and maintenance (OAM) network function, an over-the-top (OTT) server, an Application Function (AF) , or an alternative Location Management Function (LMF) .
[0170] 25. The method of any of solutions 11 to 24, wherein the network function is a Location Management Function (LMF) .
[0171] 26. The method of any of solutions 11 to 24, wherein the information of the ML model comprises at least one of an ML model file, an ML model file size, an ML model execution environment, an inference input information, parameters of the ML model, a uniform resource locator (URL) / fully qualified domain name (FQDN) link containing the ML model or the parameters of the ML model, a number of parameters of the ML model, a complexity information of the ML model, an estimated energy consumption information of the ML model, an area of interest information related to the ML model, values of ML model metrics, used ML model metrics, accuracy information for the ML model, an analytics identifier (ID) related to the ML Model, or a use of the ML Model.
[0172] 27. An apparatus for wireless communication comprising one or more processors, configured to cause the apparatus to implement the method recited in one or more of solutions 1 to 26.
[0173] 28. A non-transitory computer readable program storage medium having code stored thereon, the code, when executed by one or more processors, causing the one or more processors to implement the method recited in one or more of solutions 1 to 26.
[0174] FIG. 10 shows a block diagram of an example hardware platform 1000 that may be a part of a network device (e.g., base station) or a communication device (e.g., a user equipment (UE) ) . The hardware platform 1000 includes at least one processor 1010 and a memory 1005 having instructions stored thereupon. The instructions upon execution by the processor 1010 configure the hardware platform 1000 to perform the operations described in FIGS. 4 to 9 and in the various embodiments described in this patent document. The transmitter 1015 transmits or sends information or data to another device. For example, a network device transmitter can send a message to a user equipment. The receiver 1020 receives information or data transmitted or sent by another device. For example, a user equipment can receive a message from a network device.
[0175] The implementations as discussed above will apply to a wireless communication. FIG. 11 shows an example of a wireless communication system (e.g., a 5G or NR cellular network) that includes a base station 1120 and one or more user equipment (UE) 1111, 1112 and 1113. In some embodiments, the UEs access the BS (e.g., the network) using a communication link to the network (sometimes called uplink direction, as depicted by dashed arrows 1131, 1132, 1133) , which then enables subsequent communication (e.g., shown in the direction from the network to the UEs, sometimes called downlink direction, shown by arrows 1141, 1142, 1143) from the BS to the UEs. In some embodiments, the BS send information to the UEs (sometimes called downlink direction, as depicted by arrows 1141, 1142, 1143) , which then enables subsequent communication (e.g., shown in the direction from the UEs to the BS, sometimes called uplink direction, shown by dashed arrows 1131, 1132, 1133) from the UEs to the BS. The UE may be, for example, a smartphone, a tablet, a mobile computer, a machine to machine (M2M) device, an Internet of Things (IoT) device, and so on.
[0176] Some of the embodiments described herein are described in the general context of methods or processes, which may be implemented in one embodiment by a computer program product, embodied in a computer-readable medium, including computer-executable instructions, such as program code, executed by computers in networked environments. A computer-readable medium may include removable and non-removable storage devices including, but not limited to, Read Only Memory (ROM) , Random Access Memory (RAM) , compact discs (CDs) , digital versatile discs (DVD) , etc. Therefore, the computer-readable media can include a non-transitory storage media. Generally, program modules may include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. Computer-or processor-executable instructions, associated data structures, and program modules represent examples of program code for executing steps of the methods disclosed herein. The particular sequence of such executable instructions or associated data structures represents examples of corresponding acts for implementing the functions described in such steps or processes.
[0177] Some of the disclosed embodiments can be implemented as devices or modules using hardware circuits, software, or combinations thereof. For example, a hardware circuit implementation can include discrete analog and / or digital components that are, for example, integrated as part of a printed circuit board. Alternatively, or additionally, the disclosed components or modules can be implemented as an Application Specific Integrated Circuit (ASIC) and / or as a Field Programmable Gate Array (FPGA) device. Some implementations may additionally or alternatively include a digital signal processor (DSP) that is a specialized microprocessor with an architecture optimized for the operational needs of digital signal processing associated with the disclosed functionalities of this application. Similarly, the various components or sub-components within each module may be implemented in software, hardware or firmware. The connectivity between the modules and / or components within the modules may be provided using any one of the connectivity methods and media that is known in the art, including, but not limited to, communications over the Internet, wired, or wireless networks using the appropriate protocols.
[0178] While this document contains many specifics, these should not be construed as limitations on the scope of an invention that is claimed or of what may be claimed, but rather as descriptions of features specific to particular embodiments. Certain features that are described in this document in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable sub-combination. Moreover, although features may be described above as acting in certain combinations and even initially claimed as such, one or more features from a claimed combination can in some cases be excised from the combination, and the claimed combination may be directed to a sub-combination or a variation of a sub-combination. Similarly, while operations are depicted in the drawings in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results.
[0179] Only a few implementations and examples are described and other implementations, enhancements and variations can be made based on what is described and illustrated in this disclosure.
Claims
1.A wireless communication method, comprising:transmitting, by a wireless device to a control plane network function, a registration message comprising artificial intelligence (AI) -related information for the wireless device; andreceiving, from the control plane network function, a registration response message.2.The method of claim 1, wherein the wireless device and the control plane network function are communicatively coupled through a radio access network.3.The method of claim 1, wherein the AI-related information comprises at least one of an identifier (ID) of the wireless device, AI configuration information, a model type of a machine learning (ML) model, a maximum size of the ML model, a parameter dimension of the ML model, a maximum number of parameters of the ML model, time-related information, filter information for the ML model, location information for the wireless device, computing power information of the wireless device, complexity information of the ML model, accuracy information for the ML model, ML model metrics information, running environment information for the ML model, a use of the ML model by the wireless device, an inference input information of the ML model, an energy consumption information of the ML model, a use for the ML model, or an analytics ID of the ML model.4.The method of any of claims 1 to 3, wherein the control plane network function is configured to store the AI-related information.5.The method of any of claims 1 to 3,wherein the control plane network function is configured to:receive the registration message comprising the AI-related information, andtransmit, upon determining that the control plane network function will not store the AI-related information, the AI-related information to a data management network function or a data repository function, andwherein the data management network function or the data repository function is configured to:perform a storage operation on the AI-related information, andtransmit, to the control plane network function, a success or failure indication for the storage operation.6.The method of claim 5, wherein the control plane network function is an Access and Mobility Management Function (AMF) , the data management network function is a Unified Data Management (UDM) , and the data repository function is a Unified Data Repository (UDR) .7.A wireless communication method, comprising:transmitting, by a network function to a network storage, a network function profile comprising a capability for establishing a user plane between the network function and a wireless device for different artificial intelligence (AI) -related purposes; andreceiving, from the network storage, a response message comprising a success or failure indication associated with the network storage performing a storage operation on the network function profile.8.The method of claim 7, wherein the different AI-related purposes for establishing the user plane comprise delivering an information for a machine learning (ML) model, delivering an updated information for the ML model, and / or transferring training data for the ML model.9.The method of claim 8, wherein the user plane is configured to be used for one or more purposes of the different AI-related purposes.10.The method of any of claims 7 to 9, wherein the network function is a Location Management Function (LMF) and the network storage is a Network Repository Function (NRF) .11.A wireless communication method, comprising:transmitting, by a wireless device to a network function, a first request message to obtain at least one of an information for a machine learning (ML) model, an updated information for the ML model, or training data for the ML model over a network plane specified by the wireless device;receiving, in response to the first request message, a response message via the network plane specified by the wireless device, wherein the response message comprises the information for the ML model, the updated information for the ML model, or the training data for the ML model; andperforming, based on the response message, an ML model training operation or an ML model inference operation.12.The method of claim 11, wherein the first request message comprises a use of the ML model that includes (a) direct AI / ML positioning or (b) AI / ML assisted positioning.13.The method of claim 11, wherein the network plane specified by the wireless device comprises a user plane, a data plane, or a control plane.14.The method of claim 13, wherein, when the first request message specifies the network plane being the user plane, the network function is configured to transmit, to the wireless device, a second request message for establishing the user plane between the wireless device and the network function.15.The method of claim 14, wherein the wireless device is configured to transmit, in response to the second request message, an acknowledgement message.16.The method of claim 15, wherein the acknowledgement message or the first request message comprises artificial intelligence (AI) -related information for the wireless device.17.The method of claim 14, wherein the wireless device is configured to transmit, in response to the second request message, a rejection message including a reason for rejecting an establishment of the user plane.18.The method of claim 17, wherein the reason comprises the wireless device not supporting the establishment of the user plane, the user plane being temporarily unavailable, the wireless device being overloaded, a system of the wireless device being overloaded, or a system of the wireless device lacking a requisite computing power.19.The method of claim 13, wherein wireless device and the network function are communicatively coupled through a control plane network function configured to:receive, from the wireless device, the first request message;select the network function from a plurality of network functions; andtransmit, to the network function, the first request message.20.The method of claim 19, wherein the control plane network function is an Access and Mobility Management Function (AMF) that is configured to query a Network Repository Function (NRF) to determine the plurality of network functions.21.The method of claim 13, wherein, when the first request message comprises a request for the information for the ML model, the network function is configured to generate an information for a trained ML model.22.The method of claim 21, wherein the network function is configured to generate the information for the trained ML model.23.The method of claim 21, wherein the network function is configured to request, based on information from a network storage, the information for the trained ML model from an alternative network function.24.The method of claim 23, wherein the network storage is a Network Repository Function (NRF) and the alternative network function is a network data analytics function (NWDAF) , an operation, administration, and maintenance (OAM) network function, an over-the-top (OTT) server, an Application Function (AF) , or an alternative Location Management Function (LMF) .25.The method of any of claims 11 to 24, wherein the network function is a Location Management Function (LMF) .26.The method of any of claims 11 to 24, wherein the information of the ML model comprises at least one of an ML model file, an ML model file size, an ML model execution environment, an inference input information, parameters of the ML model, a uniform resource locator (URL) / fully qualified domain name (FQDN) link containing the ML model or the parameters of the ML model, a number of parameters of the ML model, a complexity information of the ML model, an estimated energy consumption information of the ML model, an area of interest information related to the ML model, values of ML model metrics, used ML model metrics, accuracy information for the ML model, an analytics identifier (ID) related to the ML Model, or a use of the ML Model.27.An apparatus for wireless communication comprising one or more processors, configured to cause the apparatus to implement the method recited in one or more of claims 1 to 26.28.A non-transitory computer readable program storage medium having code stored thereon, the code, when executed by one or more processors, causing the one or more processors to implement the method recited in one or more of claims 1 to 26.
Citation Information
Patent Citations
Information sending method and device, information receiving method and device, terminal and network side equipment
CN118158812A
Communication method, core network device, terminal, communication system and storage medium
CN118202675A
Method and apparatus for determining machine learning model based on network congestion information in wireless communication system
WO2023214752A1
Method and apparatus for supporting federated learning in wireless communication system
WO2023214806A1