System and method for dynamic machine learning model distribution

Abstract

The disclosure relates to a 5G or 6G communication system for supporting a higher data transmission rate. The present invention relates to a system (i.e., entity 1 or first entity) and method for providing client selection and model distribution assistance in Federated Learning / distributed Machine Learning (FL / ML) operations. The method comprising receiving, from a vertical application layer (VAL) server, an ML model training request to assist in ML model training; identifying whether the VAL server is authorized to transmit the ML model training request; and transmitting, to the VAL server, an ML model training response in response to the ML model training request.

Description

SYSTEM AND METHOD FOR DYNAMIC MACHINE LEARNING MODEL DISTRIBUTION

[0001] The present disclosure relates generally to a field of application layer support for AI / ML services, and more particularly to a system and a method for machine learning model distribution.

[0002] In Rel-19, 3GPP has been studying on application layer support for Artificial Intelligence (AI) / Machine Learning (ML) services. One of key aspects of the study is about assisting the application functions (like, application servers, VAL servers, application enabler servers etc..) for federated and distributed learning. In federated and distributed machine learning techniques, the model may be trained simultaneously at various entities (like UEs, servers etc.). These various entities (like UEs, servers etc..) may be called machine learning clients (ML clients) or federated learning clients (FL clients), which can be hosted on UEs or application servers (like VAL server, application enabler servers etc..). The machine learning server that is responsible for training the model, may select the ML / FL clients and pushes the machine learning model information to the selected ML / FL clients, so that the ML / FL clients may train the model (fully or partially) locally and share the training results back to with the machine learning server.

[0003] The application layer entities supporting the AI / ML services for application functions (like, application servers, VAL servers, etc..) may include application enabler server (like AI / ML application enabler server, ADAE server etc..) providing the server-side functionalities and application enabler client (like AI / ML application enabler client, ADAE client, FL client, ML client, etc..) providing client-side functionalities. The AI / ML Services need to have information about the conditions upon which AI / ML entities (e.g., VAL server, AI / ML Enabler Client, AI / ML Enabler Server) are available or not available to participate in the AI / ML operations (e.g., ML model training).

[0004] Further, considering that the FL / ML clients are deployed on end devices like UEs, there is potentially a very large number of FL / ML clients that may participate in machine learning operations. Out of these potentially large numbers for FL / ML clients, a set of FL / ML clients can be selected for performing various machine learning tasks. Due to various factors like mobility, changing capabilities, varying resource conditions, etc. the set of UEs (hosting FL / ML clients) may participate in federated / distributed learning changes over time and may change more frequently. Also, the model information that is exchanged between FL / ML clients and the machine learning servers may also be huge, considering the large machine learning models. So, there has to be frequent need for changing the member FL / ML clients that may participate in federated / distributed machine learning operations and also frequently the need to distribute the model information.

[0005] In such scenario, it might be highly resource (compute and network) intensive and may lead to performance overhead on machine learning servers (e.g., VAL servers, Application servers etc.) to manage the AIML operations for federated and distributed machine learning mechanisms.

[0006] Currently, 3GPP TR 23700-82 captures solutions to assist the machine learning servers in member (FL / ML client) selection for federated and distributed learning operations. However, it is not clear on how the assistance can be provided for member selection and model distribution, to reduce the load on machine learning servers, taking the dynamic and changing eligible FL / ML client's scenario.

[0007] The information disclosed in this background of the disclosure section is only for enhancement of understanding of the general background of the invention and should not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person skilled in the art.

[0008] This summary is provided to introduce a selection of concepts, in a simplified format, which is further described in the detailed description of the invention. This summary is neither intended to identify key or essential inventive concepts of the invention nor is it intended for determining the scope of the invention.

[0009] In an embodiment, the present disclosure relates to a method for machine learning model distribution. The method comprising selecting dynamically, by a first entity, at least one client from a plurality of clients based on client selection information related to the plurality of clients received from a second entity associated with the first entity, for one of federated or distributed machine learning. The method comprising sending, by the first entity, a request to train a Machine Learning (ML) model to the at least one client selected from the plurality of clients, wherein the request comprises information pertaining to a ML model to be trained by the at least one client.

[0010] In another embodiment, the present disclosure relates to a first entity for dynamic machine learning model distribution. The first entity comprising a processor; and a memory communicatively coupled to the processor, wherein the memory stores processor-executable instructions, which on execution, cause the processor to select dynamically at least one client from a plurality of clients based on client selection information related to the plurality of clients received from a second entity associated with the first entity, for one of federated or distributed machine learning. The processor is configured to send a request to train a Machine Learning (ML) model to the at least one client selected from the plurality of clients, wherein the request comprises information pertaining to a ML model to be trained by the at least one client.

[0011] According to an aspect of the present disclosure, there is provided a method by artificial intelligence distributed machine learning enabler (AIMLE) server for machine learning (ML) model distribution in a wireless communication system, the method comprising: receiving, from a vertical application layer (VAL) server, an ML model training request to assist in ML model training; identifying whether the VAL server is authorized to transmit the ML model training request; and transmitting, to the VAL server, an ML model training response in response to the ML model training request.

[0012] According to another aspect of the present disclosure, there is provided an artificial intelligence distributed machine learning enabler (AIMLE) server entity in a communication system, the AIMLE server entity comprising: a transceiver; and a processor coupled to the transceiver, and configured to: receive, from a vertical application layer (VAL) server entity, an ML model training request to assist in ML model training, identify whether the VAL server entity is authorized to transmit the ML model training request, and transmit, to the VAL server entity, an ML model training response in response to the ML model training request.

[0013] To further clarify the advantages and features of the present invention, a more particular description of the invention will be rendered by reference to specific embodiments thereof, which are illustrated in the appended drawings. It is appreciated that these drawings depict only typical embodiments of the invention and are therefore not to be considered limiting its scope. The invention will be described and explained with additional specificity and detail with the accompanying drawings.

[0014] Before undertaking the DETAILED DESCRIPTION below, it may be advantageous to set forth definitions of certain words and phrases used throughout this patent document: the terms "include" and "comprise," as well as derivatives thereof, mean inclusion without limitation; the term "or," is inclusive, meaning and / or; the phrases "associated with" and "associated therewith," as well as derivatives thereof, may mean to include, be included within, interconnect with, contain, be contained within, connect to or with, couple to or with, be communicable with, cooperate with, interleave, juxtapose, be proximate to, be bound to or with, have, have a property of, or the like; and the term "controller" means any device, system or part thereof that controls at least one operation, such a device may be implemented in hardware, firmware or software, or some combination of at least two of the same. It should be noted that the functionality associated with any particular controller may be centralized or distributed, whether locally or remotely. The phrase "at least one of," when used with a list of items, means that different combinations of one or more of the listed items may be used, and only one item in the list may be needed. For example, "at least one of: A, B, and C" includes any of the following combinations: A, B, C, A and B, A and C, B and C, and A and B and C. For example, "at least one of A, B, or C " includes any of the following combinations: A, B, C, A and B, A and C, B and C, and A and B and C.

[0015] Moreover, various functions described below can be implemented or supported by one or more computer programs, each of which is formed from computer readable program code and embodied in a computer readable medium. The terms "application" and "program" refer to one or more computer programs, software components, sets of instructions, procedures, functions, objects, classes, instances, related data, or a portion thereof adapted for implementation in a suitable computer readable program code. The phrase "computer readable program code" includes any type of computer code, including source code, object code, and executable code. The phrase "computer readable medium" includes any type of medium capable of being accessed by a computer, such as read only memory (ROM), random access memory (RAM), a hard disk drive, a compact disc (CD), a digital video disc (DVD), or any other type of memory. A "non-transitory" computer readable medium excludes wired, wireless, optical, or other communication links that transport transitory electrical or other signals. A non-transitory computer readable medium includes media where data can be permanently stored and media where data can be stored and later overwritten, such as a rewritable optical disc or an erasable memory device.

[0016] The terms used herein to describe the embodiments of the present application is not intended to limit and / or define the scope of the present application. For example, unless otherwise defined, the technical or scientific terms used in the present disclosure should have ordinary meanings as understood by ordinary skilled in the art to which the present application belongs.

[0017] It should be understood that "first", "second" and similar words used in the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. Unless clearly indicated otherwise in the context, similar words such as "a", "an", "the" and the like in the singular form do not indicate a quantitative limitation, but indicate the existence of at least one.

[0018] As used herein, any reference to "one example" or "an example", "one embodiment" or "an embodiment" means that a particular element, feature, structure or characteristic described in conjunction with the embodiment is included in at least one embodiment. The appearances of the phrases "in one embodiment" or "in one example" in different places in the specification are not necessarily all referring to the same embodiment.

[0019] As used herein, "a part of" a certain thing means "at least some of" this thing, so it may mean being less than the entirety thereof or being the entirety thereof. Therefore, "a part of" the thing includes the whole thing as a special case, that is, an example in which the whole thing is a part of the thing.

[0020] It will be further understood that words such as "include", "contain" or the like means that the elements or objects appearing preceding the word encompass the elements or objects listed behind the word as well as their equivalents, without excluding other elements or objects. Words such as "connect", "interconnect" or the like are not limited to physical or mechanical connections, but may include electrical connection, whether direct or indirect. "Up", "Down", "Left" and "Right" are only used to indicate relative positional relationships. When the absolute position of the described object changes, accordingly, the relative positional relationship may change as well.

[0021] The various embodiments discussed below for describing the principle of the present disclosure in this patent document are for illustration only, and should not be construed as limiting the scope of the present disclosure in any way. Those skilled in the art will understand that the principle of the present disclosure may be implemented in any suitably arranged wireless communication system. For example, although the following detailed description of the embodiments of the present disclosure will focus on LTE and 5G communication systems, those skilled in the art can understand that the main points of the present disclosure can also be applied to other communication systems with similar technical backgrounds and channel formats, with slight modifications and basically without departing from the scope of the present disclosure. The schemes of the embodiments of the present application may be applied to various communication systems. For example, the communication systems may include a Global System for Mobile communications (GSM) system, a Code Division Multiple Access (CDMA) system, a Wideband Code Division Multiple Access (WCDMA) system, General Packet Radio Service (GPRS), Long Term Evolution (LTE) system, LTE Frequency Division DUplex (FDD) system, LTE Time Division DUplex (TDD), Universal Mobile Telecommunication System (UMTS), worldwide interoperability for microwave access (WiMAX) communication system, fifth generation (5th generation, 5G) system or New Radio (NR), etc. In addition, the schemes of the embodiments of the present application may be applied to future-oriented communication technologies. In addition, the schemes of the embodiments of the present application may be applied to future-oriented communication technologies.

[0022] The following description with reference to the accompanying drawings is provided to assist in a comprehensive understanding of various embodiments of the present disclosure as defined by the claims and their equivalents. The description includes various specific details to assist in that understanding but should be regarded as exemplary only. Accordingly, the ordinary skilled in the art will recognize that various changes and modifications to the various embodiments described herein can be made without departing from the scope and spirit of the present disclosure. In addition, descriptions of well-known functions and structures may be omitted for clarity and conciseness.

[0023] The terms and wordings used in the following description and claims are not limited to the bibliographical meanings, but are merely used by the inventor to enable a clear and consistent understanding of the present disclosure. Accordingly, it should be apparent to those skilled in the art that the following description of various embodiments of the present disclosure is provided for illustration purpose only, but not for the purpose of limiting the present disclosure as defined by the appended claims and their equivalents.

[0024] It should be understood that the singular forms "a," "an," and "the" include plural referents, unless clearly indicated otherwise in the context. Thus, for example, reference to "a component surface" includes reference to one or more of such surfaces.

[0025] The term "include" or "may include" refers to the existence of a corresponding disclosed function, operation or component which can be used in various embodiments of the present disclosure, and does not limit the existence of one or more additional functions, operations, or components. The terms "include" and / or "have" may be construed to represent certain characteristics, numbers, steps, operations, constituent elements, components or combinations thereof, but may not be construed to exclude the possibility of existence of one or more other characteristics, numbers, steps, operations, constituent elements, components or combinations thereof.

[0026] The term "or" used in various embodiments of the present disclosure includes any of the listed terms or all combinations thereof. For example, "A or B" may include A, may include B, or may include both A and B.

[0027] Unless defined differently, all terms used in the present disclosure, including technical or scientific terms, have the same meanings as those understood by the skilled in the art as described in the present disclosure. Common terms as defined in a dictionary are to be interpreted to have meanings consistent with the context in the relevant technical field o, and are not to be interpreted ideally or excessively, unless clearly defined as such in the present disclosure.

[0028] Definitions for certain words and phrases are provided throughout this patent document, those of ordinary skill in the art should understand that in many, if not most instances, such definitions apply to prior, as well as future uses of such defined words and phrases.

[0029] The embodiments of the disclosure itself, as well as a preferred mode of use, further objectives and advantages thereof, will best be understood by reference to the following detailed description of an illustrative embodiment when read in conjunction with the accompanying drawings. One or more embodiments are now described, by way of example only, with reference to the accompanying drawing in which:

[0030] Fig. 1illustrates a procedure for dynamically selecting FL / ML member clients and distributing machine learning model information to initiate machine learning model training the selected FL / ML member clients, according to the embodiments of the present disclosure.

[0031] Fig. 2shows a detailed block diagram of an entity 1 for dynamic machine learning model distribution, according to the embodiments of the present disclosure.

[0032] Fig. 3illustrates a flowchart showing a method for dynamic machine learning model distribution, according to the embodiments of the present disclosure.

[0033] The figure depicts embodiments of the disclosure for purposes of illustration only. One skilled in the art will readily recognize from the following description that alternative embodiments of the structures and methods illustrated herein may be employed without departing from the principles of the disclosure described herein.

[0034] 5G mobile communication technologies define broad frequency bands such that high transmission rates and new services are possible, and can be implemented not only in "Sub 6GHz" bands such as 3.5GHz, but also in "Above 6GHz" bands referred to as mmWave including 28GHz and 39GHz. In addition, it has been considered to implement 6G mobile communication technologies (referred to as Beyond 5G systems) in terahertz bands (for example, 95GHz to 3THz bands) in order to accomplish transmission rates fifty times faster than 5G mobile communication technologies and ultra-low latencies one-tenth of 5G mobile communication technologies.

[0035] At the beginning of the development of 5G mobile communication technologies, in order to support services and to satisfy performance requirements in connection with enhanced Mobile BroadBand (eMBB), Ultra Reliable Low Latency Communications (URLLC), and massive Machine-Type Communications (mMTC), there has been ongoing standardization regarding beamforming and massive MIMO for mitigating radio-wave path loss and increasing radio-wave transmission distances in mmWave, supporting numerologies (for example, operating multiple subcarrier spacings) for efficiently utilizing mmWave resources and dynamic operation of slot formats, initial access technologies for supporting multi-beam transmission and broadbands, definition and operation of BWP (BandWidth Part), new channel coding methods such as a LDPC (Low Density Parity Check) code for large amount of data transmission and a polar code for highly reliable transmission of control information, L2 pre-processing, and network slicing for providing a dedicated network specialized to a specific service.

[0036] Currently, there are ongoing discussions regarding improvement and performance enhancement of initial 5G mobile communication technologies in view of services to be supported by 5G mobile communication technologies, and there has been physical layer standardization regarding technologies such as V2X (Vehicle-to-everything) for aiding driving determination by autonomous vehicles based on information regarding positions and states of vehicles transmitted by the vehicles and for enhancing user convenience, NR-U (New Radio Unlicensed) aimed at system operations conforming to various regulation-related requirements in unlicensed bands, NR UE Power Saving, Non-Terrestrial Network (NTN) which is UE-satellite direct communication for providing coverage in an area in which communication with terrestrial networks is unavailable, and positioning.

[0037] Moreover, there has been ongoing standardization in air interface architecture / protocol regarding technologies such as Industrial Internet of Things (IIoT) for supporting new services through interworking and convergence with other industries, IAB (Integrated Access and Backhaul) for providing a node for network service area expansion by supporting a wireless backhaul link and an access link in an integrated manner, mobility enhancement including conditional handover and DAPS (Dual Active Protocol Stack) handover, and two-step random access for simplifying random access procedures (2-step RACH for NR). There also has been ongoing standardization in system architecture / service regarding a 5G baseline architecture (for example, service based architecture or service based interface) for combining Network Functions Virtualization (NFV) and Software-Defined Networking (SDN) technologies, and Mobile Edge Computing (MEC) for receiving services based on UE positions.

[0038] As 5G mobile communication systems are commercialized, connected devices that have been exponentially increasing will be connected to communication networks, and it is accordingly expected that enhanced functions and performances of 5G mobile communication systems and integrated operations of connected devices will be necessary. To this end, new research is scheduled in connection with eXtended Reality (XR) for efficiently supporting AR (Augmented Reality), VR (Virtual Reality), MR (Mixed Reality) and the like, 5G performance improvement and complexity reduction by utilizing Artificial Intelligence (AI) and Machine Learning (ML), AI service support, metaverse service support, and drone communication.

[0039] Furthermore, such development of 5G mobile communication systems will serve as a basis for developing not only new waveforms for providing coverage in terahertz bands of 6G mobile communication technologies, multi-antenna transmission technologies such as Full Dimensional MIMO (FD-MIMO), array antennas and large-scale antennas, metamaterial-based lenses and antennas for improving coverage of terahertz band signals, high-dimensional space multiplexing technology using OAM (Orbital Angular Momentum), and RIS (Reconfigurable Intelligent Surface), but also full-duplex technology for increasing frequency efficiency of 6G mobile communication technologies and improving system networks, AI-based communication technology for implementing system optimization by utilizing satellites and AI (Artificial Intelligence) from the design stage and internalizing end-to-end AI support functions, and next-generation distributed computing technology for implementing services at levels of complexity exceeding the limit of UE operation capability by utilizing ultra-high-performance communication and computing resources.

[0040] In the present document, the word "exemplary" is used herein to mean "serving as an example, instance, or illustration." Any embodiment or implementation of the present subject matter described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments.

[0041] While the disclosure is susceptible to various modifications and alternative forms, specific embodiment thereof has been shown by way of example in the drawings and will be described in detail below. It should be understood, however that it is not intended to limit the disclosure to the particular forms disclosed, but on the contrary, the disclosure is to cover all modifications, equivalents, and alternative falling within the spirit and the scope of the disclosure.

[0042] The terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a setup, device or method that comprises a list of components or steps does not include only those components or steps but may include other components or steps not expressly listed or inherent to such setup or device or method. In other words, one or more elements in a device or system or apparatus proceeded by "comprises...a" does not, without more constraints, preclude the existence of other elements or additional elements in the device or system or apparatus.

[0043] The terms "dynamically" and "automatically" have been used interchangeably in the present disclosure.

[0044] The terms "model distribution" and "model provisioning" have been used interchangeably in the present disclosure.

[0045] In the following detailed description of the embodiments of the disclosure, reference is made to the accompanying drawing that form a part hereof, and in which are shown by way of illustration specific embodiments in which the disclosure may be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the disclosure, and it is to be understood that other embodiments may be utilized and that changes may be made without departing from the scope of the present disclosure. The following description is, therefore, not to be taken in a limiting sense.

[0046] REFERRAL NUMERALS:Reference numberDescription101Entity 1103Entity 2105Entity 3107FL / ML clients201I-O interface203Processor205Memory211Data213Selection data215ML model training data217Miscellaneous data221Modules223Transceiver225Monitoring module227Selecting module229Miscellaneous modules

[0047] Overview:

[0048] The present invention relates generally to the field of providing application layer support for AI / ML services, and more particularly to system and method for enabling the distributed and federated machine learning for the Application Functions (AF) (like, application servers, VAL servers etc...) by dynamically selecting the machine learning clients and distributing the machine learning model information to the selected machine learning clients for model training.

[0049] It proposes to address the explained problem on member selection and model distribution assistance in federated / distributed machine learning operations. Considering the dynamic nature (mobility, changing capabilities, varying resource conditions etc) of the UEs or servers (hosting FL / ML clients) available for machine learning operations, the AI / ML Services (like application layer enabler server, AIML application enabler server, ADAE server) may select the member FL / ML clients dynamically and distribute the model to selected FL / ML clients on behalf of machine learning server (like application server, application function, VAL server, etc.). The FL / ML client selection may be based on criteria (member selection information) shared by the machine learning server (like application server, application function, VAL server, etc.). Additionally, the AI / ML Services (application enabler server, AIML application enabler server) may monitor potential member FL / ML clients based on certain event triggers (member selection information) like FL / ML client entering in or exiting from an area of interest, new FL / ML client registration etc. These triggers and the criteria (member selection information) from machine learning server (like application server, application function, VAL server, etc.) may be used in combination in determining and selecting the member FL / ML clients.

[0050] The procedure for dynamically selecting FL / ML member clients and distributing machine learning model information to initiate machine learning model training on the selected FL / ML member clients involves selecting and distributing the machine learning model information to the FL / ML clients (i.e., VAL UEs, VAL servers) to participate in Entity-1's (e.g., VAL server, Machine learning server, Application server etc) model training (federation / distributed learning operations), based on trigger from certain events and the information (member selection information like selection criteria, indication to select FL / ML client members and distribute models automatically to FL / ML client members) received by Entity-2 (e.g., AIML enabler server, application enabler server, AIML Service server) from the Entity-1. Based on request from Entity-2, the FL / ML clients start the machine learning model training.

[0051] Entity-1 is the consumer of AIML application enabler services, which could be a VAL server, a machine learning server, an application server etc. Enity-2 is the provider of the AIML application enabler services, which could be an AIML enabler server, application enabler server, AIML service server, ADAE server, SEAL server etc. Entity-3 is the model repository where AIML model information is stored. FL / ML clients are the potential participants of the federation / distributed learning operations for Entity-1, which could be VAL UEs, VAL servers, AIML Enabler client, AIML application enabler client, AIML member, ADAE client, etc.

[0052] In step-1, the Entity-1 sends Model distribution request to Entity-2, requesting the Entity-2 to assist in its model training, that is, to monitor and select the FL / ML clients for federated / distributed machine learning, and distribute the machine learning model information to the selected FL / ML clients to initiate the machine learning model training on the selected FL / ML clients. Table-2 shows the information included in the request message. The Entity-1 may decide to include the model file directly in the request or the location / address from where the model file can be fetched.

[0053] Model distribution requestInformation elementDescriptionMember selection informationInformation related to selection of the FL / ML clients. This could include the member selection criteria / filter information like list of ML / FL clients to select the members from, VAL UE battery level, VAL UE velocity and direction, area of interest, available computer resources, etc.Machine Learning (ML) model informationInformation related to machine learning model that has to be distributed to the selected FL / ML clients. This information may consist of model identifier, ML model file, address (e.g., a URL or an FQDN) of the ML model file, model repository information related to the ML model, any additional model data etc.Time of distributionInformation related to the time or time period when the model can be distributed to FL / ML clients and / or when the model can be trained. This could include various information like date and time, day of week, schedule (combination of date, time and day of week, periodicity), etc.Members update notificationIndicates whether Entity-1 needs to be notified whenever there is update related to new FL / ML clients selected, model distributed and start of model training on the selected FL / ML clients.Existing membersInformation related to existing FL / ML member clients which are currently participating in federated / distributed learning.Allow new clientIndicates the entity-2 to select newly registered FL client or newly entered FL client in the location of interest for the model training

[0054] In step-2, if the Entity-1 is authorized to the request in step-1, then based on the information in model distribution request in step-1, the Entity-2 may accept the request and as shown in Table-3, send Model distribution response message to Entity-1 indicating that Entity-2 will monitor and select the FL / ML clients for federated / distributed machine learning, and distribute the machine learning model information to the selected FL / ML clients to automatically initiate the machine learning model training on the selected FL / ML clients.

[0055] Model distribution responseInformation elementDescriptionResultIndicates success or failure of the requestSuccessEntity-2 may monitor and select the FL / ML clients for federated / distributed machine learning, and distribute the machine learning model information to the selected FL / ML clients to automatically initiate the machine learning model training on the selected FL / ML clientsFailure causeIn case of failure, provides failure cause - like Entity-1 is not authorized or Entity-2 will not monitor and select the FL / ML clients for federated / distributed machine learning, and not distribute the machine learning model information, etc.

[0056] Based on the member selection information received in step-1, the Entity-2 monitors for potential FL / ML clients that can participate in federated / distribution learning. For instance, Entity-2 may subscribe to Service Enabler Architecture Layer (SEAL) and Network Exposure Function (NEF) location monitoring services as specified in TS 23.434, TS 23.501, TS 23.502, to identify new ML / FL clients or VAL UEs or UEs in a given area of interest, Entity-2 may monitor for new ML / FL client registrations, may use NEF or other core network capabilities to monitor the potential ML / FL clients or VAL UEs etc.

[0057] In steps 3 and 4, from the monitoring of potential FL / ML clients, when there is information available with Entity-2 on the potential FL / ML clients (like from events related to new FL / ML registrations, FL / ML client in a given area of interest etc), the Entity-2 determines and selects the new members (FL / ML clients) dynamically for federated / distributed learning, based on the availability of these potential FL / ML clients matching the member selection information.

[0058] If the machine learning model information (e.g., model file, URL or FQDN to fetch from, etc) is not available with Entity-2, then in step-5, the Entity-2 may fetch the information related to machine learning model that needs to be distributed / provisioned to ML / FL clients from Entity-3. The Entity-2 sends the Model information request message to Entity-3, including the information as shown in Table-4.

[0059] Model information requestInformation elementDescriptionModel identifierIdentifier of the machine learning model whose information is requested.Machine Learning (ML) model informationInformation related to machine learning model. This may consist of ML model file, address (e.g., a URL or an FQDN) of the ML model file, any additional model data etc.

[0060] In step-6, if the Entity-2 is authorized to request the model information, then the Entity-3 may share the model information as shown in Table-5, in the model information response message.

[0061] Model information responseInformation elementDescriptionModel identifierIdentifier of the machine learning model whose information is requested.Model fileThe model file.Model file locationAddress information (e.g., a URL or an FQDN) of the model file.Model fetch informationAny information on how to fetch the model from Entity-3.Additional informationAny additional information related to the model, like state (in training, trained), spatial validity, validity period for distribution and training of the model etcOtherAny other information related to the request.

[0062] Once the model information is available with Entity-2, then in step-7 the Entity-2 may distribute the model information to all the selected FL / ML member clients in ML Training request message, as shown in Table-6.

[0063] ML Training requestInformation elementDescriptionModel identifierIdentifier of the machine learning model whose information is requested.Model fileThe model file.Model file locationAddress information (e.g., a URL or an FQDN) of the model file.Model fetch informationAny information on how to fetch the model from Entity-3.Additional informationAny additional information related to the model, validity period for training of the model etc.Training IndicationIndicates if the FL / ML client should start the model training or stop the ongoing training (in case when FL client is exiting the area of interest). If not, then it can be an indication that the model information shared is only for the FL / ML client to have it stored locally.OtherAny other information related to the request.

[0064] In step-8, based on the model information shared by Entity-2, the ML / FL clients may either accept the model information in the request for training or may not accept the request. FL / ML clients may send the confirmation of training in ML Training response message to Entity-2. The response includes as shown in Table-7.

[0065] ML Training responseInformation elementDescriptionSuccessIndicates that the ML / FL client confirms the reception of model information (e.g., model file), will perform training.FailureIndicates that the ML / FL client cannot accept the request for model training based on the model information shared.

[0066] In Step-9, based on the number of confirmations received from ML / FL clients in step-8, the Entity-2 may notify the Entity-1 in Model distribution update message, about the set of FL / ML members selected and training a given model. The information sent is as shown in Table-8.

[0067] Model distribution updateInformation elementDescriptionList of ML / FL clients addedSet of new ML / FL clients that are selected for model training.Complete list of ML / FL clientsTotal set of ML / FL clients that are currently participating in model training.Model informationMachine learning model information related to the list of ML / FL clients. Indicates the model that is distributed and which is currently being trained by the ML / FL Clients. This information may consist of model identifier, address (e.g., a URL or an FQDN) of the ML model file, model repository information related to the ML model, any additional model data etc.

[0068] In an embodiment, the Entity-2, based on the member selection information, may identify that the existing member FL / ML clients which are currently participating in federated / distributed machine learning, are no longer eligible to be members. In such a scenario, in step 7, the Entity-2 may indicate to the FL / ML client(s) that they are no longer members to participate and removes them from the member list. The FL / ML client may stop model training in response to such request from Entity-2 and respond the same in response message.

[0069] In an embodiment, the set of member ML / FL clients participating in the federated / distributed learning for specific model(s) are maintained and managed in a group (like FL group). Entity-2 may manage the membership of this group, either on its own or with assistance of other enabler services like SEAL Group Management service as specified in TS 23.434. ML / FL client may be added or removed from the group based on availability or non-availability of the FL / ML client for federated / distributed learning. In step-3, the Entity-2 may add the newly determined ML / FL clients to the respective group related to the model.

[0070] In some embodiments, the processor may be disposed in communication with a communication network via a network interface. The communication network may couple the processor with the database. The network interface may communicate with the communication network. The network interface may employ connection protocols including, without limitation, direct connect, ethernet (e.g., twisted pair 10 / 100 / 1000 Base T), Transmission Control Protocol / Internet Protocol (TCP / IP), token ring, IEEE 802.11a / b / g / n / x, etc... The communication network 509 may include, without limitation, a direct interconnection, Local Area Network (LAN), Wide Area Network (WAN), wireless network (e.g., using Wireless Application Protocol), the Internet, etc... Using the network interface and the communication network, the computer system may communicate with a database, which may be the enrolled templates database. The network interface may employ connection protocols include, but not limited to, direct connect, ethernet (e.g., twisted pair 10 / 100 / 1000 Base T), Transmission Control Protocol / Internet Protocol (TCP / IP), token ring, IEEE 802.11a / b / g / n / x, etc.

[0071] The communication network includes, but is not limited to, a direct interconnection, a Peer-to-Peer (P2P) network, Local Area Network (LAN), Wide Area Network (WAN), wireless network (e.g., using Wireless Application Protocol), the Internet, Wi-Fi and such. The communication network may either be a dedicated network or a shared network, which represents an association of the different types of networks that use a variety of protocols, for example, Hypertext Transfer Protocol (HTTP), Transmission Control Protocol / Internet Protocol (TCP / IP), Wireless Application Protocol (WAP), etc..., to communicate with each other. Further, the communication network may include a variety of network devices, including routers, bridges, servers, computing devices, storage devices, etc...

[0072] In some embodiments, the processor may be disposed in communication with a memory (e.g., RAM 512, ROM 513, etc...) via a storage interface. The storage interface may connect to memory including, without limitation, memory drives, removable disc drives, etc..., employing connection protocols such as, Serial Advanced Technology Attachment (SATA), Integrated Drive Electronics (IDE), IEEE-1394, Universal Serial Bus (USB), fibre channel, Small Computer Systems Interface (SCSI), etc... The memory drives may further include a drum, magnetic disc drive, magneto-optical drive, optical drive, Redundant Array of Independent Discs (RAID), solid-state memory devices, solid-state drives, etc...

[0073] Furthermore, one or more computer-readable storage media may be utilized in implementing embodiments consistent with the present disclosure. A computer-readable storage medium refers to any type of physical memory on which information or data readable by a processor may be stored. Thus, a computer-readable storage medium may store instructions for execution by one or more processors, including instructions for causing the processor(s) to perform steps or stages consistent with the embodiments described herein. The term "computer-readable medium" should be understood to include tangible items and exclude carrier waves and transient signals, i.e., be non-transitory. Examples include Random Access Memory (RAM), Read-Only Memory (ROM), volatile memory, non-volatile memory, hard drives, Compact Disc (CD) ROMs, DVDs, flash drives, disks, and any other known physical storage media.

[0074] Fig. 1illustrates a procedure for dynamically selecting FL / ML member clients and distributing machine learning model information to initiate machine learning model training the selected FL / ML member clients, according to the embodiments of the present disclosure.

[0075] The procedure involves selecting and distributing machine learning model information to FL / ML clients (i.e., VAL UEs, VAL servers) to participate in Entity-2's (e.g., VAL server, Machine learning server, Application server, etc.) model training (Federation Learning / distributed Machine Learning (FL / ML) operations), based on trigger from certain events and information (member selection information like selection criteria, indication to select FL / ML client members and distribute models automatically to FL / ML client members) received by Entity-1 (e.g., Artificial Intelligence Distributed Machine Learning (AIML) enabler server, application enabler server, AIML Service server) from the Entity-2. Based on request from Entity-1, the FL / ML clients start machine learning model training.

[0076] With reference to Fig. 1, the environment comprises of an entity 1 (also, referred as first entity) 101, an entity 2 ( also, referred as second entity) 103, an entity 3 ( also, referred as third entity) 105, and a plurality of clients 107. The entity 1 101 is the provider of the AIML application enabler services, which may be one of, but not limited to, an AIML enabler server, an application enabler server, an AIML service server, an ADAE server, or a SEAL server. The entity 2 103 is the consumer of AIML application enabler services, which may be one of, but not limited to, a VAL server, a machine learning server, or an application server. The entity 3 105 may be one of, but not limited to, a model repository, or a model storage where AIML model information is stored. The plurality of clients 107 may, also, be referred as FL / ML clients or FL / ML member clients. The plurality of clients 107 are the potential participants of the FL / ML operations for the entity 2. The plurality of clients 107 may be Vertical Application Layer (VAL) UEs, VAL servers, AIML Enabler clients, AIML application enabler clients, AIML members, and / or Application Data Analytics Enablement (ADAE) clients.

[0077] The entity 1 (i.e., first entity) 101 refers to the Entity-2, and the entity 2 (i.e., second entity) 103 refers to the Entity-1 under the overview section as described above.

[0078] Hereinafter, the operation for machine learning model distribution is explained with reference to Fig. 1.

[0079] At step 111, the entity 1 101 receives a model distribution request containing the client selection information from the entity 2 103. The client selection information comprises at least one of client battery level, client velocity and direction, an area of interest, and a list of the plurality of clients 107. In detail, the entity 2 103 sends the model distribution request to the entity 1 101, requesting the entity 1 101 to assist in its model training, that is, to monitor and select the FL / ML clients for federated / distributed machine learning, and distribute the machine learning model information to the selected FL / ML clients to initiate the machine learning model training on the selected FL / ML clients. Table-9 shows the information included in the model distribution request message. The entity 1 103 may decide to include the model file directly in the request or the location / address from where the model file can be fetched.

[0080] Model distribution requestInformation elementDescriptionMember selection information (i.e., client selection information)Information related to selection of the FL / ML clients. This could include the member selection criteria / filter information like list of ML / FL clients to select the members from (i.e., a list of the plurality of clients 107), VAL UE battery level (i.e., client battery level), VAL UE velocity and direction (i.e., client velocity and direction), area of interest, available computer resources, etc.Machine Learning (ML) model informationInformation related to machine learning model that has to be distributed to the selected FL / ML clients. This information may consist of model identifier, ML model file, address (e.g., a URL or an FQDN) of the ML model file, model repository information related to the ML model, any additional model data etc.Time of distributionInformation related to the time or time period when the model can be distributed to FL / ML clients and / or when the model can be trained. This could include various information like date and time, day of week, schedule (combination of date, time and day of week, periodicity), etc.Members update notificationIndicates whether the entity 2 103 needs to be notified whenever there is update related to new FL / ML clients selected, model distributed and start of model training on the selected FL / ML clients.Existing membersInformation related to existing FL / ML member clients which are currently participating in federated / distributed learning.Allow new clientIndicates the entity 1 101 to select newly registered FL client or newly entered FL client in the location of interest for the model training

[0081] At step 113, the entity 1 101 checks whether the entity 2 103 is authorized to request to train the ML model upon receiving the model distribution request from the entity 2 103. Thereafter, the entity 1 101 sends a model distribution response containing information on success or failure of the model distribution request to the entity 2 103. In detail, if the entity 2 103 is authorized to the request in the step 111, then based on the information in the model distribution request in the step 111, the entity 1 101 may accept the request and as shown in Table-10, sends the model distribution response message to the entity 2 103 indicating that the entity 1 101 will monitor and select the FL / ML clients for federated / distributed machine learning, and distribute the machine learning model information to the selected FL / ML clients to automatically initiate the machine learning model training on the selected FL / ML clients.

[0082] Model distribution responseInformation elementDescriptionResultIndicates success or failure of the requestSuccessEntity 1 101 may monitor and select the FL / ML clients for federated / distributed machine learning, and distribute the machine learning model information to the selected FL / ML clients to automatically initiate the machine learning model training on the selected FL / ML clientsFailure causeIn case of failure, provides failure cause - like entity 2 103 is not authorized or entity 1 101 will not monitor and select the FL / ML clients for federated / distributed machine learning, and not distribute the machine learning model information, etc.

[0083] Based on the client selection information received in the step 111, the entity 1 101 monitors for potential FL / ML clients that can participate in federated / distribution learning. For instance, the entity 1 101 may subscribe to SEAL and NEF location monitoring services as specified in TS 23.434, TS 23.501, TS 23.502, to identify new ML / FL clients or VAL UEs or UEs in a given area of interest, entity 1 101 may monitor for new ML / FL client registrations, may use NEF, or other core network capabilities to monitor the potential ML / FL clients or VAL UEs, etc.

[0084] At step 115, the entity 1 101 monitors continuously the plurality of clients 107 to select the at least one client based on the client selection information. Thereafter, the entity 1 101 selects dynamically at least one client from the plurality of clients 107 based on client selection information related to the plurality of clients 107 received from the entity 2 103 associated with the entity 1 101, for one of federated or distributed machine learning. In detail, in the step 115, from the monitoring of potential FL / ML clients, when there is information available with the entity 1 101 on the potential FL / ML clients (like from events related to new FL / ML registrations, FL / ML client in a given area of interest, etc), the entity 1 101 determines / monitors and selects the new members (FL / ML clients) dynamically for federated / distributed learning, based on the availability of these potential FL / ML clients matching the member selection information.

[0085] At step 117, the entity 1 101 sends a model information request to the entity 3 105, when the information pertaining to the ML model to be sent to the at least one client is not available with the entity 1 101. The model information request comprises at least one of a ML model identifier, and ML model information. In detail, if the machine learning model information (e.g., model file, URL or FQDN to fetch from, etc) is not available with the entity 1 101, then in the step 117, the entity 1 101 may fetch the information related to machine learning model that needs to be distributed / provisioned to ML / FL clients from the entity 3 105. The entity 1 101 sends the model information request message to the entity 3 105, including the information as shown in Table-11.

[0086] Model information requestInformation elementDescriptionModel identifier (i.e., ML model identifier)Identifier of the machine learning model whose information is requested.Machine Learning (ML) model informationInformation related to machine learning model. This may consist of ML model file, address (e.g., a URL or an FQDN) of the ML model file, any additional model data etc.

[0087] At step 119, the entity 1 101 receives a model information response from the entity 3 105 upon receiving the model information request. The model information response comprises at least one of the ML model identifier, a ML model file, a ML model file location, a ML model fetch information, and a validity period for training of the ML model. In detail, if the entity 1 101 is authorized to request the model information, then the entity 3 105 may share the model information as shown in Table-12, in the model information response message.

[0088] Model information responseInformation elementDescriptionModel identifier (i.e., ML model identifier)Identifier of the machine learning model whose information is requested.Model file (i.e., ML model file)The model file.Model file location (i.e., ML model file location)Address information (e.g., a Uniform Resource Locator (URL) or a Fully Qualifed Domain Name (FQDN)) of the model file.Model fetch information (i.e., ML model fetch information)Any information on how to fetch the model from entity 3 105.Additional informationAny additional information related to the model, like state (in training, trained), spatial validity, validity period for distribution and training of the model (i.e., validity period for training of the ML model), etcOtherAny other information related to the request.

[0089] At step 121, the entity 1 101 sends a request to train a Machine Learning (ML) model to the at least one client selected from the plurality of clients 107. The request comprises information pertaining to a ML model to be trained by the at least one client. The information pertaining to the ML model to be trained by the at least one client comprises at least one of a ML model identifier, a ML model file, a ML model file location, a ML model fetch information, a validity period for training of the ML model, an indication of whether the entity 1 101 wants to be notified whenever update happens to a list of the at least one client selected for participating in training of the ML model, and a training indication. In detail, once the model information is available with entity 1 101, then in the step 121, the entity 1 101 may distribute / send the model information to all the selected FL / ML member clients in ML Training request message, as shown in Table-13.

[0090] ML Training requestInformation elementDescriptionModel identifier (i.e., ML model identifier)Identifier of the machine learning model whose information is requested.Model file (i.e., ML model file)The model file.Model file location (i.e., ML model file location)Address information (e.g., a URL or an FQDN) of the model file.Model fetch information (i.e., ML model fetch information)Any information on how to fetch the model from entity 3 105.Additional informationAny additional information related to the model, validity period for training of the model (i.e., validity period for training of the ML model), etc.Training IndicationIndicates if the FL / ML client should start the model training or stop the ongoing training (in case when FL client is exiting the area of interest). If not, then it can be an indication that the model information shared is only for the FL / ML client to have it stored locally.OtherAny other information related to the request.

[0091] At step 123, the entity 1 101 receives a response from the at least one client, containing confirmation information on acceptance or rejection of the request to train the ML model for one of federated or distributed machine learning. In detail, in the step 123, based on the model information shared by the entity 1 101, the ML / FL clients may either accept the model information in the request for training or may not accept the request. FL / ML clients may send the confirmation of training in ML Training response message to the entity 1 101. The response includes as shown in Table-14.

[0092] ML Training responseInformation elementDescriptionSuccessIndicates that the ML / FL client confirms the reception of model information (e.g., model file), will perform training.FailureIndicates that the ML / FL client cannot accept the request for model training based on the model information shared.

[0093] At step 125, the entity 1 101 sends a model distribution update message containing the information pertaining to training of the ML model by the at least one client to the entity 2 103 upon receiving the response from the at least one client. The information pertaining to training of the ML model comprises ML model information, a list of the at least one client selected for participating in training of the ML model. In detail, in the step 125, based on the number of confirmations received from ML / FL clients in the step 123, the entity 1 101 may notify / send the entity 2 103 in the model distribution update message, about the set of FL / ML members selected and training a given model. The information sent is as shown in Table-15.

[0094] Model distribution updateInformation elementDescriptionList of ML / FL clients added (i.e., a list of the at least one client selected for participating in training of the ML model)Set of new ML / FL clients that are selected for model training.Complete list of ML / FL clientsTotal set of ML / FL clients that are currently participating in model training.Model information (i.e., ML model information)Machine learning model information related to the list of ML / FL clients. Indicates the model that is distributed and which is currently being trained by the ML / FL Clients. This information may consist of model identifier, address (e.g., a URL or an FQDN) of the ML model file, model repository information related to the ML model, any additional model data etc.

[0095] In an embodiment, the entity 1 101, based on the member selection information, may identify that the existing member FL / ML clients which are currently participating in federated / distributed machine learning, are no longer eligible to be members. In such a scenario, in step 121, the entity 1 101 may indicate to the FL / ML client(s) that they are no longer members to participate and removes them from the member list. The FL / ML client may stop model training in response to such request from the entity 1 101 and respond the same in response message.

[0096] In an embodiment, the set of member ML / FL clients participating in the federated / distributed learning for specific model(s) are maintained and managed in a group (like FL group). The entity 1 101 may manage the membership of this group, either on its own or with assistance of other enabler services like SEAL Group Management service as specified in TS 23.434. ML / FL client may be added or removed from the group based on availability or non-availability of the FL / ML client for federated / distributed learning. In the step 115, the entity 1 101 may add the newly determined ML / FL clients to the respective group related to the model.

[0097] In an embodiment, a system for dynamic machine learning model distribution may correspond to the entity 1 (i.e., first entity) 101.

[0098] Fig. 2shows a detailed block diagram of an entity 1 101 for dynamic machine learning model distribution, according to the embodiments of the present disclosure.

[0099] The entity 1 (i.e., first entity) 101 is one of, but not limited to, an Artificial Intelligence Distributed Machine Learning (AIML) enabler server, an AIML service server, or an application enabler server. The entity 1 101 for dynamic machine learning model distribution includes an I-O interface 201, a processor 203, and a memory 205. In the present embodiment, data 211 is stored within the memory 205.

[0100] The I-O interface 201 is configured to receive requests, and to transmit responses. The I-O interface 201 employs communication protocols or methods such as, without limitation, audio, analog, digital, monoaural, Radio Corporation of America (RCA) connector, stereo, IEEE®-1394 high speed serial bus, serial bus, Universal Serial Bus (USB), infrared, Personal System / 2 (PS / 2) port, Bayonet Neill-Concelman (BNC) connector, coaxial, component, composite, Digital Visual Interface (DVI), High-Definition Multimedia Interface (HDMI®), Radio Frequency (RF) antennas, S-Video, Video Graphics Array (VGA), IEEE® 802.11b / g / n / x, Bluetooth, cellular e.g., Code-Division Multiple Access (CDMA), High-Speed Packet Access (HSPA+), Global System for Mobile communications (GSM®), Long-Term Evolution (LTE®), Worldwide interoperability for Microwave access (WiMax®), or the like.

[0101] The memory 205 is communicatively coupled to the processor 203 of the entity 1 101. The memory 205, also, stores processor-executable instructions which cause the processor 203 to execute the instructions for dynamic machine learning model distribution. The memory 205 includes, without limitation, memory drives, removable disc drives, etc. The memory drives may further include a drum, magnetic disc drive, magneto-optical drive, optical drive, RAID, solid-state memory devices, solid-state drives, etc.

[0102] The processor 203 includes at least one data processor for dynamic machine learning model distribution. The processor 203 may include specialized processing units such as integrated system (bus) controllers, memory management control units, floating point units, graphics processing units, digital signal processing units, etc.

[0103] The data 211 include, for example, selection data 213, ML model training data 215, and miscellaneous data 217.

[0104] Selection data 213: The selection data 213 stores client selection information. The client selection information comprises at least one of client battery level, client velocity and direction, an area of interest, and a list of the plurality of clients 107.

[0105] ML model training data 215: The ML model training data 215 comprises information pertaining to a ML model to be trained by the at least one client. The information pertaining to the ML model to be trained by the at least one client comprises at least one of a ML model identifier, a ML model file, a ML model file location, a ML model fetch information, a validity period for training of the ML model, an indication of whether the entity 1 101 wants to be notified whenever update happens to a list of the at least one client selected for participating in training of the ML model, and a training indication.

[0106] Miscellaneous data 217: The miscellaneous data 217 stores data, including meta data, and temporary files, generated by the modules of the entity 1 101 for performing the various functions of the entity 1 101.

[0107] In the embodiment of the present disclosure, the data 211 in the memory 205 are processed by the one or more modules 221 (also, referred as modules) of the entity 1 101. In the embodiment, the one or more modules 221 are implemented as dedicated hardware units (e.g., circuits). As used herein, the term modules / units refers to, for example, an ASIC, an electronic circuit, a PSoC, a combinational logic circuit, and / or other suitable components that provide the described functionality. In one embodiment of the present disclosure, the one or more modules 221 are communicatively coupled to the processor 203 for machine learning model distribution. The one or more modules 221 when configured with the functionality defined in the present disclosure results in a novel hardware.

[0108] In one implementation, the one or more modules 221 include, but are not limited to, a transceiver 223, a monitoring module 225, and a selecting module 227. The one or more modules 221, also, includes miscellaneous modules 229 to perform various miscellaneous functionalities of the entity 1 101.

[0109] Transceiver 223: The transceiver 223 of the entity 1 101 receives a model distribution request containing the client selection information from the entity 2 103. The transceiver 223 of the entity 1 101 sends a model distribution response containing information on success or failure of the model distribution request to the entity 2 103.

[0110] The transceiver 223 of the entity 1 101 sends a request to train a Machine Learning (ML) model to the at least one client selected from the plurality of clients 107. The request comprises information pertaining to a ML model to be trained by the at least one client. The transceiver 223 of the entity 1 101 receives a response from the at least one client, containing confirmation information on acceptance or rejection of the request to train the ML model for one of federated or distributed machine learning. The transceiver 223 of the entity 1 101 sends a model distribution update message containing the information pertaining to training of the ML model by the at least one client to the entity 2 103 upon receiving the response from the at least one client.

[0111] In one embodiment, the transceiver 223 of the entity 1 101 sends a model information request to an entity 3 105, when the information pertaining to the ML model to be sent to the at least one client is not available with the entity 1 101. The model information request comprises at least one of a ML model identifier, and ML model information. The transceiver 223 of the entity 1 101 receives a model information response from the entity 3 105 upon receiving the model information request. The model information response comprises at least one of the ML model identifier, a ML model file, a ML model file location, a ML model fetch information, and a validity period for training of the ML model.

[0112] Monitoring module 225: The monitoring module 225 of the entity 1 101 monitors continuously the plurality of clients 107 to select the at least one client based on the client selection information. The monitoring module 225 of the entity 1 101 checks whether the entity 2 103 is authorized to request to train the ML model upon receiving the model distribution request from the entity 2 103. The plurality of clients comprises at least one of one or more newly registered clients with the entity 1 101 and one or more newly entered clients in an area of interest.

[0113] Selecting module 227: The selecting module 227 of the entity 1 101 selects dynamically at least one client from the plurality of clients 107 based on client selection information related to the plurality of clients 107 received from the entity 2 103 associated with the entity 1 101, for one of federated or distributed machine learning. The client selection information comprises at least one of client battery level, client velocity and direction, an area of interest, and a list of the plurality of clients 107.

[0114] Fig. 3illustrates a flowchart showing a method for dynamic machine learning model distribution, according to the embodiments of the present disclosure.

[0115] As illustrated in Fig. 3, the method 300 includes one or more operation steps for machine learning model distribution in accordance with some embodiments of the present disclosure. The method 300 may be described in the general context of computer executable instructions. Generally, computer executable instructions can include routines, programs, objects, components, data structures, procedures, modules, and functions, which perform particular functions or implement particular abstract data types.

[0116] The order in which the method 300 is described is not intended to be construed as a limitation, and any number of the described method operation steps can be combined in any order to implement the method. Additionally, individual operation steps may be deleted from the methods without departing from the scope of the subject matter described herein. Furthermore, the method can be implemented in any suitable hardware, software, firmware, or combination thereof.

[0117] At operation step 301, the entity 1 101 (i.e., first entity) receives a model distribution request containing the client selection information from the entity 2 103 (i.e., second entity). The entity 1 101 is one of an Artificial Intelligence Distributed Machine Learning (AIML) enabler server, an AIML service server, or an application enabler server. The client selection information comprises at least one of client battery level, client velocity and direction, an area of interest, and a list of the plurality of clients 107.

[0118] At operation step 303, the entity 1 101 checks whether the entity 2 103 is authorized to request to train the ML model upon receiving the model distribution request.

[0119] At operation step 305, the entity 1 101 sends a model distribution response containing information on success or failure of the model distribution request to the entity 2 103.

[0120] At operation step 307, the entity 1 101 monitors continuously the plurality of clients 107 to select the at least one client based on the client selection information. The plurality of clients comprises at least one of one or more newly registered clients with the entity 1 101 and one or more newly entered clients in the area of interest.

[0121] At operation step 309, the entity 1 101 selects dynamically at least one client from the plurality of clients 107 based on client selection information related to the plurality of clients 107 received from the entity 2 103 associated with the entity 1 101, for one of federated or distributed machine learning.

[0122] At operation step 311, the entity 1 101 send a request to train a Machine Learning (ML) model to the at least one client selected from the plurality of clients 107. The request comprises information pertaining to a ML model to be trained by the at least one client. The information pertaining to the ML model to be trained by the at least one client comprises at least one of a ML model identifier, a ML model file, a ML model file location, a ML model fetch information, a validity period for training of the ML model, an indication of whether the entity 1 101 wants to be notified whenever update happens to a list of the at least one client selected for participating in training of the ML model, and a training indication.

[0123] At operation step 313, the entity 1 101 receives a response from the at least one client, containing confirmation information on acceptance or rejection of the request to train the ML model for one of federated or distributed machine learning.

[0124] At operation step 315, the entity 1 101 sends a model distribution update message containing the information pertaining to training of the ML model by the at least one client to the entity 2 103 upon receiving the response from the at least one client. The information pertaining to training of the ML model comprises ML model information, a list of the at least one client selected for participating in training of the ML model.

[0125] Some of the clauses are mentioned below.

[0126] [1]: A method for machine learning model distribution, comprising:

[0127] selecting dynamically, by a first entity 101, at least one client from a plurality of clients 107 based on client selection information related to the plurality of clients 107 received from a second entity 103 associated with the first entity 101, for one of federated or distributed machine learning; and

[0128] sending, by the first entity 101, a request to train a Machine Learning (ML) model to the at least one client selected from the plurality of clients 107,

[0129] wherein the request comprises information pertaining to a ML model to be trained by the at least one client.

[0130] [2]: The method described in [1], further comprises:

[0131] receiving, by the first entity 101, a response from the at least one client, containing confirmation information on acceptance or rejection of the request to train the ML model for one of federated or distributed machine learning.

[0132] [3]: The method described in [1], prior to selecting dynamically the at least one client from the plurality of clients 107 based on the client selection information, the method comprises:

[0133] receiving, by the first entity 101, a model distribution request containing the client selection information from the second entity 103;

[0134] checking, by the first entity 101, whether the second entity 103 is authorized to request to train the ML model upon receiving the model distribution request; and

[0135] sending, by the first entity 101, a model distribution response containing information on success or failure of the model distribution request to the second entity 103.

[0136] [4]: The method described in [1], prior to selecting dynamically the at least one client from the plurality of clients 107 based on the client selection information, the method comprises:

[0137] monitoring continuously, by the first entity 101, the plurality of clients 107 to select the at least one client based on the client selection information.

[0138] [5]: The method described in [2], further comprising:

[0139] sending, by the first entity 101, a model distribution update message containing the information pertaining to training of the ML model by the at least one client to the second entity 103 upon receiving the response from the at least one client.

[0140] [6]: The method described in [5], wherein the information pertaining to training of the ML model comprises ML model information, a list of the at least one client selected for participating in training of the ML model.

[0141] [7]: The method described in [1], prior to sending the request to the at least one client, the method comprises:

[0142] sending, by the first entity 101, a model information request to a third entity 105, when the information pertaining to the ML model to be sent to the at least one client is not available with the first entity 101,

[0143] wherein the model information request comprises at least one of a ML model identifier, and ML model information; and

[0144] receiving, by the first entity 101, a model information response from the third entity 105 upon receiving the model information request,

[0145] wherein the model information response comprises at least one of the ML model identifier, a ML model file, a ML model file location, a ML model fetch information, and a validity period for training of the ML model.

[0146] [8]: The method described in [1], wherein the information pertaining to the ML model to be trained by the at least one client comprises at least one of a ML model identifier, a ML model file, a ML model file location, a ML model fetch information, a validity period for training of the ML model, an indication of whether the first entity 101 wants to be notified whenever update happens to a list of the at least one client selected for participating in training of the ML model, and a training indication.

[0147] [9]: The method described in [1], wherein the client selection information comprises at least one of client battery level, client velocity and direction, an area of interest, and a list of the plurality of clients 107.

[0148]

[0010] : The method described in [1], wherein the plurality of clients 107 comprises at least one of one or more newly registered clients with the first entity 101 and one or more newly entered clients in an area of interest.

[0149]

[0011] : The method described in [1], wherein the first entity 101 is one of an Artificial Intelligence Distributed Machine Learning (AIML) enabler server, an AIML service server, or an application enabler server.

[0150]

[0012] : A first entity 101 for dynamic machine learning model distribution, comprising:

[0151] a processor; and

[0152] a memory communicatively coupled to the processor, wherein the memory stores processor-executable instructions, which on execution, cause the processor to:

[0153] select dynamically at least one client from a plurality of clients 107 based on client selection information related to the plurality of clients 107 received from a second entity 103 associated with the first entity 101, for one of federated or distributed machine learning; and

[0154] send a request to train a Machine Learning (ML) model to the at least one client selected from the plurality of clients 107,

[0155] wherein the request comprises information pertaining to a ML model to be trained by the at least one client.

[0156]

[0013] : The first entity 101 described in

[0012] , wherein the processor is configured to:

[0157] receive a response from the at least one client, containing confirmation information on acceptance or rejection of the request to train the ML model for one of federated or distributed machine learning.

[0158]

[0014] : The first entity 101 described in

[0012] , prior to selecting dynamically the at least one client from the plurality of clients 107 based on the client selection information, the processor is configured to:

[0159] receive a model distribution request containing the client selection information from the second entity 103;

[0160] check whether the second entity 103 is authorized to request to train the ML model upon receiving the model distribution request; and

[0161] send a model distribution response containing information on success or failure of the model distribution request to the second entity 103.

[0162]

[0015] : The first entity 101 described in

[0012] , prior to selecting dynamically the at least one client from the plurality of clients 107 based on the client selection information, the processor is configured to:

[0163] monitor continuously the plurality of clients 107 to select the at least one client based on the client selection information.

[0164]

[0016] : The first entity 101 described in

[0013] , wherein the processor is configured to:

[0165] send a model distribution update message containing the information pertaining to training of the ML model by the at least one client to the second entity 103 upon receiving the response from the at least one client.

[0166]

[0017] : The first entity 101 described in

[0016] , wherein the information pertaining to training of the ML model comprises ML model information, a list of the at least one client selected for participating in training of the ML model.

[0167]

[0018] : The first entity 101 described in

[0012] , prior to sending the request to the at least one client, the processor is configured to:

[0168] send a model information request to a third entity 105, when the information pertaining to the ML model to be sent to the at least one client is not available with the second entity 103,

[0169] wherein the model information request comprises at least one of a ML model identifier, and ML model information; and

[0170] receive a model information response from the third entity 105 upon receiving the model information request,

[0171] wherein the model information response comprises at least one of the ML model identifier, a ML model file, a ML model file location, a ML model fetch information, and a validity period for training of the ML model.

[0172]

[0019] : The first entity 101 described in

[0012] , wherein the information pertaining to the ML model to be trained by the at least one client comprises at least one of a ML model identifier, a ML model file, a ML model file location, a ML model fetch information, a validity period for training of the ML model, an indication of whether the second entity 103 wants to be notified whenever update happens to a list of the at least one client selected for participating in training of the ML model and a training indication.

[0173]

[0020] : The first entity 101 described in

[0012] , wherein the client selection information comprises at least one of client battery level, client velocity and direction, an area of interest, and a list of the plurality of clients 107.

[0174]

[0021] : The first entity 101 described in

[0012] , wherein the plurality of clients 107 comprises at least one of one or more newly registered clients with the second entity 103 and one or more newly entered clients in an area of interest.

[0175]

[0022] : The first entity 101 described in

[0012] , wherein the first entity 101 is one of an Artificial Intelligence Distributed Machine Learning (AIML) enabler server, an AIML service server, or an application enabler server.

[0176] The described operations may be implemented as a method, system or article of manufacture using standard programming and / or engineering techniques to produce software, firmware, hardware, or any combination thereof.　The described operations may be implemented as code maintained in a "non-transitory computer readable medium", where a processor may read and execute the code from the computer readable medium. The processor is at least one of a microprocessor and a processor capable of processing and executing the queries. A non-transitory computer readable medium may include media such as magnetic storage medium (e.g., hard disk drives, floppy disks, tape, etc...), optical storage (CD-ROMs, DVDs, optical disks, etc...), volatile and non-volatile memory devices (e.g., EEPROMs, ROMs, PROMs, RAMs, DRAMs, SRAMs, Flash Memory, firmware, programmable logic, etc...), etc... Further, non-transitory computer-readable media may include all computer-readable media except for a transitory. The code implementing the described operations may further be implemented in hardware logic (e.g., an integrated circuit chip, Programmable Gate Array (PGA), Application Specific Integrated Circuit (ASIC), etc...).

[0177] The illustrated steps are set out to explain the exemplary embodiments shown, and it should be anticipated that ongoing technological development will change the manner in which particular functions are performed. These examples are presented herein for purposes of illustration, and not limitation. Further, the boundaries of the functional building blocks have been arbitrarily defined herein for the convenience of the description. Alternative boundaries can be defined so long as the specified functions and relationships thereof are appropriately performed. Alternatives (including equivalents, extensions, variations, deviations, etc..., of those described herein) will be apparent to persons skilled in the relevant art(s) based on the teachings contained herein. Such alternatives fall within the scope and spirit of the disclosed embodiments. Also, the words "comprising," "having," "containing," and "including," and other similar forms are intended to be equivalent in meaning and be open ended in that an item or items following any one of these words is not meant to be an exhaustive listing of such item or items or meant to be limited to only the listed item or items. It must also be noted that as used herein, the singular forms "a," "an," and "the" include plural references unless the context clearly dictates otherwise.

[0178] Furthermore, one or more computer-readable storage media may be utilized in implementing embodiments consistent with the present disclosure. A computer readable storage medium refers to any type of physical memory on which information or data readable by a processor may be stored. Thus, a computer readable storage medium may store instructions for execution by one or more processors, including instructions for causing the processor(s) to perform steps or stages consistent with the embodiments described herein. The term "computer readable medium" should be understood to include tangible items and exclude carrier waves and transient signals, i.e., are non-transitory. Examples include Random Access Memory (RAM), Read-Only Memory (ROM), volatile memory, non-volatile memory, hard drives, CD ROMs, DVDs, flash drives, disks, and any other known physical storage media.

[0179] The illustrated operations of Fig. 3 shows certain events occurring in a certain order. In alternative embodiments, certain operations may be performed in a different order, modified, or removed. Moreover, steps may be added to the above-described logic and still conform to the described embodiments. Further, operations described herein may occur sequentially or certain operations may be processed in parallel. Yet further, operations may be performed by a single processing unit or by distributed processing units.

[0180] Finally, the language used in the specification has been principally selected for readability and instructional purposes, and it may not have been selected to delineate or circumscribe the inventive subject matter. It is therefore intended that the scope of the invention be limited not by this detailed description, but rather by any claims that issue on an application based here on. Accordingly, the disclosure of the embodiments of the invention is intended to be illustrative, but not limiting, of the scope of the invention, which is set forth in the following claims.

[0181] While various aspects and embodiments have been disclosed herein, other aspects and embodiments will be apparent to those skilled in the art. The various aspects and embodiments disclosed herein are for purposes of illustration and are not intended to be limiting, with the true scope being indicated by the following claims.

Claims

1.A method by an artificial intelligence distributed machine learning enabler (AIMLE) server for machine learning (ML) model distribution in a wireless communication system, the method comprising:receiving, from a vertical application layer (VAL) server, an ML model training request to assist in ML model training;identifying whether the VAL server is authorized to transmit the ML model training request; andtransmitting, to the VAL server, an ML model training response in response to the ML model training request.2.The method of claim 1, the ML model training request includes at least one of: member selection information, machine learning model information, time of distribution, members update notification, existing members information, or allow client information.3.The method of claim 1, wherein the ML model training response includes result of the ML model training request,wherein, in case that the VAL server is authorized, the result of the ML model training request indicates success of the ML model training request, or in case that the VAL server is not authorized, the result of the ML model training request indicates failure of the ML model training request and includes a reason for failure information.4.The method of claim 1, further comprising:monitoring clients by interacting with a network exposure function (NEF) and service enabler architecture layer (SEAL) services, to identify user equipment (UE) in a target location.5.The method of claim 1, further comprising:transmitting, to clients, a federated learning (FL) training request; andreceiving, from the clients, an FL training response.6.The method of claim 5, wherein the FL training request includes at least one of: model identifier, model file, model file location, model fetch information, additional information related to the ML model distribution, or training indication information.7.The method of claim 5, wherein the FL training response includes success information related to ML training or failure information related to the ML training.8.The method of claim 1, further comprising:notifying the VAL server to update a list of federated learning / machine learning (FL / ML) clients selected for the ML model training.9.An artificial intelligence distributed machine learning enabler (AIMLE) server entity for machine learning (ML) model distribution in a wireless communication system, the AIMLE server entity comprising:a transceiver; anda processor coupled to the transceiver, and configured to:receive, from a vertical application layer (VAL) server entity, an ML model training request to assist in ML model training,identify whether the VAL server entity is authorized to transmit the ML model training request, andtransmit, to the VAL server entity, an ML model training response in response to the ML model training request.10.The AIMLE server entity of claim 9, the ML model training request includes at least one of: member selection information, machine learning model information, time of distribution, members update notification, existing members information, or allow client information.11.The AIMLE server entity of claim 9, the ML model training response includes result of the ML model training request,wherein, in case that the VAL server entity is authorized, the result of the ML model training request indicates success of the ML model training request, or in case that the VAL server entity is not authorized, the result of the ML model training request indicates failure of the ML model training request and includes reason for failure information.12.The AIMLE server entity of claim 9, the processor further configured to:monitor clients by interacting with a network exposure function (NEF) and service enabler architecture layer (SEAL) services, to identify user equipment (UE) in a target location.13.The AIMLE server entity of claim 9, the processor further configured to:transmit, to clients entities, a federated learning (FL) training request; andreceive, from the clients entities, an FL training response.14.The AIMLE server entity of claim 13, wherein the FL training request includes at least one of: model identifier, model file, model file location, model fetch information, additional information related to the ML model distribution, or training indication information.15.The method of claim 13, wherein the FL training response includes success information related to ML training or failure information related to the ML training.

Images