Method for facilitating model interoperability / interchangeability

Abstract

Various embodiments disclosed herein provide for a method to facilitate model interoperability between different network functions, such as an Application Function (AF) and a Network Data Analytics Function (NWDAF) or other functions in a Fifth Generation (5G) core network (5GC). The handshake process disclosed herein can enabled the network functions to determine if there is model interoperability and can facilitate the sharing of models, weights, and other elements from the AF into the 5GC or vice versa. The model data can be provided to a Model Training Logical Function (MTLF) or Analytics Logical Function (AnLF) being operated by an NWDAF for Federated Learning (FL) or transfer learning.

Description

METHOD FOR FACILITA TING MODEL INTEROPERABILITY / INTERCHANGEABILITYTechnical Field

[0001] The present disclosure relates to methods and network nodes that facilitate model interoperability, training, sending, and storing of models in a wireless communication system.Background

[0002] A topic of discussion in Release 19 (Rel-19) is how the Fifth Generation (5G) Core (5GC), and Network Data and Analytics Functions (NWDAFs), can generate the Analytics IDs that require input data from Application Functions (AFs) without having the AF providing raw data to the NWDAFs. This topic is described in further detail in the discussion on Artificial Intelligence (AI)ZMachine Learning (ML) for 5G System (5GS) Service to enable 5GC and Air interface Intelligence in S2-2306503 ("The Discussion on Artificial Intelligence (AI)Machine Learning (ML) for 5GS Service to enable 5GC and Air interface Intelligence”).

[0003] Additionally, another scenario is described in S2-2310875 ("The Discussion on 8 on FS_AIML_CN”) where it is stated that the AF can subscribe to analytics from the NWDAF via a Network Exposure Function (NEF). There are operators that would like to control or filter the analytics information provided to the AF, as such the analytics output will depend on the Service Level Agreement (SLA) between the Mobile Network Operator (MNO) and the Application Service Provider (ASP) and on the user content to expose its analytics information. For example, an AF that requests Analyticsld on "User Mobility” for a Generic Public Subscription Identifier (GPSI) may request that the User Mobility predictions are on the level of a Tracking Area Identity (TAI), a Cell ID or a geographical location. The document S2-2310875 proposes that the AF provides an initial ML model to the NWDAF to be trained, and that the ML model is trained until the AF considers that the ML Model can be used to produce an Analyticsld with a threshold level of detail. For example, in the case of the Analyticsld on "User Mobility” the AF may consider that the ML model is trained with the User.

[0004] It is believed that in Rel-19 storing models coming from an AF in 5GS will be discussed.Summary

[0005] Various embodiments disclosed herein provide for a method to facilitate model interoperability between different network functions, such as an Application Function (AF) and a Network Data Analytics Function (NWDAF) or other functions in a Fifth Generation (5G) core network (5GC). The handshake process disclosed herein can enable the network functions to determine if there is model interoperability and can facilitate the sharing of models, weights, and other elements from the AF into the 5GC and vice versa. The model data can be provided to a Model Training Logical Function (MTLF) or Analytics Logical Function (AnLF) being operated by an NWDAF.

[0006] In an embodiment, a method performed by a Network Repository Function (NRF) to facilitate model interoperability includes receiving first model interoperability information from one or more AFs, and second modelinteroperability information from one or more NWDAFs, and then based on the first model interoperability information and the second model interoperability information, identifying an NWDAF of the one or more NWDAFs that can handle an AF. The method can also include providing an NWDAF identifier of the NWDAF to the AF.

[0007] In an embodiment, the first model interoperability information is received via an NEF.

[0008] In an embodiment, the first model interoperability information and the second model interoperability information comprise one or more of a requested model file format; a model execution environment; label information; feature information; interlayer information; transfer learning information; and split learning information.

[0009] In an embodiment, the first model interoperability information and the second model interoperability information comprise interoperability identifiers or interoperability parameters.

[0010] In an embodiment, an NRF can be provided with processing circuitry configured to perform any of the methods of the embodiments described above.

[0011] In another embodiment, a method can be performed by an NRF for facilitating model interoperability with the method comprising receiving a subscription to information from a Network Exposure Function (NEF) and receiving a registration from an NWDAF, the registration comprising model information, and providing the model information to the NEF. In an embodiment, an NRF can be provided with processing circuitry configured to perform this method.

[0012] In another embodiment, a method is provided that is performed by an NEF for facilitating model interoperability, the method comprising subscribing to information from an NRF, receiving model information from the NRF associated with one or more NWDAFs, receiving model interoperability information from an AF, determining, based on the model interoperability information from the AF, and the model information associated with the one or more NWDAFs, which interoperability information to use for the AF, and providing the interoperability information to the AF.

[0013] In an embodiment, the model information from the one or more NWDAFs further comprises model interoperability information.

[0014] In an embodiment, in response to the model information not comprising model interoperability information, the method further comprises subscribing to information from the one or more NWDAFs.

[0015] In an embodiment, an NEF is provided with processing circuitry configured to perform the method of the above embodiments.

[0016] In an embodiment, a method performed by an NWDAF is provided for facilitating model interoperability. The method includes providing registration information to an NRF wherein the registration information comprises model information. The method also includes receiving model interoperability information from an AF, determining, based on the model information and the model interoperability information from the AF, which interoperability information to use for the AF, and providing the interoperability information to the AF.

[0017] In an embodiment, the NWDAF comprises an MTLF.

[0018] In an embodiment, the registration information comprises model interoperability information associated with the NWDAF.

[0019] In an embodiment, the model interoperability information from the AF is received via an NEF.

[0020] In another embodiment, a method performed by an NWDAF is provided for facilitating storing or training a model. The method includes receiving a model request from an AF, registering a model associated with the model request into an NRF, performing training using the model associated with the model request according to one or more parameters from the model request, and providing the AF with at least one of a model identifier associated with the model, or the model, or a reference to the model.

[0021] In embodiment, the method includes storing the model into an Analytics Data Repository Function (ADRF).

[0022] In an embodiment, the one or more parameters comprise a type of model, a request to store the model for future use, a request to train a model using an attachment in the model request.

[0023] In an embodiment, the NWDAF is operating an MTLF. In other embodiment, the NWDAF is operating an Analytics Logical Function (AnLF).

[0024] In an embodiment, an NWDAF is provided with processing circuitry configured to perform the method of the above embodiments.Brief Description of the Drawings

[0025] The accompanying drawing figures incorporated in and forming a part of this specification illustrate several aspects of the disclosure, and together with the description serve to explain the principles of the disclosure. Figure 1 shows a message sequence chart for a method for facilitating model interoperability according to one or more embodiments of the present disclosure;Figure 2 shows another message sequence chart for another method for facilitating model interoperability according to one or more embodiments of the present disclosure;Figure 3 shows another message sequence chart for another method for facilitating model interoperability according to one or more embodiments of the present disclosure;Figure 4 shows a message sequence chart for a method for facilitating performing model training and storing according to one or more embodiments of the present disclosure;Figure 5 shows another message sequence chart for another method for facilitating model combining and analytics according to one or more embodiments of the present disclosure;Figure 6 shows another message sequence chart for another method for facilitating model training according to one or more embodiments of the present disclosure;Figure 7 illustrates a wireless communication system represented as a Fifth Generation (5G) network architecture composed of core Network Functions (NFs);Figure 8 illustrates a 5G network architecture using service-based interfaces between the NFs in the Control Plane (CP);Figure 9 is a schematic block diagram of a network node according to some embodiments of the present disclosure;Figure 10 is a schematic block diagram that illustrates a virtualized embodiment of the network node according to some embodiments of the present disclosure; andFigure 11 Is a schematic block diagram of the network node according to some other embodiments of the present disclosure.Detailed Description

[0026] Some of the embodiments contemplated herein will now be described more fully with reference to the accompanying drawings. Other embodiments, however, are contained within the scope of the subject matter disclosed herein, the disclosed subject matter should not be construed as limited to only the embodiments set forth herein; rather, these embodiments are provided by way of example to convey the scope of the subject matter to those skilled in the art.

[0027] Core Network Node: As used herein, a "core network node” is any type of node in a core network or any node that implements a core network function. Some examples of a core network node include, e.g., a Mobility Management Entity (MME), a Packet Data Network Gateway (P-GW), a Service Capability Exposure Function (SCEF), a Home Subscriber Server (HSS), or the like. Some other examples of a core network node include a node implementing an Application Function (AF), a Network Exposure Function (NEF), a Network Function (NF) Repository Function (NRF), a Network Data Analytics Function (NWDAF), Model Training Logical Function (MTLF), Analytical Data Repository Function (ADRF), Analytics Logical Function (AnLF) or the like.

[0028] There currently exist certain challenge(s). Today there are no specified way to do a handshake to know if the Application Function (AF) and Fifth Generation (5G) System (5GS) can share models or perform Federated Learning (FL) or similar.

[0029] Additionally, there are no specified way to send model or weights or similar from an AF into the 5G core network (5GC). In the present disclosure, the model is sent either to a Model Training Logical Function (MTLF) or an Analytics Logical Function (AnLF).

[0030] In the present disclosure, the term "model” is used to apply to Machine Learning (ML) models, weights or similar for FL, transfer learning, etc.

[0031] Certain aspects of the present disclosure and their embodiments may provide solutions to the aforementioned or other challenges.

[0032] Various embodiments disclosed herein provide for a method to facilitate model interoperability between different network functions, such as an Application Function (AF) and a Network Data Analytics Function (NWDAF) or other functions in a Fifth Generation (5G) core network (5GC). The handshake process disclosed herein can enabled the network functions to determine if there is model interoperability and can facilitate the sharing of models, weights, and other elements from the AF into the 5GC or vice versa. The model data can be provided to a Model Training Logical Function (MTLF) or Analytics Logical Function (AnLF) being operated by an NWDAF for Federated Learning (FL) or transfer learning.

[0033] The principles below can be applied even inside the core domain, then the AF may be replaced with another NWDAF, and the functionality of NEF is part of this NWDAF. In this case the NWDAF may be an ordinary NF with learning capabilities.

[0034] Certain embodiments may provide one or more of the following technical advantage(s). Some of the advantages provided by the embodiment disclosed herein include: the possibility to do handshakes, the possibility for an AF to send models to 5GC, the possibility to train such model in NWDAF(MTLF), the possibility to merge two models in NWDAF, the possibility to store models from AF in ADRF for later use, sending the model instead of data from an AF, and also the possibility for an AF to send models to 5GC to an NWDAF(AnLF).Figure 1

[0035] Figure 1 illustrates a message sequence chart for a method for facilitating model interoperability according to one or more embodiments of the present disclosure

[0036] In Figure 1, a handshake can occur before the model is sent, where it is determined that the model from the AF is interoperable with the NWDAF and the model in the NWDAF before sharing models or perform FL, transfer learning, split learning or similar.

[0037] In an embodiment, a message can be sent between an AF 102, an NRF 104, and an NWDAF 106, and in some embodiments, there may be an NEF between the AF 102 and the NRF 104. The NWDAF 106 can include an MTLF and may be a special NWDAF(MTLF) that registers into NRF 104 the capability ToActOnModel.

[0038] ML Model Interoperability Information is specified in TS 23.288 as vendor-specific information that conveys, e.g., requested model file format, model execution environment, etc. When FL, transfer learning, split learning or similar is to be performed information on model details, such as labels, features, inter-layer information can be included.

[0039] The registered information into NRF may be Interoperability ID or other new parameter (instead of Interoperability Information). In this case the Interoperability Information is changed to either Interoperability ID or the new parameter in the embodiments. The registered information into NRF may be ML Model Interoperability Information, ML Model Interoperability identifier or other new parameter carrying the interoperability information

[0040] To make it possible to use Interoperability ID the definition of Interoperability ID can be broadened to not only include list of vendors. But also include info that matches the external "vendor” that is supported! Either Interoperability ID or new parameter is used, it shall be used in a way that indicates that the MTLF can handle models from this "vendor”. Example of "vendor” are developing models, phone maker, modem maker, User Equipment Operating System (UE OS) maker, or application maker. Today Interoperability ID is used in the opposite direction, namely it is used for a consumer to know if it is allowed to download and can execute a model from the MTLF.

[0041] At step 108, the method can include receiving model interoperability information from an NWDAF 106 of a group of NWDAFs (second model interoperability).

[0042] At 110, the method can include receiving first model interoperability information from one or more AFs 102.

[0043] At 112, the method can include the NRF 104 identifying an NWDAF 106 of the one or more NWDAFs that can handle an AF 102 of the AFs based on the first model interoperability information and the second model interoperability information.

[0044] And then at 114, the NRF 104 can send the identifier of the NWDAF 106 that is interoperable with the AF 102 to the AF 102.

[0045] Generally, the first model interoperability information and / or the second model interoperability information may comprise one or more of: a) a requested model file format; b) a model execution environment; c) label information; d) feature information; e) interlayer information; f) transfer learning information; g) split learning information; and h) other related information.In other words, in the embodiments described herein the first model interoperability information and / or the second model interoperability information may comprise any combination of the features a-h above, e.g. one (1), two (2), three (3), four (4), five (5), six (6), seven (7), and / or eight (8) of the features a-h above. One example may be that the first model interoperability information and / or the second model interoperability information comprises any one of feature a-h, or features a and b or features a, c, and g etc.Figure 2Figure 2 illustrates another message sequence chart for another method for facilitating model interoperability according to one or more embodiments of the present disclosure.Like in Figure 1, before sharing models it should be understood that the model from an AF is interoperable with the NWDAF. Like in Figure 1, the NWDAF may include an MTLF that may be a special MTLF that registers into NRF the capability ToActOn Model. The registered information into NRF may be Interoperability ID or a new parameter (instead of Interoperability Information). In the case of Interoperability ID, it is possible to broaden the definition in stage 3 to not only include a list of vendors but also include info that reveals what Hyperscale Cloud Providers (HCPs) that are supported, etc.At step 204, the NEF 202 may subscribe to information from the NRF 104.At step 206, the NWDAF 106 may register into NRF 104 the ToActOnModel (e.g., model information) and optionally its Interoperability Information. The model interoperability information may include the model interoperability information comprises one or more of: a requested model file format; a model execution environment; label information; feature information; interlayer information; transfer learning information; and split learning information.At 208, NEF 202 has info or receives info from previous steps on which NWDAFs has support for ToActOnModel and optional also direct from NRF 104 what Interoperability information they support.At 209, if model interoperability information was not received in the previous step 208, NEF 202 performs a subscription to each of these NWDAFs 106 to receive model interoperability information.At 210, the AF 102 can provide model interoperability information to the NEF 202. In an embodiment, the consumer starts Handshakes on Interoperability, e.g. on what format to send the model with NEF 202 including what type of Interoperability Information the AF 102 may supportAt 212, NEF 202 decides which Interoperability Information to use. In an embodiment the NEF 202 determines, based on the model interoperability information from the AF 102, and the model information associated with the one or more NWDAFs 106, which interoperability information to use for the AF 102.At step 214, the NEF responds with the model interoperability information to the AF 102 in a Handshake response.Figure 3

[0046] Figure 3 illustrates another message sequence chart for another method for facilitating model interoperability according to one or more embodiments of the present disclosure.

[0047] In Figure 3, the steps 204-208 are as described in Figure 2, but in step 302, in a Handshake request message, the AF 102 can provide top the NWDAF 106 directly the model interoperability information, and the NWDAF 106 at step 3045 can determine, based on the model information and the model interoperability information from the AF 102, which interoperability information to use for the AF 102. At step 306, the NWDAF 106 responds to the AF 102 with the selected model interoperability information.Figure 4

[0048] Figure 4 shows a message sequence chart for a method for facilitating performing model training and storing according to one or more embodiments of the present disclosure.

[0049] Figure 4 depicts the MTLF 402 as operated by NWDAF 106, and MTLF 402 may be a special MTLF that registers into NRF the capability ToActOnModel. The ToActOnModel service can be a request to StoreModel (for future use, similar to provisioning of info to 5GS) or to request to train a model and may be re-using existing Nnwdaf_MLModelTraining Service.

[0050] At step 404, the AF 102 can request ToActOnModel. In the request it attaches a model or a URL to fetch a model. The request can be to store the model for future use or can request to train a model making use of the attachment. It can also add criteria to the training (e.g., what type of model is wanted, e.g. a trained version of the model or a new model using the attached model + e.g. network performance model giving a combined model on, e.g., Observed Service Experience).

[0051] At step 406, the MTLF 402 registers into NRF the new model using Model ID such as App ID (or similar identification).

[0052] At step 408, optionally, the MTLF can store the model in Analytics Data Repository Function (ADRF) 516 and can receive a response at 410.

[0053] At 412, the MTLF 402 can perform training using the model associated with the model request according to one or more parameters from the model request.

[0054] At 414, the MTLF 402 response back to the AF 102 with the model identifier and / or attached model or reference to model.

[0055] In an embodiment, the model can directly be stored from the AF 102 into the ADRF 516. AF 102 or NEF 202 (if between AF 102 and ADRF 516) makes use of same services as MTLF 402 when storing into ADRF 516.Figure 5

[0056] Figure 5 illustrates another message sequence chart for another method for facilitating model combining and analytics according to one or more embodiments of the present disclosure.

[0057] In the embodiment in Figure 5, the Request ToActOnModel goes to the AnLF 502. The ToActOnModel service can be a request to perform Analytics (using the attached or referenced model). In an embodiment, there is an assumption that training or provisioning has been made beforehand, as described above. And the AnLF 502 performs inference by using only the model or one model for, e.g., network performance and adds the model for the specific model ID or App ID. Here the model can be discoverable (either in NRF 104 or in the special MTLF 402).

[0058] At step 504, the method includes providing a request from the AF 102 to the AnLF 502. The request can be to ToActOnModel (perform Analytics). In the request, the AF 102 can attach a model identifier or application identifier. The AF 102 could also add which Analytics ID it would like to receive as output. This output could be, e.g., Observed Service Experience. Or the output could just be the Analytics being derived by using the referred model.

[0059] At 506, the AnLF 502 discovers an MTLF 402 that can handle the model the AnLF 402 needs. It could add either model ID in the request or add App ID. Today model ID is not registered into NRF 104.

[0060] In step 508, the AnLF 502 requests a model from the MTLF 402 and at step 510, the MTLF 502 can response with the model in the model provisioning response. At step 512, the AnLF 502 may either perform Analytics using the model received from MTLF 402 or it may combine the model with another model (thought to already be available in the AnLF 502 or received earlier from the MTLF 402.

[0061] At step 514, the AnLF 502 can respond to the AF 102 with the model or analytics.

[0062] In an embodiment, the AF 102 can request ToActOnModel (perform Analytics). In the request it attaches a model or a URL to fetch a model and then step 512 can happen in lieu of steps 506-510.Figure 6

[0063] Figure 6 illustrates another message sequence chart for another method for facilitating model training according to one or more embodiments of the present disclosure.

[0064] In the embodiment in Figure 6, the AnLF 502 receives the model and provides the MTLDrequest the MTLF to train using this model in step 602, and at step 604, the MTLF 402 can perform the training using the model, and at step 606 the MTLF 402 can provide the trained model to the AnLF 502.Figure 7

[0065] Figure 7 illustrates a wireless communication system represented as a 5G network architecture composed of core Network Functions (NFs), where interaction between any two NFs is represented by a point-to- point reference point / interface.

[0066] Seen from the access side the 5G network architecture shown in Figure 7 comprises a plurality of UEs 718 connected to either a RAN 716 or an Access Network (AN) as well as an AMF 700. Typically, the R(AN) 716 comprises base stations, e.g. such as eNBs or gNBs or similar. Seen from the core network side, the 5GC NFs shown in Figure 7 include a NSSF 702, an AUSF 704, a UDM 706, the AMF 700, a SMF 708, a PCF 710, and an Application Function (AF) 712.

[0067] Reference point representations of the 5G network architecture are used to develop detailed call flows in the normative standardization. The N1 reference point is defined to carry signaling between the UE 718 and AMF 700. The reference points for connecting between the AN 716 and AMF 700 and between the AN 716 and UPF 714 are defined as N2 and N3, respectively. There is a reference point, N11, between the AMF 700 and SMF 708, which implies that the SMF 708 is at least partly controlled by the AMF 700. N4 is used by the SMF 708 and UPF 714 so that the UPF 714 can be set using the control signal generated by the SMF 708, and the UPF 714 can report its state to the SMF 708. N9 is the reference point for the connection between different UPFs 714, and N14 is the reference point connecting between different AMFs 700, respectively. N15 and N7 are defined since the PCF 710 applies policy to the AMF 700 and SMF 708, respectively. N12 is required for the AMF 700 to perform authentication of the UE 718. N8 and N10 are defined because the subscription data of the UE 718 is required for the AMF 700 and SMF 708.

[0068] The 5GC network aims at separating UP and CP. The UP carries user traffic while the CP carries signaling in the network. In Figure 7, the UPF 714 is in the UP and all other NFs, i.e., the AMF 700, SMF 708, PCF 710, AF 712, NSSF 702, AUSF 704, and UDM 706, are in the CP. Separating the UP and CP guarantees each plane resource to be scaled independently. It also allows UPFs to be deployed separately from CP functions in a distributed fashion. In this architecture, UPFs may be deployed very close to UEs to shorten the Round Trip Time (RTT) between UEs and data network for some applications requiring low latency.

[0069] The core 5G network architecture is composed of modularized functions. For example, the AMF 700 and SMF 708 are independent functions in the CP. Separated AMF 700 and SMF 708 allow independent evolution and scaling. Other CP functions like the PCF 710 and AUSF 704 can be separated as shown in Figure 7.Modularized function design enables the 5GC network to support various services flexibly.

[0070] Each NF interacts with another NF directly. It is possible to use intermediate functions to route messages from one NF to another NF. In the CP, a set of interactions between two NFs is defined as service so that its reuse is possible. This service enables support for modularity. The UP supports interactions such as forwarding operations between different UPFs.Figure 8

[0071] Figure 8 illustrates a 5G network architecture using service-based interfaces between the NFs in the CP, instead of the point-to-point reference poi nts / i nterfaces used in the 5G network architecture of Figure 7.However, the NFs described above with reference to Figure 7 correspond to the NFs shown in Figure 8. The service(s) etc. that a NF provides to other authorized NFs can be exposed to the authorized NFs through the servicebased interface. In Figure 8 the service based interfaces are indicated by the letter “N” followed by the name of the NF, e.g. Namf for the service based interface of the AMF 700 and Nsmf for the service based interface of the SMF 708, etc. The NEF 800 and the NRF 802 in Figure 8 are not shown in Figure 7 discussed above. However, it should be clarified that all NFs depicted in Figure 7 can interact with the NEF 800 and the NRF 802 of Figure 8 as necessary, though not explicitly indicated in Figure 7.

[0072] Some properties of the NFs shown in Figures 7 and 8 may be described in the following manner. The AMF 700 provides UE-based authentication, authorization, mobility management, etc. A UE 718 even using multiple access technologies is basically connected to a single AMF 700 because the AMF 700 is independent of the access technologies. The SMF 708 is responsible for session management and allocates Internet Protocol (IP) addresses to UEs. It also selects and controls the UPF 714 for data transfer. If a UE 718 has multiple sessions, different SMFs 708 may be allocated to each session to manage them individually and possibly provide different functionalities per session. The AF 712 provides information on the packet flow to the PCF 710 responsible for policy control in order to support QoS. Based on the information, the PCF 710 determines policies about mobility and session management to make the AMF 700 and SMF 708 operate properly. The AUSF 704 supports authentication function for UEs or similar and thus stores data for authentication of UEs or similar while the UDM 706 stores subscription data of the UE 718. The Data Network (DN), not part of the 5GC network, provides Internet access or operator services and similar.

[0073] An NF may be implemented either as a network element on a dedicated hardware, as a software instance running on a dedicated hardware, or as a virtualized function instantiated on an appropriate platform, e.g., a cloud infrastructure.Figure 9

[0074] Figure 9 is a schematic block diagram of a network node 900 according to some embodiments of the present disclosure. Optional features are represented by dashed boxes. The network node 900 may be, for example, a core network node implementing one of AF 102, NRF 104, NWDAF 106, NEF 202, MTLF 402, ANLF 502 or ADRF 516. As illustrated, the network node 900 includes a control system 902 that includes one or more processors 904 (e.g., Central Processing Units (CPUs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs), and / or the like), memory 906, and a network interface 908. The one or more processors 904 are also referred to herein as processing circuitry. The one or more processors 904 operate to provide one or more functions of the network node 900 as described herein (e.g., one or more functions of a base station 716 or gNB described herein). In some embodiments, the function(s) are implemented in software that is stored, e.g., in the memory 906 and executed by the one or more processors 904.Figure 10

[0075] Figure 10 is a schematic block diagram that illustrates a virtualized embodiment of the network node 900 according to some embodiments of the present disclosure. Again, optional features are represented by dashedboxes. As used herein, a "virtualized” network node is an implementation of the network node 900 in which at least a portion of the functionality of the network node 900 is implemented as a virtual component(s) (e.g., via a virtual machine(s) executing on a physical processing node(s) in a network(s)). The network node 900 includes one or more processing nodes 1000 coupled to or included as part of a network(s) 1002. If present, the control system 902 are connected to the processing node(s) 1000 via the network 1002. Each processing node 1000 includes one or more processors 1004 (e.g., CPUs, ASICs, FPGAs, and / or the like), memory 1006, and a network interface 1008.

[0076] In this example, functions 1010 of the network node 900 described herein (e.g., one or more functions of a base station 716 or gNB described herein) are implemented at the one or more processing nodes 1000 or distributed across the one or more processing nodes 1000 and the control system 902 in any desired manner. In some particular embodiments, some or all of the functions 1010 of the network node 900 described herein are implemented as virtual components executed by one or more virtual machines implemented in a virtual environment(s) hosted by the processing node(s) 1000. As will be appreciated by one of ordinary skill in the art, additional signaling or communication between the processing node(s) 1000 and the control system 902 is used in order to carry out at least some of the desired functions 1010.

[0077] In some embodiments, a computer program including instructions which, when executed by at least one processor, causes the at least one processor to carry out the functionality of the network node 900 or a node (e.g., a processing node 1000) implementing one or more of the functions 1010 of the network node 900 in a virtual environment according to any of the embodiments described herein is provided. In some embodiments, a carrier comprising the aforementioned computer program product is provided. The carrier is one of an electronic signal, an optical signal, a radio signal, or a computer readable storage medium (e.g., a non-transitory computer readable medium such as memory).Figure 11

[0078] Figure 11 is a schematic block diagram of the network node 900 according to some other embodiments of the present disclosure. The network node 900 includes one or more modules 1100, each of which is implemented in software. The module(s) 1100 provides the functionality of the network node 900 described herein. This discussion is equally applicable to the processing node 1000 of Figure 10 where the modules 1100 may be implemented at one of the processing nodes 1000 or distributed across multiple processing nodes 1000 and / or distributed across the processing node(s) 1000 and the control system 902.

[0079] Any appropriate steps, methods, features, functions, or benefits disclosed herein may be performed through one or more functional units or modules of one or more virtual apparatuses. Each virtual apparatus may comprise a number of these functional units. These functional units may be implemented via processing circuitry, which may include one or more microprocessor or microcontrollers, as well as other digital hardware, which may include Digital Signal Processor (DSPs), special-purpose digital logic, and the like. The processing circuitry may be configured to execute program code stored in memory, which may include one or several types of memory such as Read Only Memory (ROM), Random Access Memory (RAM), cache memory, flash memory devices, optical storage devices, etc. Program code stored in memory includes program instructions for executing one or moretelecommunications and / or data communications protocols as well as instructions for carrying out one or more of the techniques described herein. In some implementations, the processing circuitry may be used to cause the respective functional unit to perform corresponding functions according to one or more embodiments of the present disclosure.

[0080] While processes in the figures may show a particular order of operations performed by certain embodiments of the present disclosure, it should be understood that such order is exemplary (e.g., alternative embodiments may perform the operations in a different order, combine certain operations, overlap certain operations, etc.).EmbodimentsSome of the embodiments that have been described above may be summarized in the following manner:1. A method performed by a Network Repository Function, NRF, (104) for facilitating model interoperability, the method comprising: receiving (110, 108) first model interoperability information from one or more application functions, AFs, (102) and second model interoperability information from one or more Network Data and Analytics Functions, NWDAFs (106); based on the first model interoperability information and the second model interoperability information, identifying (112) an NWDAF (106) of the one or more NWDAFs that can handle an AF (102) of the AFs (102); and providing (114) an NWDAF identifier of the NWDAF (106) to the AF (102).2. The method of embodiment 1, wherein the first model interoperability information is received via a Network Exposure Function, NEF, (202).3. The method of any of embodiments 1 to 2, wherein the first model interoperability information and / or the second model interoperability information comprise one or more of: a requested model file format; a model execution environment; label information; feature information; interlayer information; transfer learning information; split learning information; and other related information.4. The method of any of embodiments 1 to 3, wherein the first model interoperability information and the second model interoperability information comprise interoperability identifiers or interoperability parameters.5. A Network Repository Function, NRF, (104) for facilitating model interoperability, the NRF comprising processing circuitry configured to perform any of the methods of embodiments 1 to 4.6. A method performed by a Network Repository Function, NRF, (104) for facilitating model interoperability, the method comprising: receiving (204) a subscription to information from a Network Exposure Function, NEF, (202);receiving (206) a registration from a Network Data and Analytics Functions, NWDAF, (106) the registration comprising model information; and providing (208) the model information to the NEF (202).7. The method of embodiment 6, wherein the registration further comprises model interoperability information associated with the NWDAF (106), and wherein the providing the model information to the NEF (202) further comprises providing the model interoperability information to the NEF (202).8. The method of any of embodiments 6 to 7, wherein the model interoperability information comprises one or more of: a requested model file format; a model execution environment; label information; feature information; interlayer information; transfer learning information; and split learning information.9. A Network Repository Function, NRF, (104) for facilitating model interoperability, the NRF comprising processing circuitry configured to perform any of the methods of embodiments 6 to 8.10. A method performed by a Network Exposure Function, NEF, (202) for facilitating model interoperability, the method comprising: subscribing (204) to information from a Network Repository Function, NRF, (104); receiving (208) model information from the NRF (104) associated with one or more Network Data and Analytics Functions, NWDAFs, (106); receiving (210) model interoperability information from an Application Function, AF, (102); determining (212), based on the model interoperability information from the AF, and the model information associated with the one or more NWDAFs (106), which model interoperability information to use for the AF (102); and providing (214) the model interoperability information to the AF (102).11. The method of embodiment 10, wherein the model information from the one or more NWDAFs (106) further comprises model interoperability information.12. The method of embodiment 10, wherein in response to the model information not comprising model interoperability information, the method further comprises: subscribing (209) to information from the one or more NWDAFs (106).13. A Network Exposure Function, NEF, (202) for facilitating model interoperability, the NEF (202) comprising processing circuitry configured to perform any of the methods of embodiments 10 to 12.14. A method performed by a Network Data and Analytics Functions, NWDAF, (106) for facilitating storing or training a model, the method comprising: providing (206) registration information to a Network Repository Function, NRF, (104) wherein the registration information comprises model information; receiving (302) model interoperability information from an Application Function, AF, (102); determining (304), based on the model information and the model interoperability information from the AF(102), which interoperability information to use for the AF (102) providing (306) the interoperability information to the AF (102).15. The method of embodiment 14, wherein the NWDAF (106) comprises a Model Training Logical Function, MTLF, (402).16. The method of any of embodiments 14 to 15, wherein the registration information comprises model interoperability information associated with the NWDAF (106).17. The method of any of embodiments 14 to 16, wherein the model interoperability information from the AF (102) is received via a Network Exposure Function, NEF, (202).18. A Network Data and Analytics Functions, NWDAF, (106) for facilitating model interoperability, the NWDAF (106) comprising processing circuitry configured to perform any of the methods of embodiments 14 to 17.19. A method performed by a Network Data and Analytics Functions, NWDAF, (106) for facilitating storing or training a model the method comprising: receiving (404) a model request from an Application Function, AF, (102); registering (406) a model associated with the model request into a Network Repository Function, NRF (104); performing (412) training using the model associated with the model request according to one or more parameters from the model request; and providing (414) the AF (102) with at least one of a model identifier associated with the model, or the model, or a reference to the model.20. The method of embodiment 19, further comprising: storing (408, 410) the model into an Analytics Data Repository Function, ADRF, (516).21 . The method of any of embodiments 19 to 20, wherein the one or more parameters comprise a type of model, a request to store the model for future use, a request to train a model using an attachment in the model request.22. The method of any of embodiments 19 to 21, wherein the NWDAF (106) is operating a Model Training Logical Function, MTLF, (402).23. The method of any of embodiments 19 to 21, wherein the NWDAF (106) is operating an Analytics Logical Function, AnLF, (502). 24. A Network Data and Analytics Functions, NWDAF, (106) for facilitating model interoperability, the NWDAF(106) comprising processing circuitry configured to perform any of the methods of embodiments 19 to 23.

Claims

ClaimsWhat is claimed is:

1. A method performed by a Network Repository Function, NRF, (104) for facilitating model interoperability, the method comprising: receiving (110, 108) first model interoperability information from one or more application functions, AFs, (102) and second model interoperability information from one or more Network Data and Analytics Functions, NWDAFs (106); based on the first model interoperability information and the second model interoperability information, identifying (112) an NWDAF (106) of the one or more NWDAFs that can handle an AF (102) of the AFs (102); and providing (114) an NWDAF identifier of the NWDAF (106) to the AF (102).

2. The method of claim 1 , wherein the first model interoperability information is received via a Network Exposure Function, NEF, (202).

3. The method of any one of claim 1 to 2, wherein the first model interoperability information and / or the second model interoperability information comprise one or more of: a requested model file format; a model execution environment; label information; feature information; interlayer information; transfer learning information; split learning information; and other related information.

4. The method of any one of claim 1 to 3, wherein the first model interoperability information and the second model interoperability information comprise interoperability identifiers or interoperability parameters.

5. A Network Repository Function, NRF, (104) for facilitating model interoperability, the NRF comprising processing circuitry configured to perform any one of the methods of claim 1 to 4.

6. A method performed by a Network Repository Function, NRF, (104) for facilitating model interoperability, the method comprising: receiving (204) a subscription to information from a Network Exposure Function, NEF, (202);receiving (206) a registration from a Network Data and Analytics Functions, NWDAF, (106) the registration comprising model information; and providing (208) the model information to the NEF (202).

7. The method of claim 6, wherein the registration further comprises model interoperability information associated with the NWDAF (106), and wherein the providing the model information to the NEF (202) further comprises providing the model interoperability information to the NEF (202).

8. The method of any one of claim 6 to 7, wherein the model interoperability information comprises one or more of: a requested model file format; a model execution environment; label information; feature information; interlayer information; transfer learning information; and split learning information.

9. A Network Repository Function, NRF, (104) for facilitating model interoperability, the NRF comprising processing circuitry configured to perform any one of the methods of 6 to 8.

10. A method performed by a Network Exposure Function, NEF, (202) for facilitating model interoperability, the method comprising: subscribing (204) to information from a Network Repository Function, NRF, (104); receiving (208) model information from the NRF (104) associated with one or more Network Data and Analytics Functions, NWDAFs, (106); receiving (210) model interoperability information from an Application Function, AF, (102); determining (212), based on the model interoperability information from the AF, and the model information associated with the one or more NWDAFs (106), which model interoperability information to use for the AF (102); and providing (214) the model interoperability information to the AF (102).

11. The method of claim 10, wherein the model information from the one or more NWDAFs (106) further comprises model interoperability information.

12. The method of claim 10, wherein in response to the model information not comprising model interoperability information, the method further comprises: subscribing (209) to information from the one or more NWDAFs (106).

13. A Network Exposure Function, NEF, (202) for facilitating model interoperability, the NEF (202) comprising processing circuitry configured to perform any one of the methods of claim 10 to 12.

14. A method performed by a Network Data and Analytics Functions, NWDAF, (106) for facilitating storing or training a model, the method comprising: providing (206) registration information to a Network Repository Function, NRF, (104) wherein the registration information comprises model information; receiving (302) model interoperability information from an Application Function, AF, (102); determining (304), based on the model information and the model interoperability information from the AF (102), which interoperability information to use for the AF (102) providing (306) the interoperability information to the AF (102).

15. The method of claim 14, wherein the NWDAF (106) comprises a Model Training Logical Function, MTLF, (402).

16. The method of any one of claim 14 to 15, wherein the registration information comprises model interoperability information associated with the NWDAF (106).

17. The method of any one of claim 14 to 16, wherein the model interoperability information from the AF (102) is received via a Network Exposure Function, NEF, (202).

18. A Network Data and Analytics Functions, NWDAF, (106) for facilitating model interoperability, the NWDAF (106) comprising processing circuitry configured to perform any one of the methods of claim 14 to 17.

19. A method performed by a Network Data and Analytics Functions, NWDAF, (106) for facilitating storing or training a model the method comprising: receiving (404) a model request from an Application Function, AF, (102); registering (406) a model associated with the model request into a Network Repository Function, NRF (104); performing (412) training using the model associated with the model request according to one or more parameters from the model request; and providing (414) the AF (102) with at least one of a model identifier associated with the model, or the model, or a reference to the model.

20. The method of claim 19, further comprising: storing (408, 410) the model into an Analytics Data Repository Function, ADRF, (516).21 . The method of any one of claim 19 to 20, wherein the one or more parameters comprise a type of model, a request to store the model for future use, a request to train a model using an attachment in the model request.

22. The method of any one of claim 19 to 21, wherein the NWDAF (106) is operating a Model Training Logical Function, MTLF, (402).

23. The method of any one of claim 19 to 21, wherein the NWDAF (106) is operating an Analytics Logical Function, AnLF, (502).

24. A Network Data and Analytics Functions, NWDAF, (106) for facilitating model interoperability, the NWDAF(106) comprising processing circuitry configured to perform any one of the methods of claim 19 to 23.

Images