A method for providing machine learning models to mobile devices.

Abstract

A method for providing a machine learning model to a mobile terminal of a mobile communication system is described, comprising the steps of determining, by a mobile wireless communication network of the communication system, that a machine learning model to be used by the mobile terminal for a machine learning task should be distributed to the mobile terminal, and providing the machine learning model to the mobile terminal via user plane communication.

Description

Technical Field

[0001] The present disclosure relates to a method for providing a machine learning model to a mobile terminal.

Background Art

[0002] In a communication system such as a 5G mobile communication system, it is important to ensure that a specific quality of service can be maintained. For this reason, various information such as load, resource usage, available components, user mobility, component status, etc. may be monitored and may be considered when controlling the communication system. For example, in order to avoid a decrease in the quality of service, measures may be taken when an overload is imminent. The collection and / or evaluation of this information can be performed using a machine learning model appropriately trained for the corresponding prediction task. For example, using artificial intelligence (AI) and machine learning (ML) technologies is a promising approach for improving the performance of the next-generation radio access network (NG-RAN) air interface in several use cases such as improving channel state information (CSI) feedback, optimizing beam management, and improving positioning accuracy. However, this requires the use of a machine learning model on the terminal side in at least some cases. Due to the effort required for training, model training on the terminal side is difficult (due to limited power and computing resources), and model training should typically be offloaded to the network, i.e., the model should be trained on the network side (e.g., network function or application function) and provided to the mobile terminal when needed. Therefore, an efficient method for providing a machine learning model to a mobile terminal of a mobile communication system is desired.

Summary of the Invention

[0003] A method is provided for providing a machine learning model to a mobile terminal of a mobile communication system, comprising the steps of: determining that a machine learning model to be used by the mobile terminal for a machine learning task should be delivered to the mobile terminal by the mobile wireless communication network of the communication system; and providing the machine learning model to the mobile terminal via user plane communication. [Brief explanation of the drawing]

[0004] In the drawings, similar reference numerals generally indicate the same parts across different drawings. The drawings are not necessarily to scale, and rather the emphasis is on illustrating the principles of the present invention. Various embodiments are described below with reference to the following drawings. [Figure 1] This shows a communication system according to an embodiment. [Figure 2] Flowchart 200 shows the procedure for acquiring a machine learning (ML) model (on the network side). [Figure 3] This flowchart illustrates an embodiment in which an ML model is provided from the UPF to the mobile terminal via a Packet Data Unit (PDU) session between the mobile terminal and the User Plane Function (UPF). [Figure 4] This flowchart illustrates an embodiment in which an ML model is provided from the NEF to the mobile terminal via a PDU session between the mobile terminal and the UPF, and a User Plane (UP) tunnel between the UPF and the Network Exposure Function (NEF). [Figure 5]This flowchart illustrates an embodiment in which an ML model is provided to the UPF from the Model Training (Network) Function (MTF) or Model Repository (Network) Function (MRF) via a user plane tunnel, and then provided to a mobile device via a PDU session. [Figure 6] This paper presents an application function (AF)-based approach for providing ML models to mobile devices. [Figure 7] Figure 6 shows a flowchart illustrating an example of the procedure for an AF-based method. [Figure 8] This is a flowchart illustrating a method for providing a machine learning model to a mobile terminal of a mobile communication system according to an embodiment. [Figure 9] This flowchart illustrates a method for obtaining a machine learning model from a mobile communication system according to an embodiment. [Modes for carrying out the invention]

[0005] The following detailed description refers to the accompanying drawings illustrating specific details and embodiments of the present disclosure in which the present invention may be carried out. Other embodiments may be utilized without departing from the scope of the present invention, and structural, logical, and electrical modifications may be made. Since some embodiments of the present disclosure may be combined with one or more other embodiments of the present disclosure to form new embodiments, the various embodiments of the present disclosure are not necessarily mutually exclusive.

[0006] Various examples corresponding to the aspects of this disclosure are described below.

[0007] Example 1 is a method for providing a machine learning model to a mobile terminal of a mobile communication system, the method comprising the steps of determining that a machine learning model to be used by the mobile terminal for a machine learning task should be delivered to the mobile terminal by the mobile wireless communication network of the communication system, and providing the machine learning model to the mobile terminal via user plane communication.

[0008] Example 2 is the method of Example 1, and includes the step of determining whether a machine learning model should be delivered to a mobile device based on one or more of the following: temporal and / or spatial changes in operating conditions, performance degradation of the machine learning model, configuration of the mobile device, and operator policies of the mobile communication network.

[0009] Example 3 is the method of Example 1, and includes the step of determining that a machine learning model should be delivered to a mobile terminal in response to receiving a request for a machine learning model from a mobile terminal via a mobile wireless communication network, and providing the machine learning model to the mobile terminal in response to the request.

[0010] Example 4 is the method of Example 3, and includes the step of receiving a request by the user plane function of the communication network.

[0011] Example 5 is the method of Example 4, and includes the steps of a user plane function discovering the memory location of the machine learning model, retrieving the machine learning model from the discovered memory location, and providing the machine learning model to the mobile device.

[0012] Example 6 is one of the methods from Examples 3 to 5 and includes the step of providing a machine learning model to a mobile device via a packet data unit session.

[0013] Example 7 is one of the methods in Examples 3-5 and includes the steps of receiving a request via a packet data unit session and providing a machine learning model to a mobile terminal via one or another packet data unit session.

[0014] Example 8 is one of the methods of Examples 3 to 7, and includes the step of receiving a request by a core network component of a wireless communication network via a user plane tunnel.

[0015] Example 9 is one of the methods of Examples 3 to 8, and includes the step of receiving a request in a user plane function of a wireless communication network and forwarding the request to a network exposure function of the wireless communication network via a user plane tunnel.

[0016] Example 10 is the method of Example 9, and the network exposure function includes the steps of discovering a storage location of a machine learning model, obtaining the machine learning model from the discovered storage location, and providing the machine learning model to a mobile terminal via a user plane function.

[0017] Example 11 is the method of Example 3, and includes the step of receiving a request in an application function of a wireless communication network.

[0018] Example 12 is the method of Example 11, and the application function includes the steps of discovering a storage location of a machine learning model, obtaining the machine learning model from the discovered storage location, and providing the machine learning model to a mobile terminal.

[0019] Example 13 is one of the methods of Examples 3 to 12, and includes the step of receiving a request via a quality of service flow and providing a machine learning model to a mobile terminal in response to the request via the quality of service flow or another quality of service flow. [[ID=|24]]

[0020] Example 14 is one of the methods of Examples 1 to 13, and includes the step of obtaining a machine learning model from a storage device storing the machine learning model via control plane communication or via a user plane tunnel.

[0021] Example 15 is one of the methods of Examples 1 to 14, and the machine learning model is used by a mobile terminal for a prediction task and / or for a compression and / or encoding and / or decompression and / or decoding task.

[0022] Example 16 is a communication network configured to execute one of the methods of Examples 1 to 15.

[0023] Example 17 is a method for obtaining a machine learning model from a mobile communication system, including the step of transmitting a request for a machine learning model used by a mobile terminal for a machine learning task from the mobile terminal to the mobile wireless communication network of the communication system, and the step of receiving the machine learning model in response to the request via user plane communication.

[0024] Example 18 is the method of Example 17, including the step of determining by the mobile terminal that the machine learning model should be requested based on one or more of a temporal and / or spatial change in operating conditions, a performance degradation of the machine learning model, a configuration of the mobile terminal, and an operator policy of the mobile communication network.

[0025] Example 19 is a mobile terminal configured to execute the method of Example 17 or 18.

[0026] It should be noted that one or more of the features of any of the above examples may be combined with any one of the other examples. In particular, the examples described in the context of a device are equally valid for methods.

[0027] According to a further embodiment, when executed by a computer, a computer program and a computer-readable medium including instructions for causing the computer to execute one of the methods of any of the above examples are provided.

[0028] Various examples will be described in more detail below.

[0029] Figure 1 shows a communication system according to an embodiment.

[0030] In this example, the communication system includes a radio access network (RAN) 101, a core network 103, and a transport (communication) network 102 connecting the RAN 101 to the core network 103. Subscriber terminals (referred to as UEs according to 3GPP) 104 are connected to distributed units 105 of the RAN 101, which are connected to a centralized unit 106 (for implementing base stations of the RAN 101). Application functions 107 are connected to the core network 103. The core network 103, for example, the 5G core network (5GC), includes various network functions (NF), such as the Access and Mobility Management Function (AMF) 108, the Session Management Function (SMF) 109, the Network Exposure Function (NEF) 110, the User Plane Function (UPF) 111, the Network Data Analytics Function (NWDAF) 112, and the Policy Control Function (PCF) 113.

[0031] The core network 103 is connected to the Operation, Administration and Maintenance (OAM) system 114.

[0032] In a communication system as shown in Figure 1, a machine learning (ML) model may be used to perform predictive tasks for various purposes, such as optimizing the performance of the wireless link between the subscriber terminal 104 and the RAN 101. ● Improved channel state information (CSI) feedback, e.g., reduced overhead, improved accuracy, and better prediction. ○ Spatial frequency domain CSI compression using a bilateral AI model ○Time-domain CSI prediction using UE-side model ● Beam management, e.g., reducing overhead and latency, and improving beam selection accuracy through beam prediction in the time and / or spatial domains. ○ Spatial domain downlink beam prediction of the first beam set A based on measurement results of the second beam set ○Time-based downlink beam prediction for the first beam set based on the historical measurement results of the second beam set ●Improved positioning accuracy for different scenarios, including those with severe non-line of sight (NLOS) conditions. ○Direct AI / ML positioning ○AI / ML-assisted positioning Furthermore, AI / ML technology, for example, ●Network energy saving, for example, ML models, is used as follows: ○Input: UE mobility / trajectory, current energy efficiency, UE measurement report, etc. ○Output: Predicted energy efficiency, handover strategy, etc. ●Load balancing, for example, is performed using ML models as follows: ○Input: UE trajectory, UE traffic, RAN resource status, etc. ○Output: Predicted self-resource status information, etc. ● Mobility optimization, for example, is performed using ML models as follows: ○Input: UE location information, wireless measurement, UE handover, etc. ○Output: Predicted handover, UE traffic prediction, etc. However, it can be used, for example, to improve RAN performance.

[0033] For one of the above objectives, the corresponding ML model may be trained by entities, for example, RAN entities, in particular, base stations (represented as gNBs in 5G), OAM114, components of the core network 103 (e.g., network functions), or third parties (e.g., corresponding to or connected to the core network 103 via AF107). This training entity is hereafter referred to as the Model Training NF (MTF). The ML model can be trained using training data collected from other NFs, UEs, OAMs, etc. The MTF can be an NWDAF112 that includes a Model Training Logical Function (MTLF) with extensions to support training models for RAN use cases (e.g., CSI prediction models) that provide models to NFs other than Analytical Logical Functions (AnLFs) such as AMF, UPF, and SMF. The Model Training Entity or MTF collects training data and trains its respective ML model. Next, the trained model (i.e., the specifications of the trained model regarding neural network weights, for example) is stored in a storage called the Model Repository NF (MRF), which can store the trained model and provide it to other components and entities (e.g., NFs). The MRF is an Analytics Data Repository Function (ADRF) that has the ability to provide stored models to NFs other than AnLF, such as AMF, UPF, and SMF.

[0034] The trained model may then be used at multiple inference locations, which may include, in particular, one or more mobile devices. Thus, according to various embodiments, a method is provided for distributing an ML model from the network side, e.g., RAN, 5GC, OAM, or a third party or AF, to the UE (via RAN).

[0035] According to various embodiments, model distribution is enabled by extending the interfaces in the UE and 5GC NF for model distribution in the user plane (UP). It should be noted that training locations inside and outside the core network 103 (e.g., 5GC) are supported. The ML model may be, in particular, a RAN model, i.e., an ML model trained for RAN tasks, i.e., to perform predictions related to RAN operation.

[0036] In the embodiments described below, it is assumed that there is a registration and discovery procedure in which an MTF registers supported models with the network repository function (NRF) along with the associated ML model filter information, and other NFs discover and select the MTF by querying the NRF for specific models and corresponding filters.

[0037] Figure 2 shows flowchart 200 illustrating the procedure for acquiring the ML model (on the network side).

[0038] Consumer NF201 (i.e., a network function (NF) that functions as a consumer and acquires trained ML models), NRF202, MTF203, and MRF204, such as AMF, SMF, UPF, or NEF, are included in the flow.

[0039] In step 205, MTF203 registers with NRF202 and specifically notifies NRF202 that MTF203 has the capability to train and provide ML models.

[0040] At 206, consumer NF201 sends a discovery request to NRF202, and at 207, NRF202 responds to NRF220 with a discovery request response informing consumer NF201 about MTF203.

[0041] In 208, consumer NF201 requests an ML model from MTF (by get or subscribe). The request may include various parameters about the requested model. In particular, if the ML model is used on the terminal (UE) side, the request may include information about the UE type, UE vendor, and UE location.

[0042] If the requested ML model is stored in MTF203, then in 209, MTF203 provides the requested model to consumer NF201.

[0043] If the requested ML model is not stored in MTF203 but is stored in MRF204, in 210, MTF203 notifies consumer NF201 of the identification information of MRF204, in 211, consumer NF201 requests the ML model from MRF, and in 212, MRF provides the requested ML model.

[0044] Both the terminal and the network should be capable of distributing ML models from the network to the terminal; that is, the terminal should be able to request and receive models from the network. The network should also be able to train models and distribute them to the terminal. The mobile terminal's capabilities and characteristics regarding ML models, for example, ●UE model provision capability (i.e., the UE has the ability to obtain AIML models from a network (e.g., a 5G communication system (5GS, 5G communication system))) ● List of supported ML models (e.g., ML models for CSI) ● User device information (e.g., UE type, vendor, hardware, etc.) You may notify the network about this.

[0045] A mobile device may provide information about its capabilities and characteristics, for example, as part of a UE capability transfer (i.e., along with a UECapabilityInformation message to the network in response to a UECapabilityEnquiry message from the network) or in its registration request (sent to the RAN to register with the network).

[0046] The network may advertise its network ML model provision capability, that is, it may notify the UE of its capability to provide ML models, for example, through system information in a system information block (SIB), such as through SystemInformation messages (e.g., in response to a SystemInformationRequest). The network may provide, for example, the following system information: ● Network model provision capability (i.e., the network has the capability to provide ML models to the UE) ● List of supported ML models (e.g., ML models for CSI)

[0047] The following describes an embodiment in which an ML model is provided to a mobile terminal from the network side via user plane communication.

[0048] Figure 3 shows a flowchart 300 illustrating an embodiment in which an ML model is provided from the UPF 302 to the mobile terminal 301 via a PDU session 303 between the mobile terminal 301 and the UPF 302.

[0049] UE301 and UPF302 each include their respective ML agent (MLA)304. The ML agent 304 is an entity similar to a Performance Measurement Functionality (PMF). The PMF may be used by a mobile terminal to obtain access performance measurements on the user plane. Similarly, the MLA304 enables the exchange of data, i.e., ML models, between the UE and UPF on the user plane.

[0050] Furthermore, NRF305 and MTF or MRF306 are included in the flow.

[0051] In 307, UE301 establishes PDU session 303 with UPF302.

[0052] In 308, the mobile terminal 301 requests an ML model from UPF302. The request is for ML model provision content in accordance with 3GPP, for example, ● List of analysis IDs: Identifies the analysis in which the ML model is used. ● Use case context: Indicates the context in which the analysis is used to select the most relevant ML model. ●ML Model Interoperability Information ●ML Model Target Period ●Level of precision of interest ● The time required for the model However, it includes good ML model information.

[0053] Furthermore, according to various embodiments, the ML model information is extended with an instruction that the requesting model should support the fact that the UE is the final model consumer (i.e., the ML model may be used on the terminal side).

[0054] In 309, UPF302 performs MTF discovery and selection, and then in 310, ML model acquisition is performed (as described, for example, with reference to Figure 2). ML model acquisition from the MTF or MRF306 (depending on where the model is stored) is performed via the control plane.

[0055] In 311, UPF302 provides the ML model requested by the mobile terminal 301 (i.e., the specifications of the trained model, for example, regarding the neural network weights).

[0056] Communication between the mobile terminal 301 and the UPF 302 in 308 and 311 is performed via the PDU session 302 and uses the MLA 304 to enable data exchange between the mobile terminal and the UPF in the user plane.

[0057] Figure 4 shows a flowchart 400 illustrating an embodiment in which an ML model is provided from the NEF 403 to the mobile terminal 401 via a PDU session 404 between the mobile terminal 401 and the UPF 402, and a UP tunnel 405 between the UPF 402 and the NEF 403.

[0058] UE401 may include an ML agent (MLA) 406 that enables communication with NEF405 via the user plane.

[0059] Furthermore, NRF407 and MTF or MRF408 are included in the flow.

[0060] At 409, UE301 establishes a PDU session 404 and an UP tunnel 405 with UPF302.

[0061] In 410, the mobile terminal 401 requests an ML model from the NEF 403. The request includes ML model information, which may be ML model provisioning content conforming to 3GPP, as explained with reference to Figure 3.

[0062] Furthermore, according to various embodiments, the ML model information is extended with an instruction that the requesting model should support the fact that the UE is the final model consumer (i.e., the ML model may be used on the terminal side).

[0063] At 411, NEF403 performs MTF discovery and selection, and then at 412, it performs ML model acquisition (as illustrated, for example, with reference to Figure 2). ML model discovery and selection, as well as ML model acquisition from the MTF or MRF406 (depending on where the model is stored), are performed via the control plane.

[0064] In 413, NEF403 provides the ML model requested by the mobile terminal 401 (i.e., the specifications of the trained model, for example, regarding the neural network weights).

[0065] Communication between the mobile terminal 401 and the NEF 402 in 410 may be based on a method that allows the UE to directly call the NEF API via the user plane tunnel 405 (for example, according to Resource owner-aware Northbound API Access (RNAA)).

[0066] Figure 5 shows a flowchart 500 illustrating an embodiment in which an ML model is provided from the MTF 508 or MRF 509 to the UPF 507 via a user plane tunnel 519, and from there to the mobile terminal 501 via a PDU session 520.

[0067] Furthermore, RAN502, AMF503, SMF504, PCF505, and NRF506 are included in the flow.

[0068] In step 510, the mobile terminal 501 sends a request to the AMF 503 (via RAN 502) to establish a PDU session 519. In response, in step 511, the AMF 503 requests the SMF 504 to create a session management context. Then, in step 512, the SMF 504 performs authentication and policy decisions for the PDU session 519 with the PCF 505.

[0069] In step 513, the SMF504 performs MTF discovery and selection with the NRF506 (as illustrated, for example, with reference to Figure 2).

[0070] Next, at 514, SMF504 sends an N4 session establishment / modification request to UPF507, which indicates MTF508 discovered at 515.

[0071] At 515, PDU session 519 is established (following the standard 3GPP procedure).

[0072] In 516, the UPF establishes a UP tunnel 520 with the MTF 508. This is done, for example, according to an extension of user-plane tunnel establishment on an N4 interface in accordance with 3GPP.

[0073] In step 517, the mobile terminal 501 requests an ML model from the MTF 508 via the PDU session 519 and the UP tunnel 520, and in step 518, the MTF 508 provides the requested ML model to the mobile terminal 501 via the PDU session 519 and the UP tunnel 520.

[0074] If the model is stored in MRF509 instead of MTF508, MRF509 will replace MTF508 in the above procedure.

[0075] Figure 6 shows an AF-based method for providing an ML model to the mobile terminal 601.

[0076] The flow includes ML management AF602, NRF603, model training AF604, custom ML AF605, NEF 606, external ML repository 607, MTF608, and MRF609 (or at least some of these, depending on where the ML models are trained and stored).

[0077] Mobile terminal 601 includes MLA610 and AnLF611. MLA610 is an application within mobile terminal 601 that communicates with (proprietary) ML AF 605. ML605 requests an ML model by calling the API of NEF606.

[0078] The ML management AF602 acts as a gateway for retrieving ML models from ML model providers (i.e., model training AF604, external repository 607, OAM or MTF608 or MRF609) and providing them to the mobile terminal 601. The model training AF604 can train models based on available data (sample datasets, simulations, etc.). The source of the ML model (e.g., MTF608) may be discovered using NRF603.

[0079] Figure 7 shows a flowchart 700 illustrating an example of the procedure for an AF-based method as shown in Figure 6.

[0080] The flow includes mobile terminal 701, RAN702, AMF703, SMF704, PCF705, UPF706, NEF707, ML management AF (MMF)708, ML training AF (MTAF)709, MTF410, MRF411, and external ML model repository 412 (or at least some of these, depending on where the ML models are trained and stored).

[0081] In 713, the mobile terminal 701 sends a PDU session change NAS message to the AMF 703 to initiate a QoS flow in the PDU session (which is assumed to have been previously established).

[0082] At 714, AMF 703 responds by sending a request to SMF704 to update the SM context for the PDU session.

[0083] Alternatively, at 715, the ML management AF708 may initiate the establishment of a QoS flow (in the PDU session) by sending a corresponding request to the NEF707. In this case, at 716, the NEF709 sends a policy / authorization creation message to the PCF705 to trigger the establishment of the QoS flow.

[0084] A QoS flow is established in 717.

[0085] In 718, the mobile terminal 701 requests an ML model from the ML management AF 708. The request includes ML model information, which may be ML model provisioning content conforming to 3GPP, as described above with reference to Figure 3. Furthermore, according to various embodiments, the ML model information is extended with an instruction that the requested model should support the fact that the UE is the final model consumer (i.e., that the ML model may be used on the terminal side).

[0086] In 719, the ML management AF708 identifies the storage location of the requested ML. Model.

[0087] The ML management AF708, according to the storage location of the identified and requested ML model, ●From model training AF709 in 720, ●Directly and indirectly from MTF710 or MRF711 in 721, or, ●From external repository 712 in 722 Retrieve the requested ML model.

[0088] In 723, the ML management AF708 provides the requested ML model to the mobile terminal 701.

[0089] It should be noted that all request messages (for requesting an ML model) and response messages (for providing an ML model) in the above example can be implemented as subscribe and notification message pairs. This enables network-triggered ML model delivery, where a mobile device subscribes to a specific model, and the network then provides a customized / updated version of the model, taking into account, for example, operator policies, ML model performance monitoring, mobile device mobility / handover, etc.

[0090] It should be noted that the various methods for providing ML models are not mutually exclusive, and multiple methods can be implemented and used depending on the operator's policies, mobile device configuration, etc.

[0091] In short, various embodiments provide methods such as those shown in Figures 8 and 9.

[0092] Figure 8 is a flowchart 800 illustrating a method for providing a machine learning model to a mobile terminal of a mobile communication system according to an embodiment.

[0093] In 801, the mobile wireless communication network of communication system a decides that a machine learning model to be used by a mobile terminal for a machine learning task (e.g., inference, e.g., prediction task, e.g., prediction task for controlling radio access network operation or data compression or encoding (e.g., as an autoencoder)) should be delivered to the mobile terminal. For example, this is decided upon receiving a request for a machine learning model from a mobile terminal. The request may also be a subscription (i.e., a subscribe message) to the corresponding model delivery service.

[0094] In 802, the mobile wireless communication network provides the machine learning model to the mobile terminal via user plane communication (in accordance with the decision that the machine learning model should be delivered to the mobile terminal).

[0095] According to various embodiments, a communication network is provided that is configured to perform the method described with reference to Figure 8.

[0096] The communication network corresponds, for example, to the RAN101, transport network 102, and / or core network 103 of the communication system in Figure 1.

[0097] Figure 9 is a flowchart 900 showing a method for obtaining a machine learning model from a mobile communication system according to an embodiment.

[0098] In 901, the mobile terminal transmits a request for a machine learning model to be used by the mobile terminal for an ML task to the mobile wireless communication network of the communication system.

[0099] In 902, the mobile terminal receives the requested machine learning model via user plane communication upon request.

[0100] The request may include an instruction that a machine learning model is required for a machine learning task on a mobile device.

[0101] According to various embodiments, a mobile device is provided that is configured to perform the method described with reference to Figure 9.

[0102] A mobile device corresponds, for example, to one of the UE104s in the communication system shown in Figure 1.

[0103] In other words, according to various embodiments, the ML model is provided to the mobile terminal via the user plane. This allows, for example, the use of the ML model on the terminal side, for example, the air interface, for communication optimization.

[0104] As described above, various modifications are possible to provide a flexible setting method.

[0105] Providing a machine learning model via user-plane communication can be understood as transmitting the specifications of the machine learning model, i.e., the values ​​of the trained parameters of the machine learning model (such as neural network weights if the machine learning model is a neural network), via user-plane communication from the network side (i.e., the mobile wireless communication network) to the mobile terminal (i.e., over the air interface), that is, using communication channels, interfaces and / or reference points belonging to the user plane (of the mobile communication system).

[0106] Communication networks and mobile terminals include and / or are implemented by one or more data processing components such as one or more processors, memory, interfaces, receivers, transmitters, antennas, etc., and may also be implemented by, for example, one or more circuits. A “circuit” may be understood as any kind of logic implementation entity, and may be a dedicated circuit or processor that runs software, firmware, or any combination thereof stored in memory. Thus, a “circuit” may be a wired logic circuit, or a programmable processor, such as a microprocessor. A “circuit” may also be a processor that runs software, such as any kind of computer program. Any other kind of implementation of any of the above functions may also be understood as a “circuit.”

[0107] While specific embodiments are described, it should be understood by those skilled in the art that various modifications in form and detail may be made therein without departing from the spirit and scope of the embodiments of this disclosure as defined by the attached claims. Accordingly, the scope is indicated by the attached claims, and is therefore intended to encompass all modifications that fall within the meaning and scope of the equivalents of the claims.

Claims

1. A method for providing machine learning models to mobile terminals in a mobile communication system, The steps include: In response to receiving a request for the machine learning model from the mobile terminal via the mobile wireless communication network of the communication system, which includes receiving a request in the user plane function of the wireless communication network, the mobile wireless communication network determines that the machine learning model to be used by the mobile terminal for a machine learning task should be delivered to the mobile terminal; The user plane function discovers the memory location of the machine learning model, retrieves the machine learning model from the discovered memory location, and provides the machine learning model to the mobile terminal, or, The request is forwarded to the network publishing function of the wireless communication network via a user plane tunnel, the network publishing function discovers the storage location of the machine learning model, retrieves the machine learning model from the discovered storage location, and provides the machine learning model to the mobile terminal via the user plane function. In response to the aforementioned request, the steps include providing the machine learning model to the mobile terminal via user plane communication and A method that includes this.

2. The method according to claim 1, further comprising the step of determining whether the machine learning model should be delivered to the mobile terminal based on one or more of the following: temporal and / or spatial changes in operating conditions, performance degradation of the machine learning model, configuration of the mobile terminal, and operator policies of the mobile communication network.

3. The method according to claim 1, further comprising the step of receiving the request by the user plane function of the communication network.

4. The method according to claim 1, comprising the step of providing the machine learning model to the mobile terminal via a packet data unit session.

5. The method according to claim 1, comprising the steps of receiving the request via a packet data unit session and providing the machine learning model to the mobile terminal via the packet data unit session or another packet data unit session.

6. The method according to claim 1, comprising the step of receiving the request by a core network component of the wireless communication network via a user plane tunnel.

7. A communication network configured to perform the method described in any one of claims 1 to 6.

Images