Supporting federated learning in a communication system

Abstract

Various aspects of this disclosure relate to a network entity for supporting wireless communication for network data analysis, the network entity including instructions executable by at least one processor to cause the network entity to: receive a request message for providing inference output for a network analysis operation identified by an analysis identifier; output the inference request message to each of a set of one or more model training functions, the set of model training functions being participants in a federated learning training process for a trained model corresponding to the network analysis operation; receive inference output from each function of the set; and output a response message including an aggregated inference output based on a corresponding contribution weight of each of the functions in the federated learning training process.

Description

Technical Field

[0001] This disclosure relates to wireless communications, and more specifically, to network analysis in wireless communication networks. Background Technology

[0002] A wireless communication system may include one or more network communication devices, such as base stations, which may support wireless communication with one or more user communication devices, also referred to as user equipment (UE) or other suitable terms. The wireless communication system can support wireless communication with one or more user communication devices by utilizing the resources of the wireless communication system (e.g., time resources (e.g., symbols, time slots, subframes, frames, or the like) or frequency resources (e.g., subcarriers, carriers, or the like)). Furthermore, the wireless communication system may support wireless communication across a variety of radio access technologies, including third-generation (3G), fourth-generation (4G), fifth-generation (5G), and other suitable radio access technologies beyond 5G (e.g., sixth-generation (6G)). Summary of the Invention

[0003] The article “a” preceding an element is not limited and should be understood to refer to “at least one” or “one or more” of those elements. The terms “a,” “at least one,” “one or more,” and “at least one of one or more” are interchangeable. As used herein (including in the claims), “or” as used in a list of items (e.g., a list of items beginning with phrases such as “at least one of…” or “one or more of…” or “one or both of…”) indicates an inclusive list such that a list of at least one of, for example, A, B, or C represents A or B or C or AB or AC or BC or ABC (i.e., A and B and C). Furthermore, as used herein, the phrase “based on” should not be construed as a reference to a closed set of conditions. For example, an example step described as “based on condition A” may be based on both condition A and condition B without departing from the scope of this disclosure. In other words, as used herein, the phrase “based on” should be interpreted in the same manner as the phrase “at least partially based on.” Furthermore, as used herein (including in the claims), a “set” may comprise one or more elements.

[0004] Some embodiments of the methods and apparatus described herein may further include a network entity for supporting wireless communication for network data analysis, the network entity including instructions executable by at least one processor to cause the network entity to: receive a request message for providing inference output for a network analysis operation identified by an analysis identifier; output the inference request message to each of a set of one or more model training functions, the set of model training functions being participants in a federated learning training process for a trained model corresponding to the network analysis operation; receive inference output from each function of the set; and output a response message including an aggregated inference output based on a corresponding contribution weight of each of the functions in the federated learning training process.

[0005] The instruction enables the network entity to: obtain a model information message from a second network entity, the model information message including: an identifier of the model; an identifier of one or more functions participating in the federated learning training process; and a coordination identifier associated with the current example of the federated learning training process.

[0006] The model information message may include information identifying each of the model training functions as having one or more roles in the federated training process, wherein the one or more roles are one or more of a group including federated learning server, federated learning active participant and federated learning passive participant.

[0007] One or more of the model training functions may include Network Data Analysis Functions (NWDAF). Alternatively or additionally, one or more of the model training functions may include Application Functions (AF).

[0008] The instruction enables the network entity to obtain the model information message based on the analysis identifier.

[0009] The process of obtaining the model information message may include outputting a model information request message, including the analysis identifier, to the second network entity.

[0010] The second network entity may include at least one of a network repository function and an analytics data repository function.

[0011] The network entity may include at least one of the following: a network data analysis function (NWDAF) that supports analysis logic functions; and an NWDAF configured to control the federated learning process for the analysis identifier.

[0012] The instruction enables the network entity to receive the request message from a consumer entity within the communication network.

[0013] The consumer entity can provide consumer analysis logic functions.

[0014] The instruction enables the network entity to output the response message to the consumer entity.

[0015] The inference output may be the aggregated inference output, and the instruction may enable the network entity to obtain the aggregated inference output from the third network entity.

[0016] The third network entity may include a Model Training Logic Function (MTLF), which is a controller for the federated learning process of the model.

[0017] Alternatively or additionally, the third network entity may include AF.

[0018] The inference output may include a corresponding interference output from each of the one or more functions. The instructions enable the network entity to: obtain each of the interference outputs from the corresponding functions of the one or more functions; and determine the aggregated inference output based on each of the inference outputs.

[0019] The instruction enables the network entity to: obtain the corresponding contribution weight; and further determine the aggregate inference output based on the corresponding contribution weight.

[0020] The one or more network analysis operations may include at least one statistical analysis operation to be performed on the network.

[0021] The one or more network analysis operations may include at least one prediction operation performed regarding the behavior of one or more entities within the communication network.

[0022] The prediction operation may include at least one of the following: predicting a parameter associated with at least one entity within the communication network, the parameter optionally being the load of the at least one entity, the at least one entity optionally including at least one of a network entity of the communication network and a UE; and predicting the mobility characteristics of a UE in the communication network, the mobility characteristics optionally including the location of the UE.

[0023] Each of the one or more functions may be a network function for providing analytical data associated with one or more UEs within the communication network.

[0024] In some embodiments of the methods and apparatus described herein, one method is performed by a network entity of a wireless communication network, the method comprising: obtaining a request message for providing inference output for a network analysis operation identified by an analysis identifier; outputting the inference request message to each of a set of one or more model training functions, the set of model training functions being participants in a federated learning training process for a trained model corresponding to the network analysis operation; obtaining inference output from each function of the set; and outputting a response message including an aggregated inference output based on a corresponding contribution weight of each of the functions in the federated learning training process.

[0025] In some embodiments of the methods and apparatus described herein, a communication network system includes: a first network entity for wireless communication in the communication network; and a second network entity for wireless communication in the communication network, wherein: the first network entity is configured to: receive a first request message to provide inference output for a network analysis operation identified by an analysis identifier; output a second request message to the second network entity to perform an inference operation with respect to the network analysis operation, the second network entity being configured to: output a third request message to each of one or more functions participating in a federated learning training process for performing the network analysis operation; obtain inference output from each of the one or more functions; output an aggregated inference output to the first network entity based on the obtained inference output and a corresponding contribution weight of each of the functions in the federated learning training process, and the first network entity is configured to output a response message including the aggregated inference output.

[0026] The second network entity may be configured to provide training data (e.g., gradients and / or losses) to one or more functions based on the inference output for training the model.

[0027] The first network entity may be an NWDAF that supports AnLF.

[0028] The second network entity may be a federated learning server. Attached Figure Description

[0029] Figure 1 Examples of wireless communication systems according to aspects of this disclosure are described.

[0030] Figure 2 and 3 Describe the network architecture.

[0031] Figure 4A and 4B Explain federated learning technology.

[0032] Figure 5A and5B Describe the communication process of federated learning.

[0033] Figure 6 Describe the architecture of instance-based federated learning.

[0034] Figure 7A and 7B Describe the process flow of instance-based federated learning.

[0035] Figure 8 Examples of network equipment (NE) 800 according to aspects of this disclosure are described.

[0036] Figure 9 A flowchart illustrating the method performed by NE according to aspects of this disclosure.

[0037] Figure 10 A communication network according to aspects of this disclosure is schematically depicted. Detailed Implementation

[0038] Wireless communication systems comprising one or more network entities may support network analysis operations (e.g., processes, actions), such as statistical analysis of activities within these wireless communication systems. For example, analysis may be performed on parameters such as signal quality, traffic volume, and the like. In some other instances, network analysis operations may include one or more network entities performing predictions (e.g., estimations, determinations, forecasts) of characteristics of the wireless communication system (e.g., a network). For example, one or more network entities may perform predictions of the location (e.g., position) of a user equipment (UE) at a given time, or predictions of the load (e.g., data traffic) of one or more network entities at a future given time.

[0039] This disclosure provides improvements in the performance (e.g., accuracy and / or processing) of such network analysis. Various aspects of this disclosure relate to performing network analysis using one or more learning techniques (e.g., vertical federated learning techniques). Federated learning techniques, described in more detail below, are techniques in which multiple entities collaboratively train a model without centralized data. It should be understood that other learning techniques (e.g., machine learning models, artificial intelligence models, or the like) can be used to support one or more network entities and enable one or more network entities to support network analysis. Some aspects of this disclosure relate to federated learning techniques (e.g., processes) controlled by one or more network entities to efficiently provide improved network analysis. This control, described in more detail below, may include sending messages to participants in the process to coordinate their training of the model. Such use of learning techniques can provide improved statistical analysis and improved prediction quality, thereby improving the accuracy of network analysis. Improved network performance can thus be achieved.

[0040] Aspects of this disclosure are described in the context of wireless communication systems.

[0041] Figure 1 This describes an example of a wireless communication system 100 according to aspects of this disclosure. The wireless communication system 100 may include one or more NEs 102, one or more UEs 104, and a core network (CN) 106. The wireless communication system 100 may support various radio access technologies. In some embodiments, the wireless communication system 100 may be a 4G network, such as an LTE network or an advanced LTE (LTE-A) network. In some other embodiments, the wireless communication system 100 may be an NR network, such as a 5G network, an advanced 5G (5G-A) network, or a 5G ultra-wideband (5G-UWB) network. In other embodiments, the wireless communication system 100 may be a combination of 4G and 5G networks, or include other suitable radio access technologies such as IEEE 802.11 (Wi-Fi), IEEE 802.16 (WiMAX), and IEEE 802.20. The wireless communication system 100 may support radio access technologies beyond 5G, such as 6G. In addition, the wireless communication system 100 can support technologies such as Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), or Code Division Multiple Access (CDMA).

[0042] One or more NEs 102 may be distributed across a geographical area to form a wireless communication system 100. One or more of the NEs 102 described herein may be, include, or be referred to as a network node, base station, network element, network function, network entity, radio access network (RAN), NodeB, eNodeB (eNB), next-generation NodeB (gNB), or other suitable terms. NEs 102 and UE 104 may communicate via a communication link, which may be a wireless or wired connection. For example, NEs 102 and UE 104 may perform wireless communication (e.g., receive signaling, transmit signaling) via a Uu interface.

[0043] NE 102 can provide a geographic coverage area, for which NE 102 can support services of one or more UEs 104 within the geographic coverage area. For example, NE 102 and UE 104 can support wireless communication of signals related to services (e.g., voice, video, packet data, messaging, broadcasting, etc.) according to one or more radio access technologies. In some embodiments, NE 102 can be mobile, for example, a satellite associated with a non-terrestrial network (NTN). In some embodiments, different geographic coverage areas 112 associated with the same or different radio access technologies can overlap, but different geographic coverage areas can be associated with different NEs 102.

[0044] One or more UEs 104 may be distributed across a geographical area of ​​the wireless communication system 100. UE 104 may include or be referred to as a remote unit, mobile device, wireless device, remote device, subscriber device, transmitter device, receiver device, or some other suitable term. In some embodiments, UE 104 may be referred to as a unit, station, terminal, or client, and other instances thereof. Alternatively or additionally, UE 104 may be referred to as an Internet of Things (IoT) device, an Internet of Everything (IoE) device, or a Machine Type Communication (MTC) device, and other instances thereof.

[0045] UE 104 may support direct wireless communication with other UE 104 via a communication link. For example, UE 104 may support direct wireless communication with another UE 104 via a device-to-device (D2D) communication link. In some implementations (e.g., vehicle-to-vehicle (V2V) deployment, vehicle-to-everything (V2X) deployment, or cellular-V2X deployment), communication link 114 may be referred to as a side link. For example, UE 104 may support direct wireless communication with another UE 104 via a PC5 interface.

[0046] NE 102 may support communication with CN 106 or with another NE 102, or both. For example, NE 102 may interface with other NE 102 or CN 106 via one or more backhaul links (e.g., S1, N2, N2, or network interfaces). In some embodiments, NE 102 may communicate directly with each other. In some other embodiments, NE 102 may communicate with each other or indirectly (e.g., via CN 106). In some embodiments, one or more NE 102 may include sub-components, such as access network entities, which may be instances of access node controllers (ANCs). The ANC may communicate with one or more UE 104s via one or more other access network transmitting entities (which may be referred to as radio heads, smart radio heads, or transmit-receive points (TRPs)).

[0047] CN 106 can support user authentication, access authorization, tracking, connectivity, and other access, routing, or mobility functions. CN 106 can be an evolved packet core (EPC) or a 5G core (5GC), which may include control plane entities that manage access and mobility (e.g., a mobility management entity (MME), access and mobility management functions (AMF)) and user plane entities that route or interconnect packets 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 entities may manage non-access plane (NAS) functions, such as mobility, authentication, and bearer management (e.g., data bearers, signaling bearers, etc.) of one or more UEs 104 served by one or more NEs 102 associated with CN 106.

[0048] CN 106 can communicate with the packet data network via one or more backhaul links (e.g., via S1, N2, N2, or another network interface). The packet data network may contain an application server. In some implementations, one or more UEs 104 can communicate with the application server. UE 104 can establish a session (e.g., a Protocol Data Unit (PDU) session or the like) with CN 106 via NE 102. CN 106 can use the established session (e.g., an established PDU session) to route traffic (e.g., control information, data, and the like) between UE 104 and the application server. A PDU session may be an instance of a logical connection between UE 104 and CN 106 (e.g., one or more network functions of CN 106).

[0049] In the wireless communication system 100, NE 102 and UE 104 can use the resources of the wireless communication system 100 (e.g., time resources (e.g., symbols, time slots, subframes, frames, or the like) or frequency resources (e.g., subcarriers, carriers)) to perform various operations (e.g., wireless communication). In some embodiments, NE 102 and UE 104 may support different resource structures. For example, NE 102 and UE 104 may support different frame structures. In some embodiments, such as in 4G, NE 102 and UE 104 may support a single frame structure. In some other embodiments, such as in 5G and other suitable radio access technologies, NE 102 and UE 104 may support various frame structures (i.e., multiple frame structures). NE 102 and UE 104 may support various frame structures based on one or more parameter sets.

[0050] The wireless communication system 100 may support one or more parameter sets, and the parameter sets may include subcarrier spacing and cyclic prefixes. A first parameter set (e.g., μ=0) may be associated with a first subcarrier spacing (e.g., 15 kHz) and a regular cyclic prefix. In some embodiments, the first parameter set (e.g., μ=0) associated with the first subcarrier spacing (e.g., 15 kHz) may utilize one time slot per subframe. A second parameter set (e.g., μ=1) may be associated with a second subcarrier spacing (e.g., 30 kHz) and a regular cyclic prefix. A third parameter set (e.g., μ=2) may be associated with a third subcarrier spacing (e.g., 60 kHz) and a regular cyclic prefix or an extended cyclic prefix. A fourth parameter set (e.g., μ=3) may be associated with a fourth subcarrier spacing (e.g., 120 kHz) and a regular cyclic prefix. A fifth parameter set (e.g., μ=4) may be associated with a fifth subcarrier spacing (e.g., 240 kHz) and a regular cyclic prefix.

[0051] The time intervals of resources (e.g., communication resources) can be organized according to frames (also called radio frames). Each frame may have a certain duration, such as 10 milliseconds (ms). In some embodiments, each frame may contain multiple subframes. For example, each frame may contain 10 subframes, and each subframe may have a certain duration, such as 1 ms. In some embodiments, each frame may have the same duration. In some embodiments, each subframe of a frame may have the same duration.

[0052] Alternatively, the time intervals of resources (e.g., communication resources) can be organized according to time slots. For example, a subframe may contain a certain number (e.g., quantity) of time slots. The number of time slots in each subframe may also depend on one or more parameter sets supported in the wireless communication system 100. For example, the first, second, third, fourth, and fifth parameter sets (i.e., μ=0, μ=1, μ=2, μ=3, μ=4) associated with corresponding subcarrier intervals of 15 kHz, 30 kHz, 60 kHz, 120 kHz, and 240 kHz may utilize one time slot per subframe, two time slots per subframe, four time slots per subframe, eight time slots per subframe, and 16 time slots per subframe, respectively. Each time slot may contain a certain number (e.g., quantity) of symbols (e.g., OFDM symbols). In some embodiments, the number (e.g., quantity) of time slots in a subframe may depend on the parameter set. For a conventional cyclic prefix, a time slot may contain 14 symbols. For an extended cyclic prefix (e.g., applicable to a 60 kHz subcarrier spacing), a time slot may contain 12 symbols. The relationship between the number of symbols per time slot for the regular cyclic prefix and the extended cyclic prefix, and between the number of time slots per subframe and the number of time slots per frame, may depend on the parameter set. It should be understood that references to the first parameter set (e.g., μ=0) associated with the first subcarrier spacing (e.g., 15 kHz) may be used interchangeably between subframes and time slots.

[0053] In the wireless communication system 100, the electromagnetic (EM) spectrum can be divided into various categories, bands, channels, etc., based on frequency or wavelength. For example, the wireless communication system 100 may support one or more operating frequency bands, such as frequency range identifiers FR1 (410 MHz to 7.125 GHz), FR2 (24.25 GHz to 52.6 GHz), FR3 (7.125 GHz to 24.25 GHz), FR4 (52.6 GHz to 114.25 GHz), FR4a or FR4-1 (52.6 GHz to 71 GHz), and FR5 (114.25 GHz to 300 GHz). In some embodiments, NE 102 and UE 104 may perform wireless communication on one or more of the operating frequency bands. In some embodiments, FR1 may be used by NE 102 and UE 104, as well as other equipment or devices, for cellular communication services (e.g., control information, data). In some implementations, FR2 can be used by NE 102 and UE 104, as well as other equipment or devices, for short-range, high data rate capabilities.

[0054] FR1 may be associated with one or more parameter sets (e.g., at least three parameter sets). For example, FR1 may be associated with: a first parameter set (e.g., μ=0) containing a 15 kHz subcarrier spacing; a second parameter set (e.g., μ=1) containing a 30 kHz subcarrier spacing; and a third parameter set (e.g., μ=2) containing a 60 kHz subcarrier spacing. FR2 may be associated with one or more parameter sets (e.g., at least two parameter sets). For example, FR2 may be associated with: a third parameter set (e.g., μ=2) containing a 60 kHz subcarrier spacing; and a fourth parameter set (e.g., μ=3) containing a 120 kHz subcarrier spacing.

[0055] In the wireless communication system 100, NE 102 can be configured (as described in more detail below) to control the federated learning process to perform network analysis.

[0056] Figure 2 This describes the network architecture that allows data collection related to the use of the Network Data Analysis Function (NWDAF). In some cases, the Analysis Logic Function (AnLF) trains a machine learning (ML) model from a Model Training Logic Function (MTLF) request by including the analysis ID corresponding to the analysis ID requested by the analysis consumer in the request. Information about the available analysis IDs supported by AnLF is provided in 3GPP TS23.288. The MTLF trains the ML model corresponding to the requested analysis ID by collecting data from one or more data sources (NF, OAM, or UE).

[0057] NWDAF (e.g.) Figure 1 The functionality of NWDAF can be divided between AnLF and MTLF. Figure 3 This illustrates an example of this functional partitioning.

[0058] exist Figure 3 In this framework, each MTLF is trained to perform a corresponding ML model for network analysis. Each MTLF can be associated with an analysis ID. Each such analysis ID identifies a specific network analysis. A given ML model can be configured by the ML model designer.

[0059] Each MTLF can train its model based on data collected from a data producer network function (NF). A data producer NF can be any network function that generates data. For example, a data producer NF could be an Access and Mobility Function (AMF) that generates location information about the UE. Alternatively or additionally, the MTLF can train its model based on historical data received from a Data Collection Coordination Function (DCCF) or an Analytical Data Repository Function (ADRF).

[0060] AnLF then receives the trained ML model (and / or data indicating the trained ML model) for the specific analysis ID from the MTLF corresponding to that analysis ID. AnLF may potentially use the model in conjunction with new data from (several) data producer NFs and / or DCCFs to derive the analysis requested by the consumer NF.

[0061] AnLF can subscribe to MTLF to receive trained ML models for analytics IDs using a specific API (existing technology). AnLF subscriptions are based on analytics requests from consumer NFs. NWDAF supports both analytics inference and ML model training.

[0062] In some cases, direct data exchange between the data producer (NF) and the NWDAF (i.e., MTLF or AnLF) may not be possible. This could be due to privacy concerns, for example: operators of edge networks or private slices may not want to share data with external networks. As another example, this could be due to high signaling load: signaling load can increase significantly when data is provided to a centralized training function, such as one provided by the MTLF. For networks at the edge, receiving trained models is more efficient than collecting data to derive analytics.

[0063] One way to address situations where data is not directly exchanged (e.g., due to privacy concerns or signaling load efficiency) is through federated learning. Federated learning allows local model training functions to exchange model parameters and aggregate trained models and / or intermediate training results, rather than centrally training models by collecting raw data. Therefore, federated learning is a distributed machine learning framework that allows models to be jointly trained based on data distributed across different data owners. This provides the advantage of AI / ML models being trained closer to the data source, rather than sending raw data to centralized training nodes. Instead of sending raw data, the parameters, weights, or intermediate results of the ML model can be sent back to centralized nodes to aid in the training of a general model.

[0064] It should be understood that federated learning can be conceptually divided into horizontal and vertical federated learning. These are respectively... Figure 4A and 4B The image is depicted schematically.

[0065] Horizontal federated learning, or sample-based federated learning, can be applied to scenarios where datasets share the same feature space (e.g., each dataset contains the same features) but have different samples (e.g., collected from different users). This is in... Figure 4A The explanation is as follows.

[0066] Vertical federated learning, or feature-based federated learning, can be applied to scenarios with datasets that share the same sample space (e.g., each dataset is suitable for the same group of users) but have different feature spaces (e.g., different features are collected in each dataset). This is in Figure 4B The explanation is as follows.

[0067] In both horizontal and vertical federated learning models, parameters from each local model training function are sent to a model aggregator to compute an aggregated model. The model aggregator provides updated model parameters to each MTLF, and each MTLF uses the updated model parameters to retrain its own model, thereby allowing each local model training function to have a trained model using data from multiple sources.

[0068] There are different approaches to training models using Vertical Federated Learning (VFL), including both "split" and "non-split" methods. Examples of these methods are now described in "Vertical Federated Learning: Taxonomies, Threats, and Prospects" by Qun Li et al. Specifically, its... Figure 2 Describe the classification methods of the VFL algorithm, including both splitting and non-splitting approaches. Figure 3 The non-split VFL is depicted in more detail, while Figure 4 depicts the split VFL.

[0069] In both split and non-split VFLs, one party is the "label owner" or "active participant." This party can classify the input data. The other parties are "passive participants" or "workers" who participate in the VFL training process.

[0070] In a non-split VFL, passive participants send intermediate results based on locally collected data, obtained using their own models, to active participants, who then use labels and their own ML models as a basis to compute gradients / losses. Some scenarios also include a coordinator that ensures message exchange between VFL parties is encrypted (e.g., Li). Figure 3 (Depicted in Chinese).

[0071] In contrast, in a split VFL, the model is split among several parties. One party owns the top model (label owner / VLF server), while the other parties own one or more bottom models (passive participants). This is illustrated in Figure 4. The label owner can also be an active participant. Compared to a non-split VFL, the VFL server knows the labels and is able to compute gradients / losses shared with the passive participants.

[0072] Figure 5A and 5B Describe the communication flow for federated learning between different NWDAFs. These communication flows are described in 3GPP TS 23.288.

[0073] Figure 5A This describes the registration, discovery, and selection of clients for federated learning. Each MTLF registers its FL capabilities (FL client or FL server) within its NWDAF profile in the NRF. The profile also contains information about the analytics IDs that support FL and / or the location regions and / or (some) event IDs that support data collection. The NWDAF MTLF with FL server capabilities discovers candidate FL clients for training models using federated learning by receiving a list of FL clients from the NRF containing the analytics IDs and / or location regions and / or event IDs required to support training the model. The FL server prepares a request and sends it to all candidates to check their eligibility to join the FL process, i.e., in terms of data availability and required time scheduling (steps 7 through 9), and each FL client responds to whether it is eligible to join the process.

[0074] 5B describes the FL process once the FL client has been selected and joined the FL process.

[0075] The general process of FL is that the FL server sends subscription requests containing local model parameters related to federated learning for model training to each selected FL client, and each FL client uses data from its local NF to train the model.

[0076] This disclosure provides examples of solutions supporting vertical federated learning, enabling 5GS to assist collaborative AI / ML operations involving 5GC / NWDAF or AF. Specifically, to support VFL, this disclosure provides methods for VFL entities (servers, active participants, or passive participants) to participate in the VFL training process. The following description also illustrates actions that servers and active participants can take after the VFL training process is completed (e.g., model information storage).

[0077] Figure 6 Various architectures for executing VFLs according to examples of this disclosure are described. These architectures depict communication between VFL server 610, active participant 615, and passive participant 625. In this example, VFL server 710, which may also be referred to as VFL coordinator or controller, is responsible for managing the VFL process, selecting VFL participants 615 and 625 for model training, and assigning functions to act as VFL active participant 615 or passive participant 625.

[0078] In this example, VFL active participant 615 is a VFL function that possesses a portion of the ML model for a given analytics ID and knows the labels of the ML model. Active participant 615 is the main function used to train the ML model for the given analytics ID.

[0079] In this example, VFL passive participant 625 is a VFL function that has part of an ML model for a given analysis ID, is unaware of the ML model's labels, but is able to collect local data on one or more features.

[0080] A given VFL function in a 3GPP network can support a combination of the above functions, for example, supporting both VFL server and active participant functions.

[0081] Some examples of this disclosure relate to the following scenarios: where, once a model is trained using VFL, the VFL server 610 or the active participant 615 is aware of all VFL participants (active 615 and passive 625) in the model training process and the contribution weight of each participant.

[0082] Aspects of such instances involve actions performed in response to a consumer's (e.g., AnLF 605) request for inference results. Reference will now be made to... Figure 7A and 7B Describe examples of this type of aspect.

[0083] exist Figure 7AIn the first instance described, consumer 605 sends a request for the inference result to VFL server / active participant 610, 615 based on its knowledge of the participants in the VFL process when the model was trained. Then, VFL server / active participant 610, 615 sends a request for the inference value from each participant in the VFL training process based on its knowledge of the participants in the VFL process when the model was trained.

[0084] exist Figure 7B In the second instance described, consumer 605 directly sends requests for inference values ​​from each participant in the VFL training process, based on its knowledge of using VFL to train the ML model and its knowledge of the participants in the VFL process when the model was trained.

[0085] Depending on which of these instances is followed, the consumer 605 or the VFL server / active participant 610, 615 derives the aggregated inference value based on the cognitively derived contribution weights of each participant from the VFL model training process.

[0086] In these instances, a given VFL function (i.e., VFL server 610, active participant 615, and / or passive participant 625) may be a part of the NWDAF AnLF, the NWDAF MTLF, or a function within the AF. When an interaction occurs between a VFL function from the NWDAF and a VFL function from the AF, the exchange of messages may be performed via the Network Exposure Function (NEF) 628.

[0087] Figure 6 Four architectural scenarios are described, including VFL server 610, active participant 615, and passive participant 625, involving NWDAF AnLF, MTLF, and AF. Specifically, Figure 6 The top line describes instances where a VFL is executed within the core network, and the bottom line describes instances where a VFL is initiated outside the core network. Within those lines, Figure 6 The left side depicts an instance that includes a VFL server 610, and the right side depicts an instance that does not use a VFL server 610.

[0088] Figure 7A and 7B A process flow is described according to aspects of this disclosure. The process flow can be implemented as referenced. Figure 1 The described aspects of the wireless communication system 100, and more specifically, are available in... Figure 6The process can be implemented within one or more of the architectures described herein. The process flow may involve one or more of the following: NWDAF AnLF 705, VFL server 710, VFL active participant 715, VFL passive participant 725, VFL passive participant 735, NRF 745, and ADRF 750. Therefore, the process flow can be implemented within one or more of these components. Figure 6 The architecture described herein is implemented where entities with the same name have the same functionality as described above. In the following description, operations performed by one or more of entities 705 to 750 may be performed in a different order than shown or at different times. Some operations may also be omitted from the process flow, and other operations may be added.

[0089] exist Figure 7A and 7B In the example, the contributions of active and passive participant functions 715, 725, and 735 to the VLF process can be determined, and the contribution weight of each participant is known. This information can be provided to the model consumer, allowing the consumer to use the model (trained using VLF) for inference. Therefore, Figure 7A and 7B The process flow provides an efficient way to perform inference on VFL-trained models in wireless communication networks. As explained above, this improves the accuracy and performance of network analysis, thereby improving system performance.

[0090] Now refer to Figure 7A Describe the first reasoning method.

[0091] At point 1, the NWDAF AnLF 705 supporting AnLF receives requests for analytics (e.g., observed service experience analytics) and includes the analytics ID and analytics filter as specified in 3GPP TS 23.288.

[0092] At two points, the NWDAF AnLF 705 identifies the ML model required to derive the analysis and determines that distributed inference is needed (or alternatively, determines that federated learning is used to train the ML model). The NWDAF AnLF 705 obtains this information by interfacing with the NRF 745 or ADRF 750 or via prior interaction with the MTLF, which acts as the VFL server 710 or VFL active participant 715, providing the NWDAF AnLF 705 with information about the ML model details. The ML model details include the address, contribution weight, and VFL coordination ID of the VFL server 710 or VFL active participant 715 or passive participants 725, 735 (which is used by each VFL participant to identify the specific VFL training collaboration process used to train the ML model for the analysis ID).

[0093] At three points, the NWDAF AnLF 705 sends inference requests to the VFL server 710 (if available) or the VFL active participant 715. The NWDAF AnLF 705 includes an analytics ID (e.g., observed service experience), analytics filters (e.g., target UE, service area, slice information, application information, etc.), and a VFL collaboration ID (if known). Inference requests can be subscription requests (i.e., periodically providing feedback) or one-time requests.

[0094] At point 4, it is assumed that the ML model has been continuously trained and updated using VFL.

[0095] At five locations, VFL server 710 or VFL active participant 715 identifies the ML model linked to the analytics ID, and identifies the VFL participants (active 715 and passive 725, 735) and associated VFL collaboration IDs involved in the vertical federated process used to train the ML model. VFL server 710 or VFL active participant 715 obtains this information locally or via NRF 745 or by obtaining model information from ADRF 750.

[0096] At point 6, VFL server 710 or VFL active participant 715 sends an inference request to each participant in the VFL process. The request may include an analytics ID, a filter for the sample used for inference computation, and a VFL collaboration ID. The inference request may be a subscription request (i.e., providing feedback periodically) or a one-time request.

[0097] At 7 locations, each VFL participant 715, 725, 735 uses VFL to identify the local ML model linked to the ML training process and uses its local ground truth data to compute the inference output (7a, 7b, 7c).

[0098] In some instances, if the VFL server 710 is not deployed, the VFL active participant 715 does not send inference requests, but instead computes the inference output locally.

[0099] At point 8, the inference output (or intermediate results) from each participant 715, 725, and 735 is sent to the VFL server 710 and / or the VFL active participant 715.

[0100] At point 9, VFL server 710 and / or active participant 715 calculate aggregate inference outputs while taking into account the contribution weights from each participant 715, 725, 735.

[0101] At point 10, VFL server 710 and / or active participant 715 can also use the inference output as input to further train the ML model, and provide gradients / losses to each VFL participant 715, 725, 735, each VFL participant 715, 725, 735 can use the gradients / losses to further train its local ML model.

[0102] At point 11, VFL server 710 and / or active participant 715 prepare an inference response, including output data, to be transmitted to NWDAF AnLF 705.

[0103] At point 12, the NWDAF AnLF 705 prepares to analyze the output data, taking into account the inference information from the VFL server / active participants 710 and 715.

[0104] At point 13, the results are sent to the consumer being analyzed.

[0105] therefore, Figure 7A The method provides an efficient way to provide analysis using inference from a model trained with VFL.

[0106] Figure 7B The NWDAF AnLF 705 is depicted as an alternative instance to the VFL server 710. In this case, the NWDAF AnLF 705 can send inference requests directly to each VFL participant 715, 725, and 735. This will be described in the following section.

[0107] At point 1, the AnLF-enabled NWDAF 705 receives requests for analytics (e.g., observed service experience analytics) and includes the analytics ID and analytics filter as specified in 3GPP TS 23.288.

[0108] At two points, the NWDAF AnLF 705 identifies the ML model linked to the analytics ID to derive analytics and determines that distributed inference is required (or alternatively, determines that federated learning should be used to train the ML model). The NWDAF AnLF 705 obtains this information by interfacing with the NRF 745 or ADRF 750 or via prior interaction with the MTLF, which acts as the VFL server 710 or VFL active participant 715, providing information about the ML model details to the NWDAF AnLF 705. The ML model details include the address of the VFL server / active participant 710, 715 or passive participant 725, 735, contribution weights, and the VFL coordination ID (which is used by each VFL participant to identify the specific VFL training collaboration process used to train the ML model for the analytics ID).

[0109] At point 3, it is assumed that the ML model has been continuously trained and updated using VFL.

[0110] At four points, the NWDAF AnLF 705 sends inference requests to each participant in the VFL process, 715, 725, and 735. Requests may include an analytics ID, a filter for the samples used for inference computation, and a VFL collaboration ID. Inference requests can be subscription requests (i.e., providing feedback periodically) or one-time requests.

[0111] At 5 locations, each VFL participant 715, 725, 735 uses VFL to identify the local ML model linked to the ML training process and uses its local ground truth data to compute the inference output (5a, 5b, 5c).

[0112] In this example, if VFL server 710 is not deployed, then VFL active participant 715 does not send inference requests, but instead computes the inference output locally.

[0113] At point 6, the inference output (or intermediate result) from each participant is sent to the VFL server 710 and / or the VFL active participant 715.

[0114] At point 7, the NWDAF AnLF 705, acting as VFL server 710, calculates the aggregate inference output while taking into account the contribution weights from each participant, and prepares the analysis output data while taking into account the inference information from VFL servers / active participants 710, 715.

[0115] At point 8, the results are sent to the consumer being analyzed.

[0116] therefore, Figure 7B The method provides an efficient way to provide analysis using inference from a model trained with VFL.

[0117] Figure 8 This describes an example of an NE 800 according to aspects of this disclosure. The NE 800 may include a processor 802, a memory 804, a controller 806, and a transceiver 808. The processor 802, memory 804, controller 806, or transceiver 808, or various combinations thereof, or various components thereof, may be examples of components for performing the various aspects of this disclosure described herein. These components may be coupled via one or more interfaces (e.g., operatively ground, communicatively ground, functional ground, electronic ground, electrical ground).

[0118] Processor 802, memory 804, controller 806, or transceiver 808, or various combinations or components thereof, may be implemented in hardware (e.g., a circuit system). The hardware may include processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), or other programmable logic devices, or any combination thereof, configured to or otherwise support components for performing the functions described in this disclosure.

[0119] Processor 802 may include intelligent hardware devices (e.g., a general-purpose processor, DSP, CPU, ASIC, FPGA, or any combination thereof). In some embodiments, processor 802 may be configured to operate memory 804. In some other embodiments, memory 804 may be integrated into processor 802. Processor 802 may be configured to execute computer-readable instructions stored in memory 804 to cause NE 800 to perform various functions of this disclosure.

[0120] Memory 804 may comprise volatile or non-volatile memory. Memory 804 may store computer-readable, computer-executable code containing instructions that, when executed by processor 802, cause NE 800 to perform the various functions described herein. The code may be stored in a non-transitory computer-readable medium, such as memory 804 or another type of memory. Computer-readable medium includes both non-transitory computer storage media and communication media, wherein the communication media includes any medium that facilitates the transfer of a computer program from one place to another. Non-transitory storage media may be any available medium accessible by a general-purpose or special-purpose computer.

[0121] In some implementations, processor 802 and memory 804 coupled to processor 802 may be configured to cause NE 800 to perform one or more of the functions described herein (e.g., instructions stored in memory 804 are executed by processor 802). For example, processor 802 may support wireless communication at NE 800 according to examples disclosed herein. NE 800 may be configured to support components for: receiving a request message for providing inference output for a network analysis operation identified by an analysis identifier; outputting the inference request message to each of a set of one or more model training functions, the set of model training functions being participants in a federated learning training process for a trained model corresponding to the network analysis operation; receiving inference output from each function of the set; and outputting a response message including an aggregated inference output based on a corresponding contribution weight of each of the functions in the federated learning training process.

[0122] Controller 806 manages the input and output signals of NE 800. Controller 806 can also manage peripheral devices not integrated into NE 800. In some embodiments, controller 806 may utilize an operating system such as iOS®, ANDROID®, WINDOWS®, or other operating systems. In some embodiments, controller 806 may be implemented as part of processor 802.

[0123] In some embodiments, NE 800 may include at least one transceiver 808. In other embodiments, NE 800 may have more than one transceiver 808. Transceiver 808 may represent a wireless transceiver. Transceiver 808 may include one or more receiver chains 810, one or more transmitter chains 812, or a combination thereof.

[0124] Receiver chain 810 may be configured to receive signals (e.g., control information, data, packets) via wireless media. For example, receiver chain 810 may include one or more antennas for receiving signals in the air or via wireless media. Receiver chain 810 may include at least one amplifier (e.g., a low-noise amplifier (LNA)) configured to amplify the received signal. Receiver chain 810 may include at least one demodulator configured to demodulate the received signal and obtain transmitted data by reversing the modulation technique applied during signal transmission. Receiver chain 810 may include at least one decoder for decoding the processed demodulated signal to receive the transmitted data.

[0125] Transmitter chain 812 can be configured to generate and transmit signals (e.g., control information, data, packets). Transmitter chain 812 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). Transmitter chain 812 may also include at least one power amplifier configured to amplify the modulated signal to an appropriate power level suitable for transmission over a wireless medium. Transmitter chain 812 may also include one or more antennas for transmitting the amplified signal into the air or wireless medium.

[0126] Figure 9 A flowchart illustrating a method according to an aspect of this disclosure is provided. The operation of the method may be implemented by a NE as described herein. In some embodiments, the NE may execute a set of instructions to control the functional elements of the NE to perform the described functions. The operation of the method may be performed according to examples as described herein. In some embodiments, aspects of operation may be referenced... Figure 8The described NE is executed. One or more of the operations may be omitted, or executed in a different order than the example shown.

[0127] At 902, the method may include a request message to provide inference output for a network analysis operation identified by an analysis identifier;

[0128] At 904, the method may include outputting an inference request message to each of a set of one or more model training functions, the set of model training functions being participants in a federated learning training process of a trained model corresponding to a network analysis operation.

[0129] At 906, the method may include obtaining inference output from each of the functions in the set;

[0130] At 908, the method may include outputting a response message that includes aggregated inference output, the aggregated inference output being based on the corresponding contribution weight of each of the functions during the federated learning training process.

[0131] It should be noted that the method described herein describes one possible implementation, and the operation and steps may be rearranged or otherwise modified, and other implementations are possible.

[0132] Figure 10 A wireless communication system 1005 according to aspects of this disclosure is schematically depicted. The wireless communication system 1005 includes a first network entity 1010 and a second network entity 1015.

[0133] The first network entity 1010 is configured to receive a first request message that provides inference output for a network analysis operation identified by an analysis identifier; and outputs a second request message to the second network entity to perform an inference operation regarding the network analysis operation.

[0134] The second network entity 1015 is configured to: output a third request message to each of one or more functions participating in the federated learning training process of the model used to perform network analysis operations; obtain inference output from each of the one or more functions; and output aggregated inference output to the first network entity based on the obtained inference output and the corresponding contribution weight of each of the functions in the federated learning training process.

[0135] The first network entity 1010 is configured to output a response message that includes the aggregated inference output.

[0136] The description herein is provided to enable those skilled in the art to make or use this disclosure. Various modifications to this disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the scope of this disclosure. Therefore, this disclosure is not limited to the examples and designs described herein, but should be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A network entity for supporting wireless communication analysis, the network entity comprising instructions executable by at least one processor to cause the network entity to: Obtain a request message to provide inference output for a network analysis operation identified by an analysis identifier; The inference request message is output to each of a set of one or more model training functions, which are participants in the federated learning training process of the trained model corresponding to the network analysis operation. Obtain the inference output from each function of the group; and The output includes a response message of aggregated inference output, which is based on the corresponding contribution weight of each of the functions during the federated learning training process.

2. The network entity according to claim 1, wherein the instructions cause the network entity to: The model information message is obtained from the second network entity, and the model information message includes: The identifier of the model; Identifiers of one or more functions that participate in the federated learning training process; and The coordination identifier associated with the current example of the federated learning training process.

3. The network entity of claim 2, wherein the model information message includes information identifying each of the model training functions as having one or more roles in the federated training process, the one or more roles being one or more of a group including federated learning servers, active participants in federated learning, and passive participants in federated learning.

4. The network entity according to claim 3, wherein at least one of the model training functions includes at least one of the network data analysis function NWDAF and the application function AF.

5. The network entity according to any one of claims 2 to 4, wherein the instructions cause the network entity to obtain the model information message based on the analysis identifier.

6. The network entity according to claim 5, wherein obtaining the model information message includes outputting a model information request message including the analysis identifier to the second network entity.

7. The network entity according to any one of claims 2 to 6, wherein the second network entity includes at least one of a network repository function and an analytics data repository function.

8. The network entity according to any preceding claim, wherein the network entity includes at least one of the following: Network Data Analysis Function (NWDAF), wherein the NWDAF supports analysis logic functions; and NWDAF, which is configured to control the federated learning process for the analysis identifier.

9. The network entity according to any of the preceding claims, wherein the instructions cause the network entity to receive the request message from a consumer entity within a communication network.

10. The network entity according to claim 9, wherein the consumer entity is a consumer analysis logic function.

11. The network entity according to claim 9 or claim 10, wherein the instruction causes the network entity to output the response message to the consumer entity.

12. The network entity according to any of the preceding claims, wherein the inference output is the aggregated inference output, and wherein the instruction causes the network entity to obtain the aggregated inference output from a third network entity.

13. The network entity of claim 12, wherein the third network entity includes a Model Training Logic Function (MTLF), the MTLF being a controller for the federated learning process of the model.

14. The network entity of claim 12, wherein the third network entity includes application function AF.

15. The network entity according to any one of claims 1 to 11, wherein the inference output includes a corresponding interference output from each of the one or more functions, and wherein the instruction causes the network entity to: Each of the interference outputs is obtained from the corresponding function of the one or more functions; and The aggregate inference output is determined based on each of the inference outputs.

16. The network entity of claim 15, wherein the instructions cause the network entity to: Obtain the corresponding contribution weight; and Furthermore, the aggregated inference output is determined based on the corresponding contribution weights.

17. A method performed by a network entity of a wireless communication network, the method comprising: Obtain a request message to provide inference output for a network analysis operation identified by an analysis identifier; The inference request message is output to each of a set of one or more model training functions, which are participants in the federated learning training process of the trained model corresponding to the network analysis operation. Obtain the inference output from each function of the group; The output includes a response message of aggregated inference output, which is based on the corresponding contribution weight of each of the functions during the federated learning training process.

18. A communication network system comprising: The first network entity, which is used for wireless communication in a communication network; and A second network entity, used for wireless communication in the communication network, wherein: The first network entity is configured as follows: Receive the first request message to provide inference output for the network analysis operation identified by the analysis identifier; The second request message is output to the second network entity to perform inference operations regarding the network analysis operation. The second network entity is configured as follows: The third request message is output to each of one or more functions participating in the federated learning training process of the model used to perform the network analysis operation; Obtain inference output from each of the one or more functions; Based on the obtained inference output and the corresponding contribution weight of each function during the federated learning training process, the aggregated inference output is output to the first network entity, and The first network entity is configured to output a response message that includes the aggregated inference output.

19. The system of claim 18, wherein the second network entity is configured to provide training data to the one or more functions based on the inference output for training the model.

20. The system according to claim 18 or claim 19, wherein at least one of the following: The first network entity is an NWDAF that supports AnLF; and The second network entity is the federated learning server.

Images