Authorization and authentication of machine learning model migration

Through end-to-end encryption and authorization token verification, combined with security anchoring function, the problem of security threats in the sharing process of AI/ML model in 5G systems is solved, and efficient security sharing and protection of AI/ML models is achieved.

CN118975193APending Publication Date: 2024-11-15INTEL CORP
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202380032721.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Priority Date
2022-06-10
Filing Date
2023-05-01
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

In fifth generation (5G) systems, AI/ML models have security threats in the sharing process between network data analysis functions (NWDAF), especially due to the insufficient granularity of the current authorization scheme, resulting in unauthorized network function consumers It is possible to access and utilize sensitive AI/ML models, resulting in data security risks.

Method used

End-to-end encryption approach and authorization token verification ensures that only authorized network functionality consumers can access the AI/ML model, while introducing security anchoring capabilities to enable secure sharing and protection of models in a multi-vendor environment.

Benefits of technology

It effectively prevents unauthorized access and use, ensures the security of the AI/ML model, and provides a higher level of data protection, especially in model sharing scenarios in a multi-vendor environment.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118975193A_ABST
    Figure CN118975193A_ABST
Patent Text Reader

Abstract

The present disclosure relates to architecture enhanced security aspects for interaction between a network data analysis function (NWDAF) and other network functions in a fifth generation core network (5GC). Secure access mechanisms for machine learning (ML) models for authorized ML model service consumers are described, including mechanisms for secure storage by ML model service providers and acquisition by authorized ML model service consumers. Mechanisms are also provided to enable secure sharing of ML model NWDAFs provided by the same or different vendors.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to U.S. Provisional Application No. 63 / 337,376 filed on May 2, 2022 and U.S. Provisional Application No. 63 / 351,206 filed on June 10, 2022, the contents of each of which are incorporated herein by reference in their entirety. Technical Field

[0003] The present disclosure generally relates to wireless communication technology, network topology, artificial intelligence (AI) and machine learning (ML), network and information security technology, and more specifically, to security aspects of potential architectural enhancements related to the interaction between a network data analysis function (NWDAF) and other network functions in a fifth-generation core network. Background Art

[0004] In a fifth generation (5G) system, AI / ML models are shared between network data analysis functions (NWDAFs) (e.g., NWDAF to NWDAF) and / or with other network functions (NFs) (e.g., analysis data repository function (ADRF) to NWDAF, application function (AF) to NWDAF via network exposure function (NEF), etc.). In different scenarios, the NF producer (NFp) of the AI / ML model may store the AI / ML model in the ADRF, NWDAF, and / or other entities. The ADRF stores the AI / ML models to facilitate the distribution and sharing of these AI / ML models between NFs. Since AI / ML models and their algorithms are typically private (e.g., subject to the intellectual property rights of the designer, developer, or supplier), it may be desirable to ensure that only NFs that have been authenticated for access to the AI / ML models can read and use these models. In addition, the ADRF may not be considered a fully trusted entity for storing sensitive AI / ML data models because these models may be open at rest in the ADRF.

[0005] The current / existing authorization scheme defined by the 3GPP Service-Based Architecture (SBA) may only work at the service level or resource / operation level scope. This authorization granularity may be insufficient for AI / ML model sharing because the ADRF, NWDAF and / or any other NF that may store the AI / ML model cannot verify whether the NF consumer (NFc) is authorized to obtain a specific AI / ML model. This may lead to various security threats, for example, an unauthorized NFc (e.g., an NFc that is not eligible to obtain a specific model stored by the NFp) may have access to the storage entity and obtain the model. In addition, if there are no protection measures for accessing and reading the AI / ML model stored by the NFp from the ADRF, the unresistant ADRF may open the algorithm and sensitive data to unauthorized entities, which can easily abuse it and / or distribute it further to other entities, resulting in additional and potentially more serious data security breaches. BRIEF DESCRIPTION OF THE DRAWINGS

[0006] The embodiments will be easily understood by the following detailed description in conjunction with the accompanying drawings. For the convenience of this description, the same reference numerals represent the same structural elements. The embodiments are shown in the figures of the accompanying drawings by way of example and not by way of limitation, wherein:

[0007] Figure 1 depicts an example data collection architecture using data collection coordination; Figure 2 Depicts an example network data analysis open architecture using data collection coordination; Figure 3 depicts an example data storage architecture for analysis and collected data; Figure 4 depicts an example trained ML model provisioning architecture; Figure 5 and Figure 6 Describes an example process for secure machine learning (ML) model migration; Figure 7 depicts an example process for registering a new logical function called a vendor repository function (VRF) with a network repository function (NRF); Figure 8 Describes an example process for registering an NWDAF containing a Model Training Logic Function (MTLF) with an NRF; Fig. 9 Depicts examples of interoperability support for parameters identified by analysis; Fig.10 and Fig.11 depicts an example wireless network; Fig.12 depicts example hardware resources; and Fig.13 , Fig.14 , Fig.15 , Fig.16 and 17 Depicted are example procedures for practicing the various embodiments discussed herein. DETAILED DESCRIPTION

[0008] 1. Network data analysis function (NWDAF)

[0009] 5G System (5GS) (see, e.g., discussed below) Fig.10 ) may include NWDAF (e.g., Fig.10 NWDAF1062 in, which is capable of receiving data from a user equipment (UE) (e.g., Fig.10 UE 1002 in the UE), other NFs, operations, administration and maintenance (OAM) entities, application functions (AF) (e.g., Fig.10 AF 1060 in), data networks (e.g. Fig.10 DN 1036 in), cloud computing services, edge computing nodes and / or edge networks and / or other entities / elements that can be used for analysis to collect data. Network functions (NFs).

[0010] The 5GS architecture allows the NWDAF 1062 to access the NF from any NF (e.g. Fig.10 The 5GS architecture also allows the NWDAF 1062 to obtain management data from the OAM entity by calling OAM services.

[0011] The 5GS architecture also allows NWDAF 1062 to use DCCF 1063 with associated Ndccf services (see e.g. Figure 1 ), collect data from any NF or OAM (see, e.g., clause 8.2 of [TS23288]). The 5GS architecture also allows the NWDAF 1062 and DCCF 1063 to collect data from the NWDAF 1062 with an associated Nnwdaf_DataManagement service (see, e.g., Figure 2 and clause 7.4 of [TS23288]). The 5GS architecture allows the MFAF 1065 to ingest data from the NWDAF 1062 with an associated Nnwdaf_DataManagement service (see e.g. Figure 2and clause 7.4 of [TS23288]). The Nnwdaf_AnalyticsSubscription service enables a NF service consumer (NFc) to subscribe / unsubscribe to different types of analytics from the NWDAF 1062 (see, e.g., clause 7.2 of [TS23288]). The Nnwdaf_AnalyticsInfo service enables a NFc to request and get different types of analytics information from a NWDAF 1062 and / or enables a NWDAF 1062 to request migration of analytics context from another NWDAF 1062 (see, e.g., clause 7.3 of [TS23288]).

[0012] Figure 1 An example data collection architecture using data collection coordination is depicted. The data collection architecture includes: NWDAF 1062; Data Collection Coordination Function (DCCF) 1063; Messaging Framework 1064, which includes Messaging Framework Adapter Function (MFAF) 1065; and Network Node / NF 150, which can be or include NRF (e.g., Fig.10 NRF 1052), UDM (e.g. Fig.10 The UDM 1058 of [TS23288] and / or the Binding Support Function (BSF). Various DCCF services are discussed in clause 8 of [TS23288], and various MFAF services are discussed in clause 9 of [TS23288].

[0013] The NWDAF 1062 is communicatively coupled to the MFAF 1065 via an Nnf interface and to the DCCF 1063 via an Ndccf interface. The NDCCF interface is defined for the NWDAF 1062 to support subscription requests from the DCCF 1063 for data delivery, unsubscribing from data delivery, and requesting specific reports on data. If the data has not yet been collected, the DCCF 1063 requests the data from the data source (e.g., any NF) using the Nnf service (e.g., via the Nnf interface). The DCCF 1063 may collect the data and transmit it to the NWDAF 1062 (e.g., via the Ndccf interface), or the DCCF 1063 may rely on the messaging framework 1064 to collect the data from the NF and transmit it to the NWDAF 1062 (e.g., via the Nnf interface). The DCCF 1063 is communicatively coupled to the MFAF 1065 via an Nmfaf interface.

[0014] Figure 2 Depicts an example network data analysis open architecture using data collection coordination, which includes Figure 1The 5GS architecture allows any NF to include an analysis logic function (AnLF) 1062a (see e.g. Figure 4 ) requests network analysis information from the NWDAF of the DCCF 1062. The NWDAF 1062 belongs to the same PLMN as the NF consuming the analysis information. The Ndccf interface is defined for any NF to support subscription requests for network analysis (e.g., NWDAF 1062), cancelling subscriptions for network analysis, and requesting specific reports for network analysis. If the analysis has not yet been collected, the DCCF 1063 requests the analysis from the NWDAF 1062 using the Nnwdaf service. The DCCF 1063 can collect the analysis and transmit it to the NF, or the DCCF 1063 can rely on the messaging framework 1064 to collect the analysis and transmit it to the NF (e.g., via the Nnf interface).

[0015] Figure 3 An example data storage architecture for analytics and collected data is depicted, including NF, DCCF 1063, MFAF 1065 in messaging framework 1064, and analytical data repository function (ADRF) 1066. The 5GS architecture allows ADRF 1066 to store and retrieve collected data and analytics in one or more databases 1067 (which can implement any suitable database management system).

[0016] ADRF 1066 opens Nadrf services (e.g., via Nadrf interfaces) for other NFs (e.g., NWDAF 1062 and / or any other NF (e.g., any NF discussed herein)) that use Nadrf services to access data to store and retrieve data. For example, data may be stored in ADRF 1066 by a consumer sending a Nadrf_DataManagement_StorageRequest containing data or analysis to be stored to ADRF 1066. In some examples, the Nadrf_DataManagement_StorageRequest sent by the service consumer may include data to be stored, data collection timestamps, analysis with timestamps, service operations, analysis specifications or data specifications, storage processing information, DataSetTag, and / or other suitable information. The ADRF responds with a result indication to provide and / or send a Nadrf_DataManagement_StorageRequest Response message to the consumer. As examples, the response may include an indication of: storing the data and / or analysis; whether ADRF 1066 determined that the data or analysis has been stored; the storage method; and / or other suitable information (eg, any of the information discussed herein).

[0017] The consumer sends a Nadrf_DataManagement_Retrieval Request to the ADRF 1066 to retrieve data or analytics for the storage transaction identifier or ingestion instruction received in a Nadrf_DataManagement_RetrievalNotify from the ADRF 1066. The ADRF 1066 determines the availability of the data or analytics in its repository and sends the data or analytics in response to the consumer.

[0018] The ML model can be stored in the ADRF 1066 by the consumer sending a Nadrf_MLModelManagement_StorageRequest containing the ML model or the address of the ML model to be stored to the ADRF 1066. The ADRF response provides an indication of the result. The ML model can be deleted from the ADRF 1066 by the consumer sending a Nadrf_MLModelManagement_Delete request. The ADRF response provides an indication of the result.

[0019] Based on the NF request or configuration on the DCCF 1063, the DCCF 1063 can determine or identify the ADRF 1066 and interact with the ADRF 1066 directly or indirectly to request or store data. Direct interaction involves: (for example, when ADRF requests a data collection notification via DCCF1063) the DCCF 1063 requests to store data in the ADRF 1066 via the Nadrf service or via Ndccf_DataManagement_Notify. In addition, the DCCF 1063 obtains data from the ADRF 1066 via the Nadrf service. Indirect interaction involves: the DCCF 1063 requests the message transmission framework 1064 service to store data in the ADRF 1066 via the Nadrf service or via Nmfaf_3daDataManagement_Configure. The message transmission framework 1064 may include one or more adapters (for example, MFAF 1065 and / or some other adapters) that convert between 3GPP defined protocols. The NFc may indicate in a request to the DCCF 1063 that the data provided by the data source needs to be stored in the ADRF 1066. The ADRF 1066 stores the data received in a Nadrf_DataManagement_StorageRequest sent directly from the NF, or in a Ndccf_DataManagement_Notify, Nmfaf_3caDataManagement_Notify or Nnwdaf_DataManagement_Notify from the DCCF 1063, the MFAF 1065 and / or from the NWDAF 1062. The ADRF 1066 checks whether the data consumer is authorized to access the ADRF service and provides the requested data using the procedure specified in clause 7.1.4 of [TS23501].

[0020] Figure 4An example trained ML model provisioning architecture is depicted. The NWDAF 1062 may include an analysis logic function (AnLF) 1062a and / or a model training logic function (MTLF) 1062b. The NWDAF 1062 may include only the MTLF 1062b, only the AnLF 1062a, or both logic functions 1062a, 1062b. The 5GS architecture allows a NWDAF including an AnLF 1062a (also referred to herein as "NWDAF-ANLF 1062a") to use trained ML model provisioning services from the same or different NWDAF including a MTLF 1062b (also referred to herein as "NWDAF-MTLF 1062b"). The NWDAF-AnLF 1062a uses the Nnwdaf interface to request and subscribe to trained ML model provisioning services provided by the NWDAF-MTLF 1062b. The NWDAF 1062 provides the Nnwdaf_MLModelProvision service, which enables the NFc to receive notifications when ML models matching the subscription parameters become available in the NWDAF-MTLF 1062b (see, e.g., clause 7.5 of [TS23288]). The NWDAF 1062 provides the Nnwdaf_MLModelInfo service, which enables the NFc to request and get ML model information from the NWDAF-MTLF 1062b (see, e.g., clause 7.6 of [TS23288])

[0021] AnLF 1062a is a logical function in NWDAF 1062 that performs reasoning, derives analytical information (e.g., based on analytical consumer requests, derives statistics, reasoning, and / or predictions), and exposes analytical services (e.g., Nnwdaf_AnalyticsSubscription or Nnwdaf_AnalyticsInfo). The analytical information is either statistical information of past events or predictive information. MTLF 1062b is a logical function in NWDAF 1062 that trains AI / ML models and exposes new training services (e.g., provides trained ML models), as defined in clauses 7.5 and 7.6 of [TS23288].

[0022] Since multiple NWDAF 1062 instances may be deployed in the network, NFc may utilize NRF 1054 to discover NWDAF 1062 instances unless NWDAF information is available through other means (e.g., locally configured on NFc). NFc may make additional queries to UDM 1058 (when supported). The NWDAF selection function in NFc selects an NWDAF 1062 instance based on the available NWDAF 1062 instances.

[0023] To support NF discovery and selection of NWDAF 1062 instances containing MTLF 1062b, NWDAF 1062 instances containing AnLF 1062a, or both that can provide the required services (e.g., analysis opening and / or ML model provisioning) for the required type of analysis, each NWDAF 1062 instance may provide a list of supported analysis IDs (e.g., possibly per supported service) when registering with NRF 1054, in addition to other NRF 1054 registration elements of the NF profile. NFs that need to discover NWDAF 1062 instances that provide support for some specific services regarding a specific type of analysis may query NRF 1054 for NWDAF 1062 that supports the required services and required analysis IDs. Consumers (e.g., NFs, AF 1060, and OAM) decide how to use the data analysis provided by NWDAF 1062. The interaction between NFs and NWDAF 1062 occurs within the PLMN.

[0024] The NRF 1054 may return one or more candidate NWDAF 1062 instances, and each candidate NWDAF 1062 instance (based on its registered profile) supports an analysis ID with a time less than or equal to the supported analysis delay.

[0025] For NWDAF 1062 selection, NFc may consider the following factors: S-NSSAI; analysis ID; supported services, possibly with their associated analysis IDs; NWDAF service area information (e.g., a list of TAIs for which NWDAF 1062 can provide analytics, trained ML models and / or data, and / or other NWDAF services); NF type of the data source when DCCF 1063 is hosted by NWDAF 1062; NF set ID of the data source; supported analysis latencies for the requested analysis ID (see clause 6.2.6.2 of [TS23501]); and / or for multiple deployed NWDAF 1062 instances, consider NWDAF capabilities (e.g., analysis aggregation capability, analysis metadata provisioning capability, ML model training capability, ML model deployment capability, etc.). When selecting the NWDAF 1062 for ML model provisioning, the NWDAF 1062 may consider the following additional factors: ML model filtering information parameters (e.g., S-NSSAI and Area of ​​Interest (AoI) for trained ML models by analysis ID and ML model interoperability indicator by analysis ID (if available)) (see, e.g., clause 5.2 of [TS23288]). When selecting the NWDAF 1062 that supports federated learning (FL), the NWDAF 1062 may consider the following additional factors: Time period of interest (e.g., time interval [start...end] for which FL will be performed); when selecting the FL client, consider: FL capability type as FL client by analysis ID and / or data available to the FL client; and when selecting the FL server, consider: FL capability type as FL server by analysis ID and / or ML model filtering information parameters S-NSSAI and AoI for trained ML models by analysis ID (if available) (see, e.g., clause 5.2 of [TS23288]).

[0026] 1.1. ML model exchange and sharing via ADRF

[0027] Based on the security threats described previously, the following solutions provide protection for AI / ML models between entities that produce ML models or store ML models in ADRF 1066 (e.g., NWDAF-MTLF 1062b, NFp, etc.) and entities that consume models (NFc), and ADRF 1066 authorizes NFc to obtain the model. NFc is authorized to access the model in ADRF 1066 (or any other NF that can store ML models (e.g., NWDAF-MTLF 1062b)). The solutions discussed below protect AI / ML models between entities that produce ML models and / or store ML models in ADRF 1066 and entities that consume models (e.g., NFc). In these solutions, ADRF 1066 uses authorization tokens to verify that NFc is allowed to access the ML model.

[0028] These solutions address the issues related to ADRF 1066 not being a fully trusted entity and provide an end-to-end (e2e) encryption approach where NFc and ML model owners are able to access the ML model in plain text. A third-party public cloud will not have access to the ML model and its sole purpose is to store the model / file. The e2e encryption prevents leakage of confidential and / or sensitive data and ML models in both single vendor and multi-vendor environments (e.g., cloning, duplication, or internal ML model details by heuristics). In addition, when ML models are not encrypted, they can be cloned and these cloned ML models may result in loss of turnover or advantage over competitors and / or data privacy / security breaches. The e2e encryption provides security guarantees that cannot be provided by transmission security or access level control to the database alone. These solutions also assume that ML model protection is vendor specific. Use existing procedures in [TS23288] (e.g., using Nadrf_MLModel_StorageRequest containing the ML model) for ML model storage.

[0029] 1.1.1. Safety First of All in ML Model Migration

[0030] Figure 5 Depicting a first example process for secure ML model migration, which may operate as follows:

[0031] 1. The MTLF 1062b trains one or more ML models and provides the ML models to the ADRF 1066 by calling the Nadrf_DataManagement_StorageRequest (ML Model) service operation. The Nadrf_DataManagement_StorageRequest may include the ML model and / or an address / reference to a location where the ML model can be obtained. Along with each ML model (or each address / reference), data and / or metadata for each model is also included in the storage request. Examples of such data / metadata include an ML model ID per ML model, an analysis ID per ML model, a vendor ID per ML model, a digital certificate or signature per ML model (e.g., a binary MAC or SHA256 signature of an application), an execution environment / requirements per ML model, an address / reference (e.g., URL, link, etc.) for obtaining configuration and secrets and / or signing keys, certificates and / or other information for generating authentication credentials per ML model (see, e.g., Section 1.3.3 below and / or [TS23288]).

[0032] 2. ADRF 1066 stores the ML model and responds according to [TS23288] (except that ADRF 1066 stores the ML model). If the ML model address is included in the Nadrf_DataManagement_StorageRequest, ADRF 1066 downloads the ML model based on the ML model address and stores the ML model locally. ADRF 1066 sends a Nadrf_DataManagement_StorageRequest Response message to the consumer (e.g., MTLF 1062b) indicating that the ML model has been stored. The ML model and / or data may be stored in ADRF 1066 according to clause 5B.1 of [TS23288].

[0033] 3. NFc (e.g., NWDAF-ANLF 1062a) contacts NRF 1054 and requests an access token using existing procedures in 3GPP TS 33.501v18.1.0 (2023-03-30) ("[TS33501]"), [TS23288], and / or [TS23502]. The access token can be used for authorized access to ADRF 1066 and MTLF 1062b. NRF 1054 can send two tokens, one for MTLF 1062b and the other for ADRF 1066.

[0034] 4. The NRF 1054 sends an access token along with the MTLFID using existing procedures per [TS23288], [TS33501], and / or [TS23502]. Upon receiving the access token, the NFc (e.g., NWDAF-ANLF 1062a) generates a unique key pair (e.g., a public-private key pair) and creates a token-based certificate chain for the public key. If the NRF 1054 sends two tokens, one for the MTLF 1062b and the other for the ADRF 1066, the ADRF-based token is used to become the certificate chain previously described. In some examples, the certificate chain can be based on the trust roots installed in both NFs.

[0035] 5. NFc uses Nnwdaf_MLModelProvision service operation for ANLF 1062a to receive ML model ID based on analysis ID and ADRF-id to obtain ML model. This service operation may include sending access token to MTLF 1062b.

[0036] 6. MTLF 1062b verifies the access token received in step 3. MTLF 1062b points to the ML model address stored in ADRF 1066.

[0037] 7. The consumer of the ML model (e.g., NWDAF-ANLF 1062a) uses the ADRF service process to request the ML model. It also sends the unique public key generated by the NF in step 4 to the consumer.

[0038] 8. The ADRF 1066 verifies the unique public key (as indicated in step 7) from the token-based certificate chain. If the verification is successful, the ADRF 1066 generates a symmetric encryption key (e.g., a 256-bit AES key, etc.) to encrypt the stored ML model and sends it to the NFc. The ADRF 1066 also packages the symmetric encryption key with the encrypted model. Additionally or alternatively, the ADRF 1066 may encrypt the ML model using the public key. The consumer of the ML model (e.g., NWDAF-ANLF 1062a) decrypts the ML model using the symmetric key.

[0039] 1.1.2.Second security of ML model migration

[0040] Figure 6 A second example process for secure ML model migration is depicted, which may operate as follows:

[0041] 1. This step is similar to the previously described Figure 5The same as step 1 of . In addition, NWDAF-MTLF 1062b generates a security context using a logical function or a named network function NKGC for protecting the ML model information. MTLF 1062b may send the ML model encrypted using a symmetric key (e.g., an AES key, etc.) before storage. The security context includes, for example, an encryption key K enc , integrity key K int and corresponding security algorithms for encryption and integrity protection. NWDAF-MTLF 1062b uses the encryption key K enc and the integrity key K int To protect the ML model and related information. NKGC stores the security context. In some examples, the security algorithm in the security context can be used to generate the encryption key K enc and the integrity key K int , for decrypting and verifying the ML model / ML model information stored in ADRF 1066.

[0042] 2. This step is similar to the previously described Figure 5 Same as step 2.

[0043] 3. NFc (e.g., NWDAF-ANLF 1062a) contacts NRF 1054 and requests an access token using the existing procedures in [TS33501], [TS23288] and / or [TS23502].

[0044] 4. The NRF 1054 sends the access token along with the MTLFID using the existing procedures as per [TS23288], [TS33501] and / or [TS23502]. According to [TS29510], the NWDAF 1062 may add the Nnwdaf_MLModelProvision_Subscribe service operation in "allowedOperationsPerNfType" and / or "allowedOperationsPerNfInstance" for the consumer's specific NF type and / or specific instance ID and register its NF profile with the NRF 1054. When the NF Service Consumer requests an access token for Nnwdaf_MLModelProvision_Subscribe to acquire AI / ML Models, if Nnwdaf_MLModelProvision_Subscribe is present in "allowedOperationsPerNfType" (for the NF Type of the NF Service Consumer) or in "allowedOperationsPerNfInstance" (for the instance ID of the NF Service Consumer), the NRF 1054 approves the access token. The MTLF (eg, via OAM) learns the NF instance ID of the AnLF according to the existing procedures in [TS23288].

[0045] 5. NFc uses the Nnwdaf_MLModelProvision service operation for ANLF 1062a to receive the ML model ID based on the analysis ID and ADRF-id to obtain the ML model. In addition to various other information / parameters discussed in [TS23288] and / or [TS23502], this service operation may include: sending an access token to MTLF 1062b.

[0046] 6. MTLF 1062b (eg, using the integrity key K int ) validates the access token received in step 3. The MTLF 1062b may send the encryption key used in step 1 to encrypt the ML model stored in the ADRF 1066. The encryption key may be a pre-shared key or derived from an anchor key (e.g., the Authentication and Key Management for Applications (AKMA) key (K) discussed in 3GPP TS 33.535 and / or [TS33501]). AKMA)) is derived. Additionally or alternatively, the MTLF 1062b sends a one-time credential for accessing the ML model from the ADRF 1066. As an example, the one-time credential includes one or more of the following: a nonce, which is shared as part of the metadata in step 1; a message authentication code (MAC) or a binary hash or random number shared as part of the data in step 1; a signing key, which is passed as a private key of a public key in step 1; and / or the MTLF 1062b uses its signing key to generate a credential (e.g., a JSON Web Token (JWT), a certificate, etc.). Additional or alternative examples of one-time credentials include a one-time password (OTP) (e.g., time-based OTP (TOTP) and out-of-band (OOB) OTP). The one-time credential can be used to limit the number of accesses from the NFc. The one-time credential can be used as a regular authorization token for accessing the ML model multiple times (e.g., as the name implies, not just once).

[0047] 7. The consumer of the ML model (e.g., NWDAF-ANLF 1062a) requests the ML model using the ADRF service process. It also sends the one-time credentials received in step 6.

[0048] 8. The ADRF 1066 validates the one-time credentials (as specified in step 6). If the access token validation is successful, the ADRF 1066 provides the stored model to the consumer NF. The consumer of the ML model (eg, NWDAF-ANLF 1062a) decrypts the ML model using the symmetric key.

[0049] 1.1.3. Additional aspects related to security of ML model migration

[0050] In About Figure 5 and Figure 6 In the example process discussed, authorization for model acquisition at the NRF 1054 may use OAuth 2.0 token-based authorization. The NRF 1054 uses information provided by the MTLF 1062b.

[0051] NF service producers (e.g., NWDAF-MTLF 1062b) may need to register with NRF 1054, indicating NF service producer information (e.g., interoperability indicator). NF service consumers (e.g., NWDAF-ANLF 1062a) may need to register with NRF 1054 as OAuth 2.0 clients, indicating NF service consumer information used by NRF 1054 to determine whether the consumer is authorized.

[0052] According to the request of NFc for the analysis ID, MTLF 1062b performs authorization acquisition by the corresponding model of the selected model. Granularity of authorization at the AI / ML model level is performed at MTLF 1062b.

[0053] In addition, based on the decision made by MTLF 1062b, ADRF 1066 verifies that the NF consumer (e.g., MTLF 1062b or AnLF 1062a) is able to obtain the requested AI / ML model.

[0054] 1.2. Authorization and Authentication in Multi-Vendor ML Frameworks

[0055] Embodiments herein introduce new security anchoring capabilities for NWDAF 1062 and also enable model sharing in a multi-vendor environment. The motivation for introducing such anchoring capabilities is based on the following threat model for ML model sharing between multiple vendors in the same or multi-vendor environment.

[0056] The threat model involves that supplier A does not trust supplier B or supplier C, and vice versa. Supplier A, supplier B, and supplier C all trust the operator contractually. That is, they expect the operator to be a responsible integrator that protects each supplier from the other suppliers. In addition, each of supplier A, supplier B, and supplier C may also optionally require the operator to encode-decode messages that flow directly between their own MTLF 1062b.

[0057] The security anchoring function discussed herein also acts as a policy enforcer between different vendors. Additionally or alternatively, the security anchoring function discussed herein acts as a gateway for each vendor in the following scenario: The sharing of trained ML models between NWDAFs 1062 from different vendors includes: discovering and selecting NWDAF-MTLF 1062b even from different vendors that can provide interoperable trained ML models, and also obtaining interoperable trained ML models from them by NWDAF-ANLF 1062a.

[0058] Enabling ML model sharing between NWDAFs containing MTLF 1062b from the same / different vendors involves: allowing an ML model file attribute included in the NF profile of the NWDAF-MTLF 1062b indicating a list of supported ML model file serialization formats when registered with the NRF. Additionally or alternatively, the NWDAF-MTLF 1062b (NWDAF A) includes during discovery of another NWDAF-MTLF 1062b (NWDAF B) the ML model file attribute supported by the NWDAF A that only returns NWDAF-MTLF 1062b instances that support at least one matching file serialization format for ML models.

[0059] ML model sharing between multiple NWDAFs containing MTLF 1062b will be applicable to support different types of AI / ML operations (e.g., distributed learning, transfer learning, FL, etc.). For example, taking transfer learning as an example, training the weights of an ML model from scratch from random initialization may take several weeks and significantly more computing resources. Therefore, downloading weights (e.g., ML model files) from another ML model trained on NWDAF-MTLF1062b, using it as pre-training, and transferring it to a new task in another NWDAF-MTLF 1062b accelerates the progress of developing applications using a given ML model and allows more efficient use of computing resources, storage resources, and network resources.

[0060] 1.2.1 Security Anchoring Function for NWDAF

[0061] The secure anchor functionality solution relies on deploying a Vendor Repository Function (VRF) (e.g. Figure 7 The VRF 762 in the NRF 1054 operates as a gateway to each vendor and manages the ML model / MTLF 1062b of each vendor. The following deployment scenarios may be applicable to the use of the security anchor function: (1.2.1.1) VRF is co-located with NWDAF 1062 or is an independent NF; (1.2.1.2) VRF is a logical function within NRF 1054 and / or is co-located with NRF 1054.

[0062] The solution also ensures that only authenticated service consumers (e.g., service consumers that already have access to ML models and / or analytical services) have access to ML models and / or analytical services. In one example, in a multi-vendor environment, producers of ML models and / or analytical services may be prevented from (self-)accessing analytical consumers from the same vendor or multiple vendors. In another example, a NWDAF-MTLF 1062b that creates an ML model from vendor A is not self-permitted by the ANLF 1062a of vendor B; however, a NWDAF-ANLF 1062a from the same vendor (e.g., vendor B) may be allowed to use the ML model from vendor B.

[0063] 1.2.1.1. VRF Co-located with NWDAF or as a Standalone NF

[0064] Figure 7 An example process for registration of a VRF 762 with an NRF 1054 is shown. In this example, each vendor may have its own VRF 762 and a list of NWDAF-MTLFs 1062b that it supports. In these deployment scenarios, VRF 762 instances are pre-configured in NWDAF-MTLFs 1062b, which contain trained ML models to be registered with the VRF 762. In addition, the operator deploys a separate VRF 762 for all vendors, and all MTLFs 1062b of all vendors are registered with the VRF 762. Regardless of the deployment scenario, all NWDAF-MTLFs 1062b use Figure 7 The process in registers to VRF 762, which can be operated as follows.

[0065] 1. Registration of NWDAF-MTLF 1062b with VRF 762: NWDAF-MTLF 1062b sends Nvrf_NFManagement_NFRegister to VRF 762 to inform VRF 762 of its NF profile. In some examples, Nvrf_NFManagement_NFRegister is applicable to the case where VRF 762 is an independent NF, but not to the case where NWDAF-MTLF 1062b also supports VRF 762 functions. In other examples, Nvrf_NFManagement_NFRegister is applicable to the case where NWDAF-MTLF 1062b includes VRF 762 functions. Instead of registering with NRF 1054, NWDAF-MTLF 1062b registers with VRF 762. The rest of the process and content are the same as discussed in [TS23502]. The vendor ID attribute and / or model ID can also indicate the vendor.

[0066] When a trained ML model is available for one or more analysis IDs, the NWDAF-MTLF 1062b includes an ML model provisioning service (e.g., Nnwdaf_MLModelProvision, Nnwdaf_MLModelInfo) as one of the supported services during registration in the VRF 762.

[0067] The NWDAF-MTLF 1062b may provide a list (set) of analysis IDs corresponding to trained ML models and / or ML model filtering information for trained ML models by analysis ID (if available) to the VRF 762. Additionally or alternatively, Nvrf_NFManagement_NFRegister may include analysis information (e.g., analysis ID), ML model interoperability information (e.g., requested model file format, model execution environment, etc.), and / or any other information / parameters discussed in Section 1.3.3 below.

[0068] 2. NWDAF-VRF 762 (or VRF 762 in case VRF 762 is a standalone NF) sends Nnrf_NFManagement_NFRegister to NRF 1054 to inform NRF 1054 of its NF profile. Nnrf_NFManagement_NFRegister includes analysis ID (possibly by service), NWDAF service area information, and supported analysis delays by analysis ID (if available). In some examples, Nnrf_NFManagement_NFRegister includes analysis aggregation capabilities and / or analysis metadata provisioning capabilities (if NWDAF 1062 provides such capabilities). Additionally or alternatively, Nnrf_NFManagement_NFRegister also includes ML model filtering information parameters S-NSSAI and AoI for trained ML models by analysis ID and ML model interoperability indicator by analysis ID (if available) (see, e.g., clause 5.2 of [TS23288]). Additionally or alternatively, if data management services are available, Nnrf_NFManagement_NFRegister also includes the NF set ID and NF type of the NF data source (see, for example, clause 6.3.13 of [TS23501]). In addition to the NF profile parameters discussed herein and / or defined in [TS23502], Nnrf_NFManagement_NFRegister also includes a list of supported vendors, a list of ML models from each vendor (owned), ML model filtering information (e.g., S-NSSAI and / or AoI for trained ML models by analysis ID), and ML model interoperability information (e.g., ML model serialization format, etc.). Nnrf_NFManagement_NFRegister may include other information (e.g., any information discussed in Section 1.3.3 below). In some examples, the consumer ANLF 1062a belongs to the same vendor as the NWDAF-MTLF 1062b, or the ANLF 1062a and the NWDAF-MTLF 1062b belong to different vendors.

[0069] In one example, a list of vendors (mlmodelvendorList) is provided as part of the MlAnalyticsInfo parameter, as shown in Table 1.2.1-1. Here, MlAnalyticsInfo is a Nnrf_NFManagement specific data type that provides ML analysis filtering information supported by the Nnwdaf_MLModelProvision service (see, for example, clause 6.1.6.2.84 in [TS29510]). The MlAnalyticsInfo can be provided in mlAnalyticsList (e.g., array(MlAnalyticsInfo)), which includes a list of ML analysis filtering information per analysis ID supported by the NWDAF 1062 and whose status is requested to be monitored. If present (e.g., in a suitable GET request), this attribute contains a list of ML analysis filtering information per analysis ID supported by the Nnwdaf_MLModelProvision service request. The NRF 1054 returns the NWDAF profile that supports at least one MlAnalyticsInfo in the list.

[0070] Table 1.2.1-1: Definition of type MlAnalyticsInfo

[0071]

[0072]

[0073] 3-4. Discovery via NWDAF-MTLF 1062b of NRF 1054:

[0074] 3. During the discovery of NWDAF-MTLF 1062b, NFc (e.g., NWDAF-ANLF 1062a) calls Nnrf_NfDiscovery_Request service. Nnrf_NfDiscovery service enables NFc (e.g., NWDAF-ANLF 1062a) to discover a set of NF instances with a specific NF service or target NF type. As an example, the NFc (e.g., NWDAF-ANLF 1062a) includes one or more of the following in the request: analysis ID (possibly by service), TAI, analysis aggregation capability, analysis metadata provisioning capability, real-time communication indication by analysis ID, NF set ID and NF type of NF data source (e.g., target NF type (e.g., NWDAF 1062)), S-NSSAI, AoI of the required trained ML model, NF consumer information (e.g., when the target is NWDAF-MTLF 1062b), FL capability type and / or time period of interest (e.g., when the target is NWDAF-MTLF 1062b with FL capability) (see, e.g., clause 6.3.13 of [TS23501]). In some examples, the request includes NFc information (e.g., vendor ID, etc.). Additionally or alternatively, the NFc (e.g., NWDAF-ANLF 1062a) may call Nnrf_AccessToken_service to obtain an access token for accessing the stored ML model.

[0075] 4. The NRF 1054 provides a Nnrf_NFDiscovery_Response to return one or more candidates for an instance of the NWDAF-MTLF 1062b to the NFc (e.g., NWDAF-ANLF 1062a). The one or more candidates for an instance of the NWDAF-MTLF 1062b may be or include a set of candidate VRF 762 instances or a set of NWDAF-VRF 762 instances. As an example, for each candidate instance, the Nnrf_NfDiscovery_Response includes an analysis ID and (possibly) ML model filtering information for an available trained ML model (if available).

[0076] 5-7.ANLF 1062a uses the Nnwdaf_MLModelProvision service as specified in [TS23502] and / or [TS23288]:

[0077] 5a. NFc (e.g., NWDAF-ANLF 1062a) calls Nnwdaf_MLModelProvision_Request service operation to subscribe / request a set of trained ML models associated with a set of analysis IDs. Nnwdaf_MLModelProvision_Request is sent to an instance selected from the set of instances obtained in step 4. As an example, Nnwdaf_MLModelProvision_Request may include analysis information (e.g., analysis ID), ML model interoperability information (e.g., ML model file serialization format, supported hardware (HW) platforms / requirements, software (SW) requirements, etc.), ML model filtering information, and / or other information discussed herein.

[0078] 5b. NWDAF-VRF 762 calls the Nnwdaf_MLModelProvision_Response service operation with a token (eg, OAuth2 2.0 access token, etc.).

[0079] In some examples, when the NFc (e.g., NWDAF-ANLF 1062a) sends a Nnwdaf_MLModelProvision_Request to the NWDAF including the VRF 762 using the Nnwdaf_MLModelProvision_Request service, the NWDAF including the VRF 762 notifies the NFc (e.g., NWDAF-ANLF 1062a) of the ML model information (or ML model filtering (e.g., address / reference of the ML model file (e.g., URL, FQDN, IP address, etc.) etc.) only when the requested ML model format in the input of the Nnwdaf_MLModelProvision_Request matches.

[0080] VRF 762 checks whether NFc (e.g., NWDAF-ANLF 1062a) can consume the ML model in the described architecture / environment. If the check is successful, VRF 762 generates an access token. The access token received by the consumer from VRF 762 contains information that the consumer (e.g., NWDAF-ANLF 1062a) can obtain from the ML model from NWDAF-MTLF 1062b. In addition to the existing parameters, the token also contains the following information: the vendor ID of the NFc (e.g., NWDAF-ANLF 1062a) and the vendor ID of the service producer (e.g., NWDAF-MTLF 1062b), the NF ID (if any) and the name of the data that is allowed to use the data, the timeline of the allowed access, and / or any other information related to the contractual agreement between the two vendors (e.g., if vendor A's NWDAF-ANLF 1062a can access vendor B's NWDAF-MTLF 1062b, then VRF762 provides a token to subscribe to vendor B's NWDAF-MTLF 1062b). The rest of the process can be similar to that of [TS23502].

[0081] 6. The NWDAF service consumer (e.g., NWDAF-MTLF 1062b) subscribes / requests a set of trained ML models associated with a set of analysis IDs by calling the Nnwdaf_MLModelProvision_Subscribe (or Nnwdaf_MLModelProvision_Request) service operation. In some examples, Nnwdaf_MLModelProvision_Subscribe (or Nnwdaf_MLModelProvision_Request) includes the token obtained in step 7 or some information included in the token.

[0082] 7. If the NWDAF service consumer subscribes to and (is associated with) a set of trained ML models of a set of analysis IDs, the NWDAF service producer (e.g., NWDAF-MTLF 1062b) notifies the NWDAF service consumer (e.g., NWDAF-MTLF 1062b) of a set of unique ML model identifier and ML model information pairs associated with each analysis ID requested by the NWDAF service consumer by calling the Nnwdaf_MLModelProvision_Notify service operation. The NWDAF includes the NWDAF-MTLF 1062b service consumer with a set of unique ML model identifier and ML model information pairs associated with each analysis ID requested by the service consumer (e.g., NWDAF-MTLF 1062b). In some examples, for each ML model, Nnwdaf_MLModelProvision_Notify includes the analysis ID, the ML model file address (e.g., URL, FQDN, IP address, etc.), notification relevance information, etc.). Additional or alternative content of the trained ML model information that may be provided by the NWDAF-MTLF 1062b is specified in clause 6.2A.2 of [TS23288] and / or in section 1.3.3 below.

[0083] 1.2.1.1.1.MTLF Discovery and Selection

[0084] In some implementations, MTLF 1062b discovery and selection works as follows: The NWDAF service consumer selects a VRF 762 that supports the requested analytics information and required analytics capabilities and / or the requested ML model information by using the previously described NWDAF discovery principles.

[0085] To discover NWDAF-MTLF 1062b via VRF 762:

[0086] 1. When trained ML models are available for one or more analysis IDs, NWDAF-MTLF 1062b includes a set of ML model provisioning services (e.g., Nnwdaf_MLModelProvision, Nnwdaf_MLModelInfo, etc.) as one of the supported services during registration in VRF 762. NWDAF-MTLF 1062b may provide (a list of) analysis IDs corresponding to trained ML models and possibly ML model filtering information for trained ML models by analysis ID to VRF 762, if available. In some examples, during NWDAF-MTLF 1062b registration, only S-NSSAI and AoI from the ML model filtering information for trained ML models by analysis ID are registered into VRF 762.

[0087] 2. VRF 762 registers with NRF 1054 using a list of trained ML models for one or more analysis IDs, including ML model filtering information for the trained ML models by analysis ID (if available).

[0088] 3. During the discovery of the NWDAF-MTLF 1062b, the consumer (e.g., NWDAF-ANLF 1062a or another NWDAF-MTLF 1062b) may include the target NF type (e.g., NWDAF), analysis ID, S-NSSAI, and AoI of the desired trained ML model in the (discovery) request. The NRF 1054 returns one or more candidates for the instance of the VRF 762 containing the MTLF to the NF consumer, and each candidate for the instance of the NWDAF-MTLF 1062b includes the analysis ID and possibly ML model filtering information for the available trained ML models (if available).

[0089] 1.2.1.1.2.NR Role

[0090] In addition to the roles and / or functions described in [TS23501], [TS23502], and / or [TS23288], the following NFs include the following roles / functions:

[0091] NRF 1054 stores the NF profile for the VRF 762 instance, which includes "Vendor Information" and "Analysis Metadata Provisioning Capability" for each NWDAF 1062 (when supported by the NWDAF 1062); and returns the VRF 762 that matches the attributes provided in the Nnrf_NFDiscovery_Request, as specified in clause 5.2.7.3 of TS23.502.

[0092] The NWDAF service consumer requests or subscribes to receive analytics for one or more analytics IDs; uses the discovery mechanism from NRF 1054 defined in clause 6.3.13 of [TS23501] to identify VRF 762 with analytics aggregation capabilities and other capabilities (e.g., providing data / analytics for a specific TAI); and if multiple VRFs 762 are returned in the NWDAF discovery response, can distinguish and select a preferred VRF 762 based on its internal selection criteria (taking into account the NWDAF capabilities and information registered in NRF 1054 or UDM 1058).

[0093] VRF 762 stores the NF profile of the NWDAF MTLF instance, which includes the "vendor policy" for the vendor NWDAF and the "analysis it supports"; and returns the NWDAF MTLF matching the attributes provided in the Nnrf_NFDiscovery_Request, as specified in clause 5.2.7.3 of [TS23502].

[0094] 1.2.1.2. VRF as a logical function / entity within NRF and / or co-located with NRF

[0095] 1.2.1.2.1.MTLF Discovery and Selection

[0096] In implementations where VRF 762 is a logical function that is part of or co-located with NRF 1054, Figure 7 The MTLF discovery and selection process can be adapted to operate as follows:

[0097] 1. When trained ML models are available for one or more analysis IDs, the NFp (e.g., NWDAF-MTLF 1062b) includes an ML model provisioning service (e.g., Nnwdaf_MLModelProvision, Nnwdaf_MLModelInfo) as one of the supported services during registration in the VRF 762. The NWDAF-MTLF 1062b may provide the VRF 762 with a list or set of analysis IDs corresponding to the trained ML models, and possibly ML model filtering information for the trained ML models by analysis ID, if available. In some examples, during NWDAF-MTLF 1062b registration, only the S-NSSAI and AoI from the ML model filtering information for the trained ML models by analysis ID are registered in the VRF 762.

[0098] 2. NRF 1054 returns one or more candidates for instances of NWDAF-MTLF 1062b to NFc (or a set of candidate VRF 762 instances or a set of NWDAF-VRF 762 instances) via VRF 762. Each candidate instance includes an analysis ID and possibly ML model filtering information for an available trained ML model (if available).

[0099] Except that VRF 762 is a logical function of NRF 1054 and / or is co-located with NRF 1054, Figure 7 The rest of the MTLF discovery and selection process is the same as previously discussed in Section 1.2.1.1.

[0100] 1.2.2. Discovery

[0101] The embodiments discussed herein enable ML model sharing between NWDAFs containing MTLF 1062b, and discovery and selection of a NWDAF-MTLF 1062b by another NWDAF-MTLF 1062b. The solution described below is applicable to NWDAFs belonging to the same vendor or different vendors.

[0102] 1.2.2.1. Registration of NWDAF including MTLF to NRF

[0103] Figure 8 An example process for NWDAF-MTLF registration with NRF 1054 is depicted.

[0104] Figure 8 The process can be operated as follows:

[0105] 1. The NF service producer (NFp) (e.g., NWDAF-MTLF 1062b) sends Nnrf_NFManagement_NFRegister to NRF 1054 to inform its NF profile to NRF 1054. The Nnrf_NFManagement_NFRegister service registers NFc (e.g., NWDAF-MTLF 1062b) in NRF 1054 by providing the NF profile of NFc to NRF 1054, and NRF 1054 marks the available NFc.

[0106] In addition to the NF profile parameters discussed herein and / or defined in [TS23502], the NF profile also includes supported ML model file serialization formats for trained ML models and interoperability support by analysis ID in the ML model filtering information. The ML model file serialization format included in the ML model filtering information indicates supported ML model serialization formats for trained ML models available for consumption by service consumers at NWDAF-MTLF 1062b. The consumer of the service provided by NWDAF-MTLF 1062b can be NWDAF-ANLF 1062a or NWDAF-MTLF 1062b. The consumer NF can belong to the same vendor as NWDAF-MTLF 1062b or a different vendor. The interoperability support parameter by analysis ID indicates whether NWDAF-MTLF 1062b (in the role of a producer) has support for ML model interoperability. Fig. 9 An example is shown.

[0107] 2. NFc (e.g., NWDAF-MTLF 1062b) sends Nnrf_NfDiscovery_Request to NRF 1054 to obtain information about NF service instances (e.g., NFp( Figure 8 In addition to the parameters discussed herein and / or defined in [TS23502], the Nnrf_NfDiscovery_Request may also include ML model support information (e.g., ML model serialization format, etc.), required interoperability support (see e.g., Fig. 9 etc.

[0108] 3. As previously described, the NRF 1054 provides a Nnrf_NFDiscovery_Response to return one or more candidate instances of the NWDAF-MTLF 1062b.

[0109] Fig. 9 An example of interoperability support by analysis ID is depicted. In this example, NWDAF-MTLF 1062b supports providing trained ML models for four (4) types of analysis (e.g., analysis ID 1, analysis ID 2, analysis ID 3, and analysis ID 4).

[0110] There are four (4) trained ML models for analysis ID 1, including ML Model 1 and ML Model 3, both of which support interoperability, while ML Model 2 and ML Model 4 do not support interoperability with NWDAF 1062. During NWDAF registration with NRF 1054, if MTLF 1062b has at least one ML model that supports interoperability, then NWDAF-MTLF 1062b will include the supported interoperability. In this case, NWDAF-MTLF 1062b will indicate interoperability support for analysis ID 1.

[0111] There are two (2) trained ML models for analysis ID 4, including ML Model 5 and ML Model 6, each of which does not support interoperability. During NWDAF registration with NRF 1054, NWDAF-MTLF 1062b will not provide interoperability support for analysis ID4.

[0112] 1.2.2.2. Discovery of NWDAF including MTLF via NRF

[0113] In some implementations, the NWDAF-MTLF 1062b invokes Nnrf_NfDiscovery_Request to the appropriately configured NRF 1054. It includes analysis ID, S-NSSAI, AoI, supported ML model file serialization formats for trained ML models in ML model filtering information, interoperability support, and / or other suitable information. The NRF 1054 determines a set of NWDAF-MTLF 1062b instances that match at least one of the ML model file serialization format and interoperability support in the Nnrf_NFDiscovery_Request and the internal policy of the NRF 1054, and sends the NF profile (including the ML model file serialization format and / or other suitable information) of the determined NWDAF-MTLF 1062b instance in the discovery response.

[0114] 1.3.ML Model Provisioning

[0115] Various aspects related to ML model provisioning and sharing discussed below are applicable to the various embodiments discussed previously.

[0116] The NWDAF-ANLF 1062a may be locally configured with a set of IDs of NWDAFs including MTLFs and analysis IDs supported by each NWDAF-MTLF 1062b to obtain trained ML models, or the NWDAF discovery process specified in clause 5.2 of [TS23288] may be used to discover the NWDAF including the MTLF 1062b. When the NWDAF-MTLF 1062b receives an ML model subscription or ML model request, it may determine that further training of the existing ML model is required. Various aspects of secure ML model provisioning / sharing between multiple MTLFs 1062b and / or multiple ML model vendors are discussed in Sections 1.1 and 1.2 above.

[0117] 1.3.1.ML model subscription / unsubscription

[0118] For analyzing the subscription / unsubscription process (see e.g. [TS23288] Figure 6 .2A.1-1) is used by an NWDAF service consumer (e.g., NWDAF-ANLF 1062a) to subscribe / unsubscribe at an NWDAF service producer (e.g., NWDAF-MTLF 1062b) using the Nnwdaf_MLModelProvision service defined in clause 7.5 of [TS23288] (see, e.g., Figure 6.2A.1-1) to be notified when ML model information for relevant analytics becomes available. ML model information is used by NWDAF-ANLF 1062a to derive analytics. The service is also used by NWDAF 1062 to modify existing ML model subscriptions. NWDAF 1062 can be both a consumer of the service provided by other NWDAF 1062 and a provider of the service to other NWDAF 1062.

[0119] The NWDAF service consumer (e.g., NWDAF-ANLF 1062a) subscribes, modifies, or unsubscribes to a set of trained ML models associated with a set of analysis IDs by calling the Nnwdaf_MLModelProvision_Subscribe / Nnwdaf_MLModelProvision_Unsubscribe service operations. The parameters that may be provided by the NWDAF service consumer are listed in clause 6.2A.2 of [TS23288] and / or in section 1.3.3 below. The service consumer may indicate its support for multiple ML models (if available).

[0120] When receiving a subscription for a trained ML model associated with an analysis ID, the NWDAF service producer (e.g., NWDAF-MTLF 1062b) may determine whether an existing trained ML model can be used for the subscription, and / or determine whether further training of the existing trained ML model needs to be triggered for the subscription. If the NWDAF-MTLF 1062b determines that further training is required, the NWDAF 1062 may initiate data collection from relevant NFs (e.g., AMF 1044, DCCF 1063, ADRF 1066, etc.), UE applications (via AF 1060), and / or OAM to generate the ML model as described in clause 6.2 of [TS23288]. If the service call is for subscription modification or subscription cancellation, the NWDAF service consumer includes the identifier to be modified (e.g., subscription correlation ID) in the call to Nnwdaf_MLModelProvision_Subscribe.

[0121] If the NWDAF service consumer subscribes to and (with) a set of trained ML models for a set of analysis IDs, the NWDAF service producer (e.g., NWDAF-MTLF 1062b) notifies the NWDAF service consumer (e.g., NWDAF-MTLF 1062b) of a set of unique ML model identifier and ML model information pairs associated with each analysis ID requested by the NWDAF service consumer by calling the Nnwdaf_MLModelProvision_Notify service operation. The content of the trained ML model information that can be provided by the NWDAF-MTLF 1062b is specified in clause 6.2A.2 of [TS23288] and / or in section 1.3.3 below.

[0122] When the NWDAF-MTLF 1062b determines that the previously provided trained ML model needs to be retrained (as previously described), the NWDAF-MTLF 1062b also calls the Nnwdaf_MLModelProvision_Notify service operation to notify the available retrained ML model. When a subscription modification is requested (e.g., including a subscription correlation ID), the NWDAF-MTLF 1062b can provide a new trained ML model that is different from the previously provided ML model, or provide a retrained ML model by calling the Nnwdaf_MLModelProvision_Notify service operation.

[0123] Additionally, the Nnwdaf_MLModelProvision_Notify service operation may also be used to convey the analysis ID, the address of / reference to the requested / subscribed ML model file, notification dependency information, and / or other information previously discussed.

[0124] 1.3.2.ML Model Request

[0125] ML model request process (see e.g. [TS23288] Figure 6 .2A.3-1) is used by NWDAF service consumers (e.g., NWDAF-ANLF 1062a) to request and obtain ML model information from NWDAF service producers (e.g., NWDAF-MTLF 1062b) using the Nnwdaf_MLModelInfo service defined in clause 7.6 of [TS23288]. The ML model information is used by NWDAF-ANLF 1062a to derive analytics. NWDAF 1062s can be both consumers of the services provided by other NWDAFs 1062 and providers of the services to other NWDAFs 1062.

[0126] The NWDAF service consumer (e.g., NWDAF-ANLF 1062a) requests a set of ML models associated with a set of analysis IDs by calling the Nnwdaf_MLModelInfo_Request ML service operation. The parameters that may be provided by the NWDAF service consumer (e.g., NWDAF-ANLF 1062a) are listed in clause 6.2A.2 of [TS23288] and / or in section 1.3.3 below. The service consumer may optionally indicate its support for multiple ML models, if available. Upon receiving a request for ML model information for analysis, the NWDAF-MTLF 1062b may: determine whether an existing trained ML model may be used for the request and / or determine whether further training of an existing trained ML model needs to be triggered for the request. If the NWDAF-MTLF 1062b determines that further training is required, the NWDAF 1062 may initiate data collection from relevant NFs (e.g., AMF 1044, DCCF 1063, ADRF 1066, etc.), UE applications (via AF 1060), and / or OAM to generate an ML model as described in clause 6.2 of [TS23288].

[0127] The NWDAF-MTLF 1062b responds to the NWDAF service consumer (e.g., NWDAF-ANLF 1062a) by invoking a Nnwdaf_MLModelInfo_Request response service operation including, for example, a set of unique ML model identifier and ML model information pairs for each analysis ID requested by the NWDAF service consumer. The content of the ML model information that may be provided by the NWDAF-MTLF 1062b is specified in clause 6.2A.2 of [TS23288] and / or in section 1.3.3 below.

[0128] 1.3.3. Contents of ML model allocation

[0129] A consumer of the ML model provisioning service described in clauses 7.5 and 7.6 of [TS23288] (e.g., NWDAF-ANLF 1062a) may provide one or more of the following input parameters.

[0130] - Analysis Information: This parameter includes information about the analysis that the requested ML model is to be used for. As an example, this parameter may include a list (set) of analysis IDs (e.g., identifying the analysis that the ML model is to be used for) and NF consumer information (e.g., identifying the vendor of the NWDAF-ANLF 1062a). The NF consumer information (e.g., vendor ID) may include any suitable identifier / network address (e.g., including any of the identifiers / network addresses discussed herein), may be generated using any suitable mechanism (e.g., a hashing algorithm, a random number / string generator, etc.), and / or may be implementation specific.

[0131] - Use case context: This parameter indicates the context in which the analysis is used to select the most relevant ML model. In some examples, when several ML models are available for the requested analysis ID, the NWDAF-MTLF 1062b may use the "Use case context" parameter to select the most relevant ML model.

[0132] -ML model interoperability information: includes vendor-specific information that conveys information for interoperability and / or sharing of ML models. Examples of ML model interoperability information include requested ML model file format, ML model serialization format, model execution environment (e.g., hardware platform requirements, software / runtime environment, etc.), hyperparameters, etc. The encoding, format, and value of ML model interoperability information can be vendor-specific information agreed upon between vendors for sharing purposes.

[0133] -ML model filter information: enables selection of the ML model for which analysis is requested. Examples of ML model filter information include S-NSSAI; Area of ​​Interest (AoI); list / set of ML models (e.g., ML model IDs), TAI; Cell ID; list / set of application IDs, DNNs and / or DNAIs; etc. The parameter types in the ML model filter information are the same as those in the analysis filter information discussed in [TS23288].

[0134] -Target of ML model reporting: indicates the object for which the ML model is requested for a specific UE, a group of UEs, or any or all UEs.

[0135] - Requested representative ratio: When the target reported by the ML model is a group of UEs, the minimum percentage of UEs in the group whose data is a non-empty set and can be used in model training.

[0136] - ML model reporting information with the following parameters: For Nnwdaf_MLModelProvision_Subscribe, the ML model reporting information parameters of the event reporting information parameters defined in Table 4.15.1-1 in [TS23502].

[0137] -ML Model Target Period: Indicates the time interval [start, end] for which the ML model is requested for analysis. The time interval is expressed in terms of actual start time and actual end time (e.g., via UTC time, etc.).

[0138] - Inference input data information: contains information about various settings expected to be used by the AnLF 1062a during inference (e.g., the input data expected to be used and the data source expected to be used as a list of NF instance (or NF set) identifiers). Each input data expected to be used may be accompanied by a metric showing the granularity at which that data will be used (e.g., sampling rate, maximum number of input values, and / or maximum time interval between sampling of that input data). This may be a subset of the possible input data specified for a particular analysis type.

[0139] - The notification target address (+ notification correlation ID) defined in clause 4.15.1 of [TS23502], allowing notifications received from the NWDAF-MTLF 1062b to be correlated with this subscription.

[0140] - Indications for supporting multiple ML models.

[0141] - Accuracy level of interest.

[0142] -Number of ML models, indicating the maximum number of multiple ML models that an ML model provider / supplier (e.g., NWDAF-MTLF 1062b) can provide to consumers of ML models. The multiple ML model filtering information consists of an indication of supporting multiple ML models, the accuracy level of interest, the number of ML models, etc.

[0143] NWDAF-MTLF 1062b provides the following listed output information to consumers of the ML model provisioning service operations described in clauses 7.5 and 7.6 of [TS23288]:

[0144] - Notify dependency information (e.g., for Nnwdaf_MLModelProvision_Notify).

[0145] - For each analysis ID requested by the service consumer, a set of unique ML model identifier and ML model information pairs.

[0146] -ML model information, including, for example, an ML model file address (e.g., a URL, FQDN, IP address, and / or some other reference, identifier, or network address (e.g., any reference, identifier, or network address discussed herein)).

[0147] - Validity period: Indicates the period of time that the provided ML model information is applicable. In some examples, the validity period is determined by the internal logic of the MTLF 1062b and is a subset of the AoI (if provided in the ML model filter information) and the ML model target period, respectively.

[0148] - Spatial validity: Indicates the region to which the provided ML model information is applicable. The spatial validity and validity period are determined by MTLF internal logic and are subsets of the AoI (if provided in the ML model filter information) and the ML model target period, respectively.

[0149] -ML model representative ratio: When the target of ML model reporting is a group of UEs, indicates the percentage of UEs in the group whose data are used in ML model training.

[0150] - Training input data information: contains information about various settings that the MTLF has used during training (e.g., input data that has been used and data sources that have been used as lists of NF instance (or NF set) identifiers). The input data that has been used may be accompanied by metrics showing the granularity at which this data has been used (e.g., sampling rate, maximum number of input values, and / or maximum time interval between sampling of this input data). This may be a subset of possible input data specified for a particular analysis type.

[0151] 2. Network, system and equipment configuration and layout

[0152] Fig.10 An example network architecture 1000 is depicted. The network 1000 may operate in a manner consistent with 3GPP technical specifications for LTE or 5G / NR systems. However, the example embodiments are not limited thereto, and the described examples may be applied to other networks (e.g., future 3GPP systems, etc.) that benefit from the principles described herein.

[0153] The network 1000 includes a UE 1002, which is any mobile or non-mobile computing device designed to communicate via an air connection with a RAN 1004. The UE 1002 is communicatively coupled to the RAN 1004 via a Uu interface that may be applicable to both LTE systems and NR systems. Examples of UE 1002 include, but are not limited to, smartphones, tablet computers, wearable devices (e.g., smart watches, fitness trackers, smart glasses, smart clothing / fabrics, head-mounted displays, smart shows, etc.), desktop computers, workstations, laptop computers, in-vehicle infotainment systems, in-vehicle entertainment systems, instrument clusters, heads-up display (HUD) devices, onboard diagnostic devices, dashboard mobile devices, mobile data terminals, electronic engine management systems, electronic / engine control units, electronic / engine control modules, embedded systems, sensors, microcontrollers, control modules, engine management systems, connected appliances, machine type communication devices, machine to machine (M2M), device to device (D2D), machine type communication (MTC) devices, Internet of Things (IoT) devices, smart appliances, flying drones or unmanned aerial vehicles (UAVs), ground drones or autonomous vehicles, robots, electronic signs, single board computers (SBCs) (e.g., Raspberry Pi, Arduino, Intel Edison, etc.), plug-in computers and / or any type of computing device (e.g., any device discussed herein). UE 1002 may be the same as or similar to any other UE discussed herein (eg, UE 1102, hardware resources 1200, and / or any other UE discussed herein).

[0154] The network 1000 may include a group of UEs 1002 that are directly coupled to each other via device-to-device (D2D), proximity services (ProSe), PC5 and / or sidelink (SL) interfaces and / or any other suitable interface (e.g., any interface discussed herein). These UEs 1002 may be M2M, D2D, MTC and / or IoT devices and / or V2X systems that communicate using physical sidelink channels (e.g., but not limited to PSBCH, PSDCH, PSSCH, PSCCH, PSFCH, etc.). UE 1002 may perform blind decoding attempts of SL channels / links according to various examples herein.

[0155] In some examples, the UE 1002 may additionally communicate with the AP 1006 via an over-the-air (OTA) connection. The AP 1006 manages a WLAN connection, which may be used to offload some / all network traffic from the RAN 1004. The connection between the UE 1002 and the AP 1006 may conform to any IEEE 802.11 protocol. Additionally, the UE 1002, the RAN 1004, and the AP 1006 may utilize cellular-WLAN aggregation / integration (e.g., LWA / LWIP). Cellular-WLAN aggregation may involve the UE 1002 being configured by the RAN 1004 to utilize both cellular radio resources and WLAN resources.

[0156] The RAN 1004 includes one or more network access nodes (NAN) 1014 (also referred to as "RAN nodes 1014"). The NAN 1014 terminates the air interface for the UE 1002 by providing access layer protocols including RRC, PDCP, RLC, MAC, and PHY / L1 protocols. In this way, the NAN 1014 implements data / voice connectivity between the core network (CN) 1040 and the UE 1002. The NAN 1014 can be a macrocell base station, or a low-power base station for providing a femtocell, a picocell, or other similar cells with a smaller coverage area, smaller user capacity, or higher bandwidth than a macrocell; or some combination thereof. In these implementations, the NAN 1014 may be referred to as a base station (BS), a next-generation nodeB (gNB), a RAN node, an eNodeB (eNB), a next-generation (ng)-eNB, a NodeB, an RSU, a TRP, etc.

[0157] One example implementation is a "CU / DU split" architecture, where the NAN 1014 is embodied as a gNB-central unit (CU) communicatively coupled to one or more gNB-distributed units (DUs), where each DU may be communicatively coupled to one or more radio units (RUs) (also referred to as RRHs, RRUs, etc.). In some implementations, the one or more RUs may be separate RSUs. In some implementations, the CU / DU split may also include an ng-eNB-CU and one or more ng-eNB-DUs, respectively, instead of or in addition to the gNB-CU and gNB-DU. The NAN 1014 employed as a CU may be implemented in a discrete device or as one or more software entities running on a server computer as part of, for example, a virtual network (including a virtual baseband unit (BBU) or a pool of BBUs, a cloud RAN (CRAN), a radio equipment controller (REC), a radio cloud center (RCC), a centralized RAN (C-RAN), a virtualized RAN (vRAN), etc. (although these terms may refer to different implementation concepts)). Any other type of architecture, arrangement, and / or configuration may be used.

[0158] The set of NANs 1014 are coupled to each other via respective X2 interfaces (if the RAN 1004 is an LTE RAN or an Evolved Universal Terrestrial Radio Access Network (E-UTRAN) 1010) or respective Xn interfaces (if the RAN 1004 is an NG-RAN 1004). The X2 / Xn interfaces (which in some examples may be separated into control / user plane interfaces) may allow the ANs to communicate information related to handover, data / context transfer, mobility, load management, interference coordination, etc.

[0159] The ANs of the RAN 1004 may each manage one or more cells, cell groups, component carriers, etc. to provide an air interface for network access to the UE 1002. The UE 1002 may be simultaneously connected to a group of cells provided by the same or different NANs 1014 of the RAN 1004. For example, the UE 1002 and the RAN 1004 may use carrier aggregation to allow the UE 1002 to connect to a group of component carriers, each corresponding to a Pcell or an Scell. In a dual connectivity scheme, the first AN 1008 may be a primary node providing an MCG, and the second AN 1008 may be a secondary node providing an SCG. The first / second NAN 1014 may be any combination of eNBs, gNBs, ng-eNBs, etc.

[0160] RAN 1004 may provide an air interface through a licensed spectrum or an unlicensed spectrum. In order to operate in an unlicensed spectrum, a node may use LAA, eLAA, and / or feLAA mechanisms based on CA technology with PCells / Scell. Before accessing the unlicensed spectrum, a node may perform a medium / carrier sensing operation based on, for example, a listen-before-talk (LBT) protocol.

[0161] Additionally or alternatively, each UE 1002 provides radio information to one or more NANs 1014 and / or one or more edge computing nodes (e.g., edge servers / hosts, etc.). The radio information may be in the form of one or more measurement reports, and / or may include, for example, signal strength measurements, signal quality measurements, etc. Each measurement report is tagged with a timestamp and location of the measurement (e.g., the current location of the UE 1002). As an example, the UE 1002 may provide a radio information report to the NAN 1014 and / or one or more edge computing nodes (e.g., edge servers / hosts, etc.). The measurements collected at 1002 and / or included in the measurement report may include one or more of the following: bandwidth (BW), network or cell load, latency, jitter, round trip time (RTT), number of interruptions, out-of-order delivery of data packets, transmit power, bit error rate, bit error ratio (BER), block error rate (BLER), packet error rate (PER), packet loss rate, packet reception rate (PRR), data rate, peak data rate, end-to-end (e2e) delay, signal-to-noise ratio (SNR), signal-to-interference-plus-noise ratio (SINR), signal-plus-noise-plus-distortion to noise-plus-distortion (SINAD) ratio, carrier-to-interference-plus-noise ratio (CINR), additive white Gaussian noise (AWGN), energy per bit to noise power density ratio (Eb / N0), energy per chip to interference power density ratio (Ec / I0), energy per chip to noise power density ratio (Ec / N0), peak to average power ratio (PAPR), reference signal received power (RSRP), reference signal received quality (RSRQ), received signal strength indication (RSSI), received channel power indication (RCPI), received signal to noise indication (RSNI), received signal code power (RSCP), average noise plus interference (ANPI), GNSS timing of cell frames for UE positioning for E-UTRAN or 5G / NR (e.g., AP 1006 or RAN node 1008 reference time and a GNSS-specific reference time for a given GNSS), GNSS code measurements (e.g., the GNSS code phase (integer and fractional parts) of the spreading code of the ith GNSS satellite signal), GNSS carrier phase measurements (e.g., the number (integer and fractional parts) of the carrier phase cycles of the ith GNSS satellite signal measured since locking onto the signal; also known as accumulated delta range (ADR)), channel interference measurements, thermal noise power measurements, received interference power measurements, power histogram measurements, channel load measurements, STA statistics, and / or other similar measurements.RSRP, RSSI and / or RSRQ measurements may include RSRP, RSSI and / or RSRQ measurements of cell-specific reference signals, channel state information reference signals (CSI-RS) and / or synchronization signals (SS) or SS blocks for 3GPP networks (e.g., LTE or 5G / NR) and RSRP, RSSI, RSRQ, RCPI, RSNI and / or ANPI measurements of various beacons, fast initial link setup (FILS) discovery frames or probe response frames for WLAN / WiFi (e.g., [IEEE80211]) networks. Other measurements may be used in addition or alternatively, such as those discussed in 3GPP TS 36.214 v17.0.0 (2022-03-31) (“[TS36214]”), 3GPP TS 38.215 v17.3.0 (2023-03-30) (“[TS38215]”), 3GPP TS 38.314 v17.2.0 (2023-01-13) (“[TS38314]”), IEEE Standard for Information Technology—Telecommunications and Information Exchange between Systems—Local and Metropolitan Area Networks—Specific Requirements—Part 11: Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications, IEEE Std 802.11-2020, pp. 1-4379 (26 Feb. 2021) ("[IEEE 80211]"), etc. Additionally or alternatively, any of the foregoing measurements (or combinations of measurements) may be collected by one or more NANs 1014 and provided to edge computing nodes.

[0162] Additionally or alternatively, the measurements may include one or more of the following measurements: measurements related to data radio bearers (DRBs) (e.g., the number of DRBs attempted to be established, the number of DRBs successfully established, the number of active DRBs released, the intra-session activity time for DRBs, the number of DRBs attempted to be restored, the number of DRBs successfully restored, etc.); measurements related to RRC (e.g., the average number of RRC connections, the maximum number of RRC connections, the average number of inactive RRC connections stored, the maximum number of inactive RRC connections stored, the number of attempted, successful and / or failed RRC connection establishments, etc.); measurements related to UE context (UECNTX); measurements related to radio resource utilization (RRU) (e.g., DL total PRB utilization, UL total PRB utilization, distribution of DL total PRB utilization, distribution of UL total PRB utilization, DL PRBs used for data services, UL PRBs used for data services). PRB, DL total available PRB, UL total available PRB, etc.); measurements related to registration management (RM); measurements related to session management (SM) (e.g., the number of PDU sessions requested to be established, the number of PDU sessions successfully established, the number of PDU sessions that failed to be established, etc.); measurements related to GTP management (GTP); measurements related to IP management (IP); measurements related to policy association (PA); measurements related to mobility management (MM) (e.g., for hetero-RAT, homo-RAT and / or homo-frequency / hetero-frequency handover and / or conditional handover): the requested , number of successful and / or failed handover preparations; number of requested, successful and / or failed handover resource allocations; number of requested, successful and / or failed handover executions; average and / or maximum time of requested handover executions; number of successful and / or failed handover executions per beam pair, etc.); measurements related to virtualized resources (VR); measurements related to the operator (CARR); measurements related to QoS flows (QF) (e.g., number of active QoS flows released, number of QoS flows attempted to be released, session activity time for QoS flows, session activity time for UE 1002, the number of QoS flows attempted to be established, the number of QoS flows successfully established, the number of QoS flows that failed to be established, the number of initial QoS flows attempted to be established, the number of initial QoS flows successfully established, the number of initial QoS flows that failed to be established, the number of QoS flows attempted to be modified, the number of QoS flows successfully modified, the number of QoS flows that failed to be modified, etc.); measurements related to application trigger (AT); measurements related to short message service (SMS); measurements related to power, energy and environment (PEE); measurements related to NF service (NFS); measurements related to packet flow description (PFD); measurements related to random access channel (RACH); measurements related to measurement report (MR);Measurements related to Layer 1 Measurements (L1M); measurements related to Network Slice Selection (NSS); measurements related to Paging (PAG); measurements related to Non-IP Data Delivery (NIDD); measurements related to External Parameter Provisioning (EPP); measurements related to Traffic Impact (TI); measurements related to Connection Establishment (CE); measurements related to Service Parameter Provisioning (SPP); measurements related to Background Data Delivery Policy (BDTP); measurements related to Data Management (DM); and / or any other performance measurements (e.g., performance measurements discussed in 3GPP TS 28.552 v18.2.0 (2023-03-30) (“[TS28552]”), 3GPP TS 32.425 v17.1.0 (2021-06-24) (“[TS32425]”), etc.). ;

[0163] Radio information may be reported in response to a trigger event and / or in a periodic manner. Additionally or alternatively, each UE 1002 reports radio information with low or high periodicity depending on the data transmission to occur and / or other information about the data transmission. Additionally or alternatively, the edge computing node may request measurements from the NAN 1014 with low or high periodicity, or the NAN 1014 may provide measurements to the edge computing node with low or high periodicity. Additionally or alternatively, the edge computing node may obtain other relevant data (e.g., key performance indicators (KPIs)) from other edge computing nodes, core network functions (NFs), application functions (AFs), and / or other UEs 1002 together with or separately from the measurement report.

[0164] Additionally or alternatively, in the event that there are discrepancies in the observed data from one or more UEs, one or more RAN nodes and / or core network NFs (e.g., reports are missing, data is erroneous, etc.), simple imputation may be performed to supplement the observed data obtained (e.g., replacing values ​​from previous reports and / or historical data, applying extrapolation filtering, etc.). Additionally or alternatively, acceptable boundaries for observed data may be predetermined or configured. For example, CQI and MCS measurements may be configured to be within a range defined only by an appropriate 3GPP standard. In the event that the reported data values ​​are meaningless (e.g., values ​​are outside an acceptable range / boundary, etc.), such values ​​may be discarded for the current learning / training episode or cycle. For example, a packet delivery delay boundary may be defined or configured, and packets determined to be received only after the packet delivery delay boundary may be discarded.

[0165] UE 1002 may also perform reference signal (RS) measurement and reporting procedures to provide the network with information about the quality of one or more wireless channels and / or the general communication medium, and this information may be used to optimize various aspects of the communication system. As an example, the measurement and reporting procedures performed by UE 1002 may include those discussed in 3GPP TS 38.211 v17.4.0 (2023-01-04) (“[TS38211]”), 3GPP TS 38.212 v17.5.0 (2023-03-30) (“[TS38212]”), 3GPP TS 38.213 v17.5.0 (2023-03-30) (“[TS38213]”), 3GPP TS 38.214 v17.5.0 (2023-03-30) (“[TS38214]”), 3GPP TS 38.215 v17.5.0 (2023-03-30) (“[TS38216]”), 3GPP TS 3GPP TS 38.101-1v18.1.0 (2023-04-07) (“[TS38101-1]”), 3GPP TS 38.104v18.1.0 (2023-04-07) (“[TS38104]”), 3GPP TS 38.133v18.1.0 (2023-04-07) (“[TS38133]”), [TS38331], etc. The physical signal and / or reference signal may include a demodulation reference signal (DM-RS), a phase tracking reference signal (PT-RS), a positioning reference signal (PRS), a channel state information reference signal (CSI-RS), a synchronization signal block (SSB), a primary synchronization signal (PSS), a secondary synchronization signal (SSS), and a sounding reference signal (SRS).

[0166] In any of the examples discussed herein, any suitable data collection and / or measurement mechanism may be used to collect observation data. For example, data tagging (e.g., sequence numbering, etc.), packet tracking, signal measurement, data sampling, and / or timestamping techniques may be used to determine any of the above metrics / observations. The collection of data may be based on the occurrence of an event that triggers the collection of data. Additionally or alternatively, data collection may occur at the initiation or termination of an event. Data collection may be continuous, discontinuous, and / or have a start and stop time. Data collection techniques / mechanisms may be specific to HW configuration / implementation, or may be non-HW specific, or may be based on various software parameters (e.g., OS type and version, etc.). Various configurations may be used to define any of the above data collection parameters. Such configurations may be provided by appropriate specifications / standards (e.g., 3GPP (e.g., [SA6Edge]), ETSI (e.g., [MEC]), O-RAN (e.g., [O-RAN]), Smart Edge Open (formerly OpenNESS) (e.g., [ISEO]), IETF (e.g., MAMS [RFC8743]), IEEE / WiFi (e.g., [IEEE80211], etc.), and / or any other similar standards (e.g., the standards discussed in this article)).

[0167] In some examples, the RAN 1004 is an E-UTRAN having one or more eNBs and provides an LTE air interface (Uu) having parameters and characteristics as discussed in at least 3GPP TS 36.300 v17.2.0 (2022-09-30) ("[TS36300]"). In some examples, the RAN 1004 is a next generation (NG)-RAN 1004 having a set of RAN nodes 1014 (including gNBs 1014a and ng-eNBs 1014b). Each gNB 1014a connects to a 5G-enabled UE 1002 using a 5G-NR Uu interface having parameters and characteristics as discussed in [TS38300] (in addition to many other 3GPP standards, including any standards discussed herein). In the case where the NG-RAN 1004 includes a group of ng-eNBs 1014b, one or more ng-eNBs 1014b are connected to the UE 1002 via a 5G Uu and / or LTE Uu interface. The gNB 1014a and the ng-eNB 1014b are connected to the 5GC 1040 via corresponding NG interfaces including N2 interfaces, N3 interfaces and / or other interfaces. The gNB 1014a and the ng-eNB 1014b are connected to each other via an Xn interface. Additionally, each gNB 1014a is connected to each other via a corresponding Xn interface, and each ng-eNB 1014b is connected to each other via a corresponding Xn interface. In some examples, the NG interface can be divided into two parts: an NG user plane (NG-U) interface, which carries service data between the nodes of the NG-RAN 1004 and the UPF 1048 (e.g., the N3 interface); and an NG control plane (NG-C) interface, which is a signaling interface between the nodes of the NG-RAN 1004 and the AMF 1044 (e.g., the N2 interface).

[0168] NG-RAN 1004 may provide a 5G-NR air interface (which may also be referred to as a Uu interface) with the following characteristics: variable SCS; CP-OFDM for DL, CP-OFDM and DFT-s-OFDM for UL; polarization codes, repetition codes, simplex codes and Reed-Muller codes for control, and LDPC for data. Similar to the LTE air interface, the 5G-NR air interface may rely on CSI-RS, PDSCH / PDCCH DMRS. The 5G-NR air interface may not use CRS, but may use: PBCH DMRS for PBCH demodulation; PTRS for phase tracking of PDSCH; and tracking reference signals for time tracking. The 5G-NR air interface may operate on a FR1 band including a sub-6 GHz band or a FR2 band including a band from 24.25 GHz to 52.6 GHz. The 5G-NR air interface may include an SSB, which is an area in the DL resource grid including PSS / SSS / PBCH.

[0169] The 5G-NR air interface can utilize BWP for various purposes. For example, BWP can be used for dynamic adaptation of SCS. For example, UE 1002 can be configured with multiple BWPs, each BWP configuration with a different SCS. When the BWP change is indicated to UE 1002, the SCS of the transmission also changes. Another use case example of BWP is related to power saving. Specifically, UE 1002 can be configured with multiple BWPs with different amounts of frequency resources (e.g., PRBs) to support data transmission under different traffic load scenarios. A BWP containing a smaller number of PRBs can be used for data transmission with a small traffic load while allowing power saving at UE 1002 and in some cases at gNB 1014a. A BWP containing a larger number of PRBs can be used in scenarios with higher traffic loads.

[0170] In some implementations, each gNB 1014a may include a gNB-CU and a set of gNB-DUs. Additionally or alternatively, the gNB 1014a may include one or more RUs. In these implementations, the gNB-CU may be connected to each gNB-DU via a corresponding F1 interface. In the case of a shared network with multiple cell ID broadcasts, each cell identity associated with a PLMN subset corresponds to a gNB-DU, and the gNB-CU to which it is connected shares the same physical layer cell resources. For resiliency, the gNB-DU may be connected to multiple gNB-CUs through appropriate implementations. In addition, the gNB-CU may be separated into gNB-CU control plane (gNB-CU-CP) and gNB-CU user plane (gNB-CU-UP) functions. The gNB-CU-CP is connected to the gNB-DU via the F1 control plane interface (F1-C), the gNB-CU-UP is connected to the gNB-DU via the F1 user plane interface (F1-U), and the gNB-CU-UP is connected to the gNB-CU-CP via the E1 interface. In some implementations, a gNB-DU is connected to only one gNB-CU-CP, and a gNB-CU-UP is connected to only one gNB-CU-CP. For resiliency, a gNB-DU and / or gNB-CU-UP may be connected to multiple gNB-CU-CPs by appropriate implementation. A gNB-DU may be connected to multiple gNB-CU-UPs under the control of the same gNB-CU-CP, and a gNB-CU-UP may be connected to multiple DUs under the control of the same gNB-CU-CP. Xn-U may support data forwarding between gNB-CU-UPs during intra-gNB-CU-CP handover within a gNB. Similarly, each ng-eNB 1014b may include an ng-eNB-CU and a set of ng-eNB-DUs. In these implementations, the ng-eNB-CU and each ng-eNB-DU are connected to each other via a corresponding W1 interface. The ng-eNB may include an ng-eNB-CU-CP, one or more ng-eNB-CU-UPs, and one or more ng-eNB-DUs. The ng-eNB-CU-CP and ng-eNB-CU-UP are connected via the E1 interface. The ng-eNB-DU is connected to the ng-eNB-CU-CP via the W1-C interface and to the ng-eNB-CU-UP via the W1-U interface. If not explicitly specified otherwise, the general principles described in this document with respect to the gNB aspect also apply to the ng-eNB aspect and the corresponding E1 and W1 interfaces.

[0171] The node hosting the user plane part of the PDCP protocol layer (e.g., gNB-CU, gNB-CU-UP, and for EN-DC, MeNB or SgNB, depending on the bearer offload) performs user inactivity monitoring and further notifies the node having the control plane connection towards the core network (e.g., via E1, X2, etc.) of its inactivity or (re)activation. The node hosting the RLC protocol layer (e.g., gNB-DU) may perform user inactivity monitoring and further notifies the node hosting the control plane (e.g., gNB-CU or gNB-CU-CP) of its inactivity or (re)activation.

[0172] In these implementations, the NG-RAN 1004 is layered into a Radio Network Layer (RNL) and a Transport Network Layer (TNL). The NG-RAN 1004 architecture (e.g., NG-RAN logical nodes and the interfaces between them) is part of the RNL. For each NG-RAN interface (e.g., NG, Xn, F1, etc.), relevant TNL protocols and functions are specified. The TNL provides services for user plane transport and / or signaling transport. In an NG-Flex configuration, each NG-RAN node is connected to all AMFs 1044 within an AMF area that support at least one AMF set that is also supported by the NG-RAN node. AMF sets and AMF areas are defined in [TS23501].

[0173] The RAN 1004 is communicatively coupled to the CN 1040, which includes network elements and / or network functions (NFs) for providing various functions supporting data and telecommunication services to customers / subscribers (e.g., UE 1002). The components of the CN 1040 may be implemented in one physical node, or in separate physical nodes. In some examples, NFV may be utilized to virtualize any or all functions provided by the network elements of the CN 1040 onto physical computing / storage resources in servers, switches, etc. A logical instantiation of the CN 1040 may be referred to as a network slice, and a logical instantiation of a portion of the CN 1040 may be referred to as a network sub-slice.

[0174] exist Fig.10In the example of , CN 1040 is 5GC 1040, 5GC 1040 includes authentication server function (AUSF) 1042, access and mobility management function (AMF) 1044, session management function (SMF) 1046, user plane function (UPF) 1048, network slice selection function (NSSF) 1050, network open function (NEF) 1052, network repository function (NRF) 1054, policy control function (PCF) 1056, unified data management (UDM) 1058, unified data repository (UDR), application function (AF) 1060 and network data analysis function (NWDAF) 1062, which are coupled with each other through various interfaces shown. The following briefly introduces the NF in 5GC 1040.

[0175] The NWDAF 1062 includes one or more of the following functions: support for data collection from NF and AF 1060; support for data collection from OAM; NWDAF service registration and metadata openness to NF and AF 1060; support for analysis information provisioning to NF and AF 1060; support for machine learning (ML) model training and provisioning to NWDAF 1062 (e.g., NWDAF containing analysis logic functions). Some or all NWDAF functions can be supported in a single instance of NWDAF 1062. NWDAF 1062 also includes analysis reporting capabilities, which include means to allow discovery of types of analysis that can be consumed by external parties and / or requests to consume analysis information generated by NWDAF 1062.

[0176] The NWDAF 1062 interacts with different entities for different purposes, such as one or more of the following: data collection based on subscriptions to events provided by the AMF 1044, SMF 1046, PCF 1056, UDM 1058, NSACF, AF 1060 (directly or via NEF 1052), and OAM (not shown); using a Data Collection Coordination Function (DCCF) (e.g., Figure 1 , Figure 2 and Figure 3 DCCF 1063 in the system) performs analysis and data collection; obtains information from the data repository (e.g., obtains subscriber-related information from the UDR via UDM1058); performs data collection of location information from the LCS system; obtains data from the analysis data repository function (ADRF) (e.g., Figure 4 ADRF 1066 in the storage and retrieval of information; from the message transmission framework adapter function (MFAF) (for example, Figure 1 , Figure 2 and Figure 31065); perform analysis and data collection; obtain information about the NF (e.g., obtain NF-related information from the NRF 1054); provide analysis to consumers on demand, as specified in clause 6 of [TS23288]; and / or provide batch data associated with an analysis ID. The NWDAF discovery and selection process is discussed in clause 6.3.13 of [TS23501] and clause 5.2 of [TS23288].

[0177] A single instance or multiple instances of NWDAF 1062 may be deployed in a PLMN. If multiple NWDAF 1062 instances are deployed, the architecture supports deploying NWDAF 1062 as a central NF, a collection of distributed NFs, or a combination of the two. If multiple NWDAF 1062 instances are deployed, NWDAF 1062 may act as an aggregation point (e.g., an aggregator NWDAF 1062) and collect analysis information from other NWDAFs 1062 that may have different service areas to generate aggregated analysis (e.g., by analysis ID) (possibly with self-generated analysis). When there are multiple NWDAFs 1062, not all NWDAFs need to be able to provide the same type of analysis results. For example, some NWDAFs 1062 may be dedicated to providing a specific type of analysis. The analysis ID information element is used to identify the type of supported analysis that NWDAF 1062 can generate. In some implementations, NWDAF 1062 instances may be co-located with 5GSNFs. Additional aspects of the NWDAF 1062 functionality are defined in 3GPP TS 23.288 v18.1.0 (2023-03-31) (“[TS23288]”).

[0178] Different NWDAF 1062 instances may exist in the 5GC 1040, possibly specialized by type of analysis. The capabilities of the NWDAF 1062 instances are described in the NWDAF profile stored in the NRF 1054. The NWDAF architecture allows multiple NWDAF 1062 instances to be arranged in a hierarchy / tree with a flexible number of layers / branches. The number and organization of hierarchy layers and the capabilities of each NWDAF 1062 instance remain deployment choices and may vary depending on the implementation and / or use case. In a hierarchical deployment, when DCCF 1063 and / or MFAF 1065 are not present in the network, the NWDAF 1062 may provide data collection open capabilities for generating analytics based on data collected by other NWDAFs 1062.

[0179] The AUSF 1042 stores data for authenticating the UE 1002 and handles authentication related functions. The AUSF 1042 may facilitate a common authentication framework for various access types.

[0180] AMF 1044 allows other functions of 5GC 1040 to communicate with UE 1002 and RAN 1004 and subscribe to notifications about mobility events for UE 1002. AMF 1044 is also responsible for registration management (e.g., for registering UE 1002), connection management, reachability management, mobility management, and legal interception of AMF-related events, as well as access authentication and authorization. AMF 1044 provides transmission of SM messages between UE 1002 and SMF 1046 and acts as a transparent agent for routing SM messages. AMF 1044 also provides transmission of SMS messages between UE 1002 and SMSF. AMF 1044 interacts with AUSF 1042 and UE1002 to perform various security anchoring and context management functions. In addition, AMF 1044 is the termination point of the RAN-CP interface, which includes the N2 reference point between RAN 1004 and AMF 1044. AMF 1044 is also the termination point for NAS (N1) signaling and performs NAS encryption and integrity protection.

[0181] AMF 1044 also supports NAS signaling with UE 1002 over the N3IWF interface. N3IWF provides access to untrusted entities. N3IWF can be the termination point for the N2 interface between (R)AN 1004 and AMF 1044 (for the control plane) and can be the termination point for the N3 reference point between (R)AN 1004 and 1048 (for the user plane). Therefore, AMF1044 handles N2 signaling from SMF 1046 and AMF 1044 for PDU sessions and QoS, encapsulates / decapsulates packets for IPSec and N3 tunnels, marks N3 user plane packets in UL, and implements QoS corresponding to N3 packet marking taking into account the QoS requirements associated with such marking received over N2. The N3IWF may also relay UL and DL control plane NAS signaling between the UE 1002 and the AMF 1044 via the N1 reference point between the UE 1002 and the AMF 1044, and relay UL and DL user plane packets between the UE 1002 and the UPF 1048. The N3IWF also provides a mechanism for IPsec tunnel establishment with the UE 1002. The AMF 1044 may expose an interface based on Namf services and may be used for the N14 reference point between the two AMFs 1044 and the AMF 1044 and the 5G-EIR ( Fig.10 In addition to the functions of AMF 1044 described herein, AMF 1044 may also provide support for network slice restrictions and network slice instance restrictions based on NWDAF analysis.

[0182] SMF 1046 is responsible for SM (e.g., session establishment, tunnel management between UPF 1048 and AN 1008); UE IP address allocation and management (including optional authorization); selection and control of UP functions; configuration of service steering at UPF 1048 to route services to the correct destination; termination of interfaces toward policy control functions; control of policy enforcement, charging, and part of QoS; lawful interception (for SM events and interface to LI system); termination of the SM portion of NAS messages; DL data notification; initiation of AN-specific SM information sent to AN 1008 via AMF 1044 over N2; and determination of the SSC mode for the session. SM refers to the management of PDU sessions, while PDU sessions or "sessions" refer to PDU connection services that provide or enable the exchange of PDUs between UE 1002 and DN 1036. SMF 1046 may also include the following functions to support edge computing enhancements (see, e.g., [TS23548]): select EASDF 1061 and provide its address to the UE as a DNS server for the PDU session; use the EASDF 1061 services defined in [TS23548]; and support the application layer architecture defined in [TS23558] to provision and update ECS address configuration information to the UE. The discovery and selection process for EASDF 1061 is discussed in [TS23501] §6.3.23.

[0183] The UPF 1048 acts as an anchor point for intra-RAT and inter-RAT mobility, an external PDU session point for interconnection to the data network 1036, and a branch point for supporting multi-homed PDU sessions. The UPF 1048 also performs packet routing and forwarding, packet inspection, implements the user plane portion of policy rules, lawful interception of packets (UP collection), performs service usage reporting, performs QoS processing for the user plane (e.g., packet filtering, gating, UL / DL rate enforcement), performs UL service verification (e.g., SDF to QoS flow mapping), transport level packet marking in UL and DL, and performs DL packet buffering and DL data notification triggering. The UPF 1048 may include a UL classifier to support routing of service flows to the data network.

[0184] The NSSF 1050 selects a set of network slice instances to serve the UE 1002. The NSSF 1050 also determines the allowed NSSAI and the mapping to the subscribed S-NSSAI (if necessary). The NSSF 1050 also determines the AMF set or list of candidate AMFs 1044 to be used to serve the UE 1002 based on appropriate configuration and possibly by querying the NRF 1054. The selection of a set of network slice instances for the UE 1002 can be triggered by the AMF 1044 to which the UE 1002 is registered, by interacting with the NSSF 1050, which may result in a change of the AMF 1044. The NSSF 1050 interacts with the AMF 1044 via the N22 reference point; and may communicate with another NSSF in the visited network via the N31 reference point (not shown).

[0185] NEF 1052 securely exposes services and capabilities provided by 3GPP NFs to third parties, internal opening / re-opening, AF 1060, edge computing networks / frameworks, etc. In these examples, NEF 1052 can authenticate, authorize, or restrict AF 1060. NEF 1052 uses the Nudr interface to the unified data repository (UDR) to store / retrieve information as structured data. NEF 1052 can also convert information exchanged with AF 1060 and information exchanged with internal NFs. For example, NEF 1052 can convert between AF service identifiers and internal 5GC information (e.g., DNN, S-NSSAI) as described in clause 5.6.7 of [TS23501]. Specifically, NEF 1052 handles the shielding of network and user sensitive information of external AF 1060 according to network policies. NEF 1052 also receives information from other NFs based on the capabilities of the other NFs that are exposed. This information may be stored at the NEF 1052 as structured data, or at a data storage NF using standardized interfaces. The stored information may then be reopened by the NEF 1052 to other NFs and AFs, or used for other purposes (e.g., analysis). For example, NWDAF analysis may be securely opened to external parties by the NEF 1052, as specified in [TS23288]. In addition, data provided by external parties may be collected by the NWDAF 1062 via the NEF 1052 for analysis generation purposes. The NEF 1052 processes and forwards requests and notifications between the NWDAF 1062 and the AF 1060, as specified in [TS23288].

[0186] NRF 1054 supports service discovery functionality, receives NF discovery requests from NF instances, and provides information of discovered NF instances to requesting NF instances. NRF 1054 also maintains NF profiles of available NF instances and their supported services. The NF profile of the NF instance maintained in the NRF 1054 includes the following information: NF instance ID; NF type; PLMN ID (in case of PLMN), PLMN ID+NID (in case of SNPN); network slice related identifiers (e.g., S-NSSAI, NSI ID); network address of the NF (e.g., FQDN, IP address, etc.), NF capacity information, NF priority information (e.g., for AMF selection), NF set ID, NF service set ID of the NF service instance; NF specific service authorization information; names of supported services (if applicable); endpoint addresses of instances of each supported service; identification of the data / information stored (e.g., for UDR profile and / or other NF profiles); other service parameters (e.g., DNN or DNN list, LADN DNN or LADN DNN list, notification endpoint for each type of notification that the NF service is interested in receiving, etc.); location information for the NF instance (e.g., geographical location, data center, etc.); TAI; NF load information; routing indicator, home network public key identifier (for UDM 1058 and AUSF 1042); for UDM 1058, AUSF 1042, and NSSAAF, in the case of accessing the SNPN using credentials owned by the credential holder with an AAA server, the identity of the credential holder (e.g., a security realm of a SUPI based on a network-specific identifier); for UDM 1058 and AUSF 1042, and if UDM 1058 / AUSF 1042 is used to access the SNPN using credentials owned by the credential holder (e.g., a security realm if a SUPI based on a network-specific identifier is used, or MCC and MNC if a SUPI based on an IMSI is used); for AUSF 1042 and NSSAAF, in the case of SNPN access using a DCS with an AAA server, the identity of the DCS (e.g., a security realm of a SUPI based on a network-specific identifier); for UDM 1058 and AUSF 1042, and if UDM 1058 / AUSF 1042 is used to access the SNPN using credentials owned by the credential holder 1042 is used as DCS (in case of SNPN access), the identity of the DCS (e.g., security domain if SUPI is based on network specific identifier, or MCC and MNC if SUPI is based on IMSI); one or more GUAMIs (in case of AMF 1044); for UPF 1048, see clause 5.2.7.2 of [TS23502].2; for UDM 1058, UDM group ID, range of SUPI, range of GPSI, range of internal group identifier, range of external group identifier; for UDR, UDR group ID, range of SUPI, range of GPSI, range of external group identifier; for AUSF 1042, AUSF group ID, range of SUPI; for PCF 1056, PCF group ID, range of SUPI; for HSS, HSS group ID, IMPI set, IMPU set, IMSI set, PSI set, MSISDN set; event IDs supported by AF 1060 (in case of NEF 1052); event IDs supported by event open services of UPF 1048; application identifiers supported by AF 1060 (in case of NEF 1052); range of external identifiers, or range of external group identifiers, or domain names served by NEF (in case of NEF 1052) (for example, when NEF 1052 is used when exposing AF information for analysis purposes, as detailed in [TS23288]); In addition, the NRF 1054 can store mappings between UDM group ID and SUPI, UDR group ID and SUPI, AUSF group ID and SUPI, and PCF group ID and SUPI to enable the use of SUPI and SUPI range to discover UDM 1058, UDR, AUSF 1042 and PCF 1056, as specified in clause 6.3 of [TS23501], and / or interact with the UDR to resolve UDM group ID / UDR group ID / AUSF group ID / PCF group ID based on UE identity (e.g., SUPI); 3GPP List of IP domains, range of (UE) IPv4 addresses or range of (UE) IPv6 prefixes, range of SUPI or range of GPSI or BSF Group ID (in case of BSF) as described in clause 6.1.6.2.21 of TS29.510v18.2.0 (2023-03-29) ("[TS29510]"); SCP domain to which the NF belongs; DCCF service area information, NF type of data source, NF set ID of data source (if available) (in case of DCCF 1063); list of supported DNAIs (in case of SMF 1046); for SNPN, capability to support SNPN onboarding (in case of AMF) and capability to support remote provisioning of user plane (in case of SMF 1046); for UPF 1048, IP address range, DNAI; additional V2X related NF profile parameters defined in 3GPP TS23.287; additional ProSe related NF profile parameters defined in 3GPP TS23.304; additional MBS-related NF profile parameters are defined in 3GPP TS23.247; additional UAS-related NF profile parameters are defined in TS23.256; and many other contents discussed in [TS23501]. In some examples, in cases where, for example, a NF instance has exceptional service authorization information, the service authorization information provided by the OAM system is also included in the NF configuration file. .

[0187] For NWDAF 1062, the NF profile includes: supported analysis IDs (possibly by service), NWDAF service area information (e.g., a list of TAIs for which NWDAF can provide services and / or data), supported analysis delays by analysis ID (if available), NF type of NF data source, NF set ID of NF data source (if available), analysis aggregation capability (if available), analysis metadata provisioning capability (if available), ML model filtering information parameters S-NSSAI and region of interest for trained ML models by analysis ID (if available), federated learning (FL) capability type (e.g., FL server or FL client (if available)), time interval for supporting FL (if available). The service area information of NWDAF 1062 is common to all analysis IDs it supports. Analysis IDs supported by NWDAF 1062 can be associated with supported analysis delays, for example, analysis reports can be generated at a time less than or equal to the supported analysis delay (including data collection delay and inference delay). Determining supported analysis delays and how the NWDAF 1062 avoids frequently updating its supported analysis delays in the NRF may be NWDAF implementation specific.

[0188] The PCF 1056 provides policy rules to the control plane functions to implement them and may also support a unified policy framework to manage network behavior. The PCF 1056 may also implement a front end to access subscription information related to policy decisions in the UDR 1059 of the UDM 1058. In addition to communicating with the functions through reference points as shown, the PCF 1056 also exposes an interface based on the Npcf service.

[0189] UDM 1058 processes subscription-related information to support the processing of communication sessions by network entities, and stores subscription data of UE 1002. For example, subscription data can be transferred via the N8 reference point between UDM 1058 and AMF 1044. UDM 1058 may include two parts: application front end and UDR. UDR may store subscription data and policy data for UDM 1058 and PCF 1056, and / or structured data for openness and application data for NEF 1052 (including PFD for application detection, application request information for multiple UEs 1002). UDR may expose Nudr service-based interfaces to allow UDM 1058, PCF 1056, and NEF 1052 to access a specific set of stored data, as well as read, update (e.g., add, modify), delete, and subscribe to notifications of changes in related data in UDR. UDM 1058 may include UDM-FE, which is responsible for handling credentials, location management, subscription management, etc. Several different front ends can serve the same user in different transactions. The UDM-FE accesses subscription information stored in the UDR and performs authentication credential processing, user identity processing, access authorization, registration / mobility management, and subscription management. In addition to communicating with other NFs through reference points as shown, the UDM 1058 can also expose interfaces based on Nudm services.

[0190] The Edge Application Server Discovery Function (EASDF) 1061 exposes an interface based on the Neasdf service and connects to the SMF 1046 via the N88 interface. One or more EASDF instances may be deployed within the PLMN, and the interaction between the 5GC NF and the EASDF 1061 occurs within the PLMN. The EASDF 1061 includes one or more of the following functions: registering with the NRF 1054 for EASDF 1061 discovery and selection; processing DNS messages according to instructions from the SMF 1046; and / or terminating DNS security (if used). Processing DNS messages according to instructions from SMF 1046 includes one or more of the following functions: receiving DNS message processing rules and / or BaselineDNSPattern from SMF1046; exchanging DNS messages from / with UE 1002; forwarding DNS messages to C-DNS or L-DNS for DNS queries; adding EDNS Client Subnet (ECS) options to DNS queries about FQDNs; reporting information related to received DNS messages to SMF 1046; and / or buffering / discarding DNS messages from UE1002 or DNS servers. EASDF has a direct user plane connection with PSA UPF through N6 (e.g., without any NAT) for transporting DNS signaling exchanged with the UE. Deployment of NAT between EASDF 1061 and PSA UPF 1048 may or may not be supported. Additional aspects of EASDF 1061 are discussed in [TS23548].

[0191] AF 1060 provides application influence on service routing, provides access to NEF 1052, and interacts with the policy framework for policy control. AF 1060 can influence UPF 1048 (re)selection and service routing. Based on operator deployment, when AF 1060 is considered a trusted entity, the network operator can allow AF 1060 to interact directly with the relevant NF. In some implementations, AF 1060 can be used for edge computing implementations.

[0192] NFs that need to collect data from AF 1060 can subscribe / unsubscribe to notifications about data collected from AF 1060 directly from AF 1060 or via NEF 1052. The data collected from AF 1060 is used as input for analysis performed by NWDAF 1062. Details about the data collected from AF 1060 and the interactions between NEF 1052, AF 1060, and NWDAF 1062 are described in [TS23288].

[0193] 5GC 1040 can implement edge computing by selecting the operator / third party service to be geographically close to the point where UE 1002 attaches to the network. This can reduce latency and load on the network. In the edge computing implementation, 5GC 1040 can select UPF 1048 close to UE 1002 and perform the steering of the service from UPF 1048 to DN 1036 via the N6 interface. This can be based on UE subscription data, UE location and information provided by AF 1060, which allows AF 1060 to influence UPF (re)selection and service routing.

[0194] Data network (DN) 1036 can represent various network operator services, Internet access or third-party services that can be provided by one or more servers (including, for example, application (app) / content server 1038). DN 1036 can be a public network external to the operator, a private PDN network, or a packet data network within the operator (for example, for provisioning IMS services). In this example, app server 1038 can be coupled to IMS via S-CSCF or I-CSCF. In some implementations, DN 1036 can represent one or more local area network DNs (LADNs), which are DNs 1036 (or DN names (DNNs)) that can be accessed by UE 1002 in one or more specific areas. Outside these specific areas, UE 1002 cannot access LADN / DN 1036.

[0195] Additionally or alternatively, DN 1036 may be an edge DN 1036, which is a (local) DN that supports an architecture for implementing edge applications. In these examples, app server 1038 may represent a physical hardware system / device that provides app server functionality and / or application software that performs server functionality and resides in the cloud or at an edge computing node. In some examples, app / content server 1038 provides an edge hosting environment that provides the support required for edge application server execution.

[0196] In some examples, 5GS may use one or more edge computing nodes to provide an interface and offload the processing of wireless communication services. In these examples, the edge computing nodes may be included in one or more RAN 1004 or RAN nodes 1014, or co-located with them. For example, the edge computing node may provide a connection between the RAN 1004 and the UPF 1048 in the 5GC 1040. The edge computing node may use one or more NFV instances instantiated on a virtualized infrastructure within the edge computing node to handle wireless connections to and from the RAN 1014 and the UPF 1048.

[0197] In some implementations, the edge computing node provides a distributed computing environment for application and service hosting, and also provides storage resources and processing resources so that data and / or content can be processed in close proximity to the subscriber (e.g., the user of UE 1002) for faster response time. The edge computing node also supports multi-tenancy runtimes and hosting environments for applications (including virtual device applications that can be delivered as encapsulated virtual machine (VM) images, middleware applications and infrastructure services, content delivery services (including content caching, mobile big data analysis, and computing offloading), etc.). Computing offloading involves: offloading computing tasks, workloads, applications, and / or services from UE 1002, CN 1040, DN 1036, and / or servers 1038 to edge computing nodes, and vice versa. For example, a device application or client application operating in UE 1002 can offload application tasks or workloads to one or more edge computing nodes. In another example, an edge computing node can offload application tasks or workloads to a group of UE 1002 (e.g., for distributed machine learning computing, etc.).

[0198] The edge computing node may include, or be part of, an edge system that employs one or more edge computing technologies (ECTs) (also referred to as "edge computing frameworks," etc.). An edge computing node may also be referred to as an "edge host" or "edge server." The edge system includes a collection of edge servers and edge management systems (not shown) necessary to run edge computing applications within an operator network or a subset of an operator network. An edge server is a physical computer system that may include an edge platform and / or a virtualized infrastructure and provides computing resources, storage resources, and network resources to edge computing applications. Each edge server is disposed at the edge of a corresponding access network and is arranged to provide computing resources and / or various services (e.g., computing tasks and / or workload offloading, cloud computing capabilities, IT services, and other similar resources and / or services discussed herein) in relatively close proximity to UE 1002. The VI of the edge computing node provides a virtualized environment and virtualized resources for the edge host, and the edge computing application can run on top of the VI as a VM and / or application container.

[0199] In one example implementation, the ECT is a MEC framework and / or operates according to a MEC framework, such as those discussed in: ETSI GR MEC 001 v3.1.1 (2022-01), ETSI GS MEC 003 v3.1.1 (2022-03), ETSI GSMEC 009 v3.1.1 (2021-06), ETSI GS MEC 010-1 v1.1.1 (2017-10), ETSI GS MEC 010-2 v2.2.1 (2022-02), ETSI GS MEC 011 v2.2.1 (2020-12), ETSI GS MEC 012 V2.2.1 (2022-02), ETSI GS MEC 013 V2.2.1 (2022-01), ETSI GS MEC 014v2.1.1 (2021-03), ETSI GS MEC 015v2.1.1 (2020-06), ETSI GS MEC 016v2.2.1 (2020-04), ETSI GS MEC 021v2.2.1 (2022-02), ETSI GR MEC 024v2.1.1 (2019-11), ETSI GS MEC 028v2.2.1 (2021-07), ETSI GS MEC 029v2.2.1 (2022-01), ETSI GS MEC 030v2.1.1 (2020-04) and ETSI GR MEC 031v2.1.1 (2020-10) (collectively referred to herein as "[MEC]"), the contents of each of which are hereby incorporated by reference in their entirety.This example implementation (and / or any other example implementation discussed herein) may also include NFV and / or other similar virtualization technologies, such as those discussed in: ETSI GR NFV 001 V1.3.1 (2021-03), ETSI GS NFV 002 V1.2.1 (2014-12), ETSI GR NFV 003 V1.6.1 (2021-03), ETSI GS NFV 006 V2.1.1 (2021-01), ETSI GS NFV-INF001 V1.1.1 (2015-01), ETSI GS NFV-INF 003 V1.1.1 (2014-12), ETSI GS NFV-INF004 V1.1.1 (2015-01), ETSI GS NFV-MAN 001v1.1.1 (2014-12) and / or OSM Release FIVE Technical Overview by Israel et al., ETSI Open Source MANO, OSM White Paper, 1st ed. (Jan. 2019), https: / / osm.etsi.org / images / OSM-Whitepaper-TechContent-ReleaseFIVE-FINAL.pdf (collectively referred to as “[ETSI NFV]”), the contents of each of which are hereby incorporated by reference in their entirety. Other virtualization technologies and / or service orchestration and automation platforms may be used, such as those discussed in: E2E Network Slicing Architecture, GSMA, Official Doc.NG.127, v1.0 (03Jun.2021), https: / / www.gsma.com / newsroom / wp-content / uploads / / NG.127-v1.0-2.pdf, Open Network Automation Platform (ONAP) documentation, Release Istanbul, v9.0.1 (17Feb.2022), https: / / docs.onap.org / en / latest / index.html (“[ONAP]”), 3GPP Service-Based Management Architecture (SBMA) discussed in 3GPP TS28.533 v17.1.0 (2021-12-23) (“[TS28533]”), the contents of each of which are hereby incorporated by reference in their entirety.

[0200] In another example implementation, the ECT is an ORAN framework and / or operates according to an ORAN framework. Typically, front-end and back-end equipment vendors and operators have worked closely to ensure compatibility. The flip side of this working mode is that plug-and-play with other devices becomes difficult and this can hinder innovation. To overcome this problem and promote openness and interoperability at every level, several key players interested in the wireless space (e.g., operators, equipment manufacturers, academic institutions, etc.) formed the Open RAN Alliance ("O-RAN") in 2018. The O-RAN network architecture is a building block for designing a virtualized RAN on programmable hardware, where radio access control is supported by AI / ML.Various aspects of the O-RAN architecture are described in the following: O-RAN Working Group 1 (Use Cases and Overall Architecture): O-RAN Architecture Description, O-RAN Alliance WG1, O-RAN Architecture Description v08.00, Release R003 (March 2023); O-RAN Operations and Maintenance Architecture Specification v04.00, O-RAN Alliance WG1 (February 2021); O-RAN Working Group 2 AI / ML Workflow Description and Requirements v01.03 O-RAN Alliance WG2 (October 2021); O-RAN Working Group 2 (non-RT RIC and A1 Interface WG): R1 Interface: General Aspects and Principles 4.0, v04.00, Release R003 (March 2023); O-RAN Working Group 2 (non-RT RIC and A1 Interface WG): non-RT RIC Architecture v02.01 (October 2022); O-RAN Working Group 3 (Near Real-Time RAN Intelligent Controller and E2 Interface Working Group): near-RT RIC Architecture, v04.00, Release R003 (March 2023); O-RAN Working Group 4 (Open Fronthaul Interface WG) Control, User and Synchronization Plane Specification, v11.00, Release R003 (March 2023); O-RAN Fronthaul Working Group 4 Collaborative Transport Interface Transport Control Plane Specification, v03.00 (October 2022); O-RAN Fronthaul Working Group 4 Collaborative Transport Interface Transport Management Plane Specification, v11.00, Release R003 (March 2023); O-RAN Open X-haul Transport Working Group Management Interface for Transport Network Elements, v05.00, Release R003 (March 2023); O-RAN Open Transport Working Group 9Xh aul Packet Switching Architecture and Solutions, v03.00, Release R003 (March 2023); O-RAN Open X-haul Transport Working Group Synchronized Architecture and Solutions Specification, v03.00 (October 2022); O-RAN Open X-haul Transport WG9 WDM-based Fronthaul Transport, v03.00, Release R003 (March 2023); O-RAN Operation and Maintenance Architecture, v08.00, Release R003 (March 2023); O-RAN Operation and Maintenance Interface Specification, v09.00, Release R003 (March 2023) (collectively referred to as "[O-RAN]"), the contents of each of which are hereby incorporated by reference in their entirety.

[0201] In another example implementation, the ECT is and / or operates in accordance with the 3rd Generation Partnership Project (3GPP) System Aspects Working Group 6 (SA6) architecture for implementing edge applications, as discussed in: 3GPP TS 23.558 v18.1.0 (2022-12-23) (“[TS23558]”), 3GPP TS 23.501 v18.0.0 (2022-12-21) (“[TS23501]”), 3GPP TS 23.502 v18.1.1 (2023-04-05) (“[TS23502]”), 3GPP TS 23.548 v18.1.0 (2023-04-06) (“[TS23548]”), 3GPP TS28.538v18.2.0(2023-03-30)("[TS28538]"), 3GPP TR 23.700-98v18.0.0(2022-12-23)("[TR23700-98]"), 3GPP TS 23.222v18.0.0(2022-12-23)("[TS23222]"), 3GPP TS 33.122v18.0.0(2022-12-16)("[TS33122]"), 3GPPTS29.222v17.1.0(2021-06-25)("[TS29222]"), 3GPP TS29.522v18.0.0(2022-12-16)(“[TS29522]”), 3GPP TS29.122v18.0.0(2022-12-16)(“[TS29122]”), 3GPP TS23.682v17.3.0(2022-06-15)(“[TS23682]”), 3GPP TS23.434v18.3.0(2022-12-23)(“[TS23434]”) and 3GPP TS23.401v18.0.0(2022-12-21) (collectively “[SA6Edge]”), the contents of each of which are hereby incorporated by reference in their entirety.

[0202] In another example implementation, the ECT is Smart Edge Open framework (formerly known as OpenNESS) and / or operations therein, as discussed in the following: Smart Edge Open Developer Guide, Version 21.09 (September 30, 2021), available at: https: / / smart-edge-open.github.io / (“[ISEO]”), the contents of which are hereby incorporated by reference in their entirety.

[0203] In another example implementation, the ECT operates according to a Multiple Access Management Service (MAMS) framework, as discussed in: Multi-Access Management Services (MAMS) by Kanugovi et al., Internet Engineering Task Force (IETF), Request for Comments (RFC) 8743 (March 2020) (“[RFC8743]”), TCP Extensions for Multipath Operation with Multiple Addresses by Ford et al., IETF RFC 8684 (March 2020), Multipath Extensions for QUIC (MP-QUIC) by De Coninck et al., IETF draft-deconinck-quic-multipath-07, IETA, QUIC Working Group (May 3, 2021), User-Plane Protocols for Multiple Access Management Service by Zhu et al., IETF draft-zhu-intarea-mams-user-protocol-09, IETA, INTAREA (March 4, 2020), and Generic Protocols by Zhu et al. Multi-Access (GMA) Convergence Encapsulation Protocols, IETF RFC 9188 (February 2022) (collectively, “[MAMS]”), the contents of each of which are hereby incorporated by reference in their entirety.

[0204] It should be understood that the above edge computing framework / ECT and service deployment examples are only illustrative examples of ECT, and the present disclosure may be applicable to many other or additional edge computing / networking technologies in various combinations and layouts of devices at the edge of a network including the various edge computing networks / systems described herein. In addition, the technology disclosed herein may involve other IoT edge network systems and configurations, and other intermediate processing entities and architectures may also be applicable to the present disclosure. Examples of these edge computing / networking technologies include [MEC]; [O-RAN]; [ISEO]; [SA6Edge]; content delivery network (CDN) (also known as "content distribution network", etc.); mobility service provider (MSP) edge computing and / or mobility as a service (MaaS) provider system (e.g., used in AECC architecture); nebula edge cloud system; fog computing system; micro cloud edge cloud system; mobile cloud computing (MCC) system; central office re-architecture as data center (CORD), mobile CORD (M-CORD) and / or converged multi-access and core (COMAC) system; etc. Furthermore, the techniques disclosed herein may relate to other IoT edge network systems and configurations, and other intermediate processing entities and architectures may also be used for the purposes of the present disclosure.

[0205] The interfaces of 5GC 1040 include reference points and service-based interfaces. Reference points include: N1 (between UE 1002 and AMF 1044), N2 (between RAN 1014 and AMF 1044), N3 (between RAN 1014 and UPF 1048), N4 (between SMF 1046 and UPF 1048), N5 (between PCF 1056 and AF 1060), N6 (between UPF 1048 and DN 1036), N7 (between SMF 1046 and PCF 1056), N8 (between UDM 1058 and AMF 1044), N9 (between two UPFs 1048), N10 (between UDM 1058 and SMF 1046), N11 (between AMF 1044 and SMF 1046), N12 (between AUSF 1042 and AMF 1044), N13 (between AUSF 1042 and UDM 1058), N14 (between two AMFs 1044; not shown), N15 (between PCF 1056 and AMF 1044 in the case of non-roaming scenario, or between PCF 1056 and AMF 1044 in the visited network in the case of roaming scenario), N16 (between two SMFs 1046; not shown), and N22 (between AMF 1044 and NSSF 1050). Fig.10 Other reference points not shown are indicated. Fig.10The service-based representation represents the NF within the control plane that enables other authorized NFs to access its services. The service-based interfaces (SBIs) include: Namf (SBI presented by AMF 1044), Nsmf (SBI presented by SMF 1046), Nnef (SBI presented by NEF 1052), Npcf (SBI presented by PCF 1056), Nudm (SBI presented by UDM 1058), Naf (SBI presented by AF 1060), Nnrf (SBI presented by NRF 1054), Nnssf (SBI presented by NSSF 1050), Nausf (SBI presented by AUSF 1042). You can also use Fig.10 Other service-based interfaces (e.g., Nudr, N5g-eir, and Nudsf) not shown in the figure. In some examples, the NEF 1052 can provide an interface to the edge computing node 1036x, which can be used to handle the wireless connection with the RAN 1014.

[0206] Although Fig.10 Not shown, but the system 1000 may further include NFs not shown (e.g., UDR, unstructured data storage function (UDSF), network slice admission control function (NSACF), network slice specific and independent non-public network (SNPN) authentication and authorization function (NSSAAF), UE radio capability management function (UCMF), 5G equipment identity registrar (5G-EIR), charging function (CHF), time sensitive network (TSN) AF 1060, time sensitive communication and time synchronization function (TSCTSF), DCCF (e.g., Figure 1 , Figure 2 and Figure 3 DCCF 1063 in ), Analytical Data Repository Function (ADRF) (e.g., Figure 4 ADRF1066 in), MFAF (for example, Figure 1 , Figure 2 and Figure 3 MFAF 1065 in [TS23501]), Non-Seamless WLAN Offload Function (NSWOF), Service Communication Proxy (SCP), Security Edge Protection Proxy (SEPP), Non-3GPP Interworking Function (N3IWF), Trusted Non-3GPP Gateway Function (TNGF), Wired Access Gateway Function (W-AGF) and / or Trusted WLAN Interworking Function (TWIF), as discussed in [TS23501]).

[0207] Fig.11A wireless network 1100 is schematically shown. The wireless network 1100 includes a UE 1102 that wirelessly communicates with a NAN 1104. The UE 1102 may be the same or similar to any other UE discussed herein (e.g., UE 1002, hardware resources 1200, and / or any other UE discussed herein), and may be substantially interchangeable. The AN 1104 may be the same or similar to any AN (network access node (NAN)) discussed herein (e.g., AP 1006, NAN 1014, RAN 1004, hardware resources 1200, and / or any other AN / NAN discussed herein), and may be substantially interchangeable.

[0208] UE 1102 may be communicatively coupled to AN 1104 via connection 1106. Connection 1106 is shown as an air interface for implementing the communicatively coupling and may conform to a cellular communication protocol (e.g., an LTE protocol or a 5G NR protocol operating at mmWave or sub-6 GHz frequencies).

[0209] UE 1102 includes a host platform 1108, which is coupled to a modem platform 1110. Host platform 1108 includes application processing circuitry 1112, which can be coupled to protocol processing circuitry 1114 of modem platform 1110. Application processing circuitry 1112 can run various applications for giving / sinking application data for UE 1102. Application processing circuitry 1112 can also implement one or more layer operations to send / receive application data to / from a data network. These layer operations include transport (e.g., UDP) and Internet (e.g., IP) operations.

[0210] The protocol processing circuitry 1114 may implement one or more layer operations to facilitate sending or receiving data over the connection 1106. The layer operations implemented by the protocol processing circuitry 1114 may include, for example, MAC, RLC, PDCP, RRC, and NAS operations.

[0211] The modem platform 1110 may also include digital baseband circuitry 1116, which may implement one or more layer operations that are "lower" layer operations in the network protocol stack performed by the protocol processing circuitry 1114. These operations include, for example, PHY operations, which include one or more of the following: HARQ-ACK functions, scrambling / descrambling, encoding / decoding, layer mapping / demapping, modulation symbol mapping, received symbol / bit metric determination, multi-antenna port precoding / decoding (which may include one or more of space-time, space-frequency, or space coding), reference signal generation / detection, preamble sequence generation and / or decoding, synchronization sequence generation / detection, control channel signal blind decoding, and other related functions.

[0212] The modem platform 1110 may also include a transmit circuit 1118, a receive circuit 1120, an RF circuit 1122, and an RF front end (RFFE) 1124 (which may include or be connected to one or more antenna panels 1126). In short, the transmit circuit 1118 includes a digital-to-analog converter, a mixer, an intermediate frequency (IF) component, etc.; the receive circuit 1120 includes an analog-to-digital converter, a mixer, an IF component, etc.; the RF circuit 1122 includes a low noise amplifier, a power amplifier, a power tracking component, etc.; the RFFE 1124 includes a filter (e.g., a surface / bulk acoustic wave filter), a switch, an antenna tuner, a beamforming component (e.g., a phased array antenna component), etc. The selection and arrangement of the components of transmit circuitry 1118, receive circuitry 1120, RF circuitry 1122, RFFE 1124, and antenna panel 1126 (generally referred to as "transmit / receive components" or "Tx / Rx components") may be specific to the details of a particular implementation (e.g., whether the communication is TDM or FDM, in mmWave or sub-6gHz frequencies, etc.). In some examples, the transmit / receive components may be arranged in multiple parallel transmit / receive chains, may be provided in the same or different chips / modules, etc.

[0213] In some examples, protocol processing circuitry 1114 includes one or more instances of control circuitry (not shown) to provide control functionality for the transmit / receive components.

[0214] UE reception may be established by and via antenna panel 1126, RFFE 1124, RF circuitry 1122, receive circuitry 1120, digital baseband circuitry 1116, and protocol processing circuitry 1114. In some examples, antenna panel 1126 may receive transmissions from AN 1104 via receive beamformed signals received by a set of antennas / antenna elements of one or more antenna panels 1126.

[0215] UE transmission may be established by and via protocol processing circuitry 1114, digital baseband circuitry 1116, transmit circuitry 1118, RF circuitry 1122, RFFE 1124, and antenna panel 1126. In some examples, the transmit component of UE 1104 may apply a spatial filter to the data to be transmitted to form a transmit beam emitted by the antenna elements of antenna panel 1126.

[0216] Similar to UE 1102, AN 1104 includes a host platform 1128, which is coupled to a modem platform 1130. The host platform 1128 includes an application processing circuit 1132, which is coupled to a protocol processing circuit 1134 of the modem platform 1130. The modem platform may also include a digital baseband circuit 1136, a transmit circuit 1138, a receive circuit 1140, an RF circuit 1142, an RFFE circuit 1144, and an antenna panel 1146. The components of AN 1104 may be similar to the components of the same name of UE 1102 and are substantially interchangeable. In addition to performing data transmission / reception as described above, the components of AN 1108 may also perform various logical functions, including, for example, RNC functions (e.g., radio bearer management, UL and DL dynamic radio resource management, and data packet scheduling).

[0217] Examples of antenna elements of antenna panel 1126 and / or antenna elements of antenna panel 1146 include a planar inverted-F antenna (PIFA), a monopole antenna, a dipole antenna, a loop antenna, a patch antenna, a Yagi antenna, a parabolic dish antenna, an omnidirectional antenna, etc.

[0218] Fig.12 Components capable of reading instructions from a machine-readable or computer-readable medium (e.g., a non-transitory machine-readable storage medium) and executing any one or more of the methods discussed herein are shown. Specifically, Fig.12 A diagrammatic representation of hardware resources 1200 is shown, including one or more processors (or processor cores) 1210, one or more memory / storage devices 1220, and one or more communication resources 1230, each of which can be communicatively coupled via a bus 1240 or other interface circuitry. For examples utilizing node virtualization (e.g., NFV), a hypervisor 1202 can be executed to provide an execution environment for one or more network slices / subslices to utilize the hardware resources 1200.

[0219] Processor 1210 may include, for example, processor 1212 and processor 1214. Processor 1210 may be, for example, a central processing unit (CPU), a reduced instruction set computing (RISC) processor, a complex instruction set computing (CISC) processor, a graphics processing unit (GPU), a DSP (e.g., a baseband processor), an ASIC, an FPGA, a radio frequency integrated circuit (RFIC), another processor (including those discussed herein), or any suitable combination thereof.

[0220] The memory / storage device 1220 may include main memory, disk storage, or any suitable combination thereof. The memory / storage device 1220 may include, but is not limited to, any type of volatile, non-volatile, or semi-volatile memory (e.g., dynamic random access memory (DRAM), static random access memory (SRAM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), flash memory, solid-state storage, etc.).

[0221] The communication resources 1230 may include interconnect or network interface controllers, components, or other suitable devices to communicate with one or more peripheral devices 1204 or one or more databases 1206 or other network elements via the network 1208. For example, the communication resources 1230 may include a wired communication component (e.g., for coupling via USB, Ethernet, etc.), a cellular communication component, an NFC component, (or low power consumption ) components, components and other communication components.

[0222] The instructions 1250 may include software, programs, applications, applet, apps, or other executable code for causing at least any processor 1210 to perform any one or more of the methods discussed herein. The instructions 1250 may reside in whole or in part in at least one of the processor 1210 (e.g., within a cache memory of the processor), the memory / storage device 1220, or any suitable combination thereof. In addition, any portion of the instructions 1250 may be transmitted to the hardware resources 1200 from any combination of the peripheral device 1204 or the database 1206. Therefore, the memory of the processor 1210, the memory / storage device 1220, the peripheral device 1204, and the database 1206 are examples of computer-readable and machine-readable media.

[0223] 3. Example Implementation

[0224] Fig.13 An example process to be performed by the NFc is shown. Fig.13 The process includes: identifying credentials associated with access to an artificial intelligence / machine learning (AI / ML) model (1301); providing an indication of the credentials to the ADRF 1066 (1302); and accessing the AI / ML model based on the credentials (1303).

[0225] Fig.14 An example process to be performed by ADRF 1066 is shown. Fig.14 The process includes: identifying an indication of credentials related to access to an AI / ML model received from a NFc (1401); and authorizing the NFc to access the AI / ML model based on the credentials (1402).

[0226] Fig.15 An example process to be performed by MTLF 1062b is shown. Fig.15 The process includes: identifying that NFc is going to access the AI / ML model (1501); and sending the authorization credential related to the AI / ML model to NFc (1502).

[0227] Fig.16 An example process to be performed by VRF 762 is shown. Fig.16 The process includes: identifying information related to the MTLF 1062b of the supplier (1601); and registering the MTLF 1062b based on the identified information (1602).

[0228] Fig.17 An example process to be performed by the NRF 1054 is shown. Fig.17 The process includes: identifying an ML model file attribute included in the NF configuration file of NWDAF-MTLF1062b, the attribute indicating a list of ML model file serialization formats supported by NWDAF-MTLF 1062b (1701); and registering NWDAF-MTLF 1062b with NRF 1054 based on the ML model file attribute (1701).

[0229] Additional examples of the presently described methods, devices, systems, and networks discussed herein include the following non-limiting example implementations. Each of the following non-limiting examples may exist independently or may be combined with any one or more other examples provided below or throughout the present disclosure in any permutation or combination.

[0230] Example 1 includes: a method for securing AI / ML models between an entity (e.g., NWDAF, NFp including MtLF) that produces ML models and / or stores ML models in an analytical data repository (ADRF) and an entity (NFc) that consumes the models.

[0231] Example 2 includes: a method, including: ADRF provides authorization to NFc to obtain the AI / ML model. NF service consumers should be authorized to access the AI / ML model in ADRF (or any other NF (e.g., NWDAF MtLF) that can store ML models).

[0232] Example 3 includes: a method as described in Example 2 and / or some other examples herein, wherein the MTLF provides a one-time credential to the consumer NF to access the ML model.

[0233] Example 4 includes: as described in Example 3 and / or some other examples herein: ADRF verifies the one-time credential to authorize the consumer NF to access the ML model.

[0234] Example 5 may include: as described in Examples 2-4 and / or some other examples herein: the consumer NF generates a unique public / private key using an authorization token from the NRF.

[0235] Example 6 includes a method as described in Example 5 and / or some other examples herein, wherein the consumer NF uses a token from the NRF as a certificate chain.

[0236] Example 7 includes: a method as described in Examples 5-6 and / or some other examples herein, wherein: the consumer NF sends the public key to the ADRF for verification, and the ADRF uses a token-based certificate chain for verification.

[0237] Example 8 includes: a method as described in Examples 1-7 and / or some other examples herein, wherein the MTLF generates a symmetric key and encrypts the AI ​​model and then sends it for storage in the ADRF.

[0238] Example 9 includes the method of Examples 1-8 and / or some other examples herein, wherein the ADRF generates a symmetric key and encrypts the ML model, and sends the encrypted model and key to the consumer NF.

[0239] Example 10 includes: a method to be performed by a consumer network function (NFc) of a network, one or more elements of the NFc, or an electronic device that includes and / or implements one or more elements of the NFc, wherein the method includes: identifying credentials associated with access to an artificial intelligence / machine learning (AI / ML) model; providing an indication of the credentials to an analytical data repository function (ADRF) of the network; and accessing the AI / ML model based on the credentials.

[0240] Example 11 includes: a method as described in Example 10 and / or some other examples herein, wherein the AI / ML model is stored in the ADRF and / or another network function of the network (e.g., a network data analysis function (NWDAF) model training logic function (MTLF)).

[0241] Example 12 includes: a method as described in any of Examples 10-11 and / or some other examples herein, wherein the credential is provided by the MTLF.

[0242] Example 13 includes: a method as described in any of Examples 10-12 and / or some other examples herein, wherein the credential is a one-time credential.

[0243] Example 14 includes: a method as described in any of Examples 10-13 and / or some other examples herein, wherein the ADRF verifies the indication of the credential.

[0244] Example 15 includes: a method as described in any of Examples 10-14 and / or some other examples herein, wherein the ADRF sends an indication of the verification of the credential to one or both of the NFc or the MTLF.

[0245] Example 16 may include: a method to be performed by an analytical data repository function (ADRF) of a network, one or more elements of the ADRF, and / or an electronic device including or implementing one or more elements of the ADRF, wherein the method includes: identifying an indication of credentials received from a network function consumer (NFc) of the network related to access to an artificial intelligence / machine learning (AI / ML) model; and based on the credentials, authorizing the NFc to access the AI / ML model.

[0246] Example 17 includes: a method as described in Example 16 and / or some other examples herein, wherein the AI / ML model is stored in the ADRF and / or another network function of the network (e.g., a network data analysis function (NWDAF) model training logic function (MTLF)).

[0247] Example 18 includes: a method as described in any of Examples 16-17 and / or some other examples herein, wherein the credential is provided to the NFc by the MTLF.

[0248] Example 19 includes: a method as described in any of Examples 16-18 and / or some other examples herein, wherein the credential is a one-time credential.

[0249] Example 20 includes: a method as described in any of Examples 16-19 and / or some other examples herein, wherein the ADRF sends an indication of the verification of the credential to one or both of the NFc or the MTLF.

[0250] Example 21 may include: a method to be performed by a model training logic function (MTLF) of a network, one or more elements of the MTLF, and / or an electronic device including or implementing the MTLF, wherein the method includes: identifying that a network function consumer (NFc) of the network is about to access an artificial intelligence / machine learning (AI / ML) model; and sending authorization credentials associated with the model to the NFc.

[0251] Example 22 includes: a method as described in Example 21 and / or some other examples herein, wherein the AI / ML model is stored in an analysis data repository function (ADRF) of the network and / or another network function (e.g., MTLF).

[0252] Example 23 includes: a method as described in any of Examples 21-22 and / or some other examples herein, wherein the credential is a one-time credential.

[0253] Example 24 includes: a method as described in any of Examples 21-23 and / or some other examples herein, wherein the ADRF verifies the indication of the credential.

[0254] Example 25 includes: a method as described in any of Examples 21-24 and / or some other examples herein, wherein the ADRF sends an indication of the verification of the credential to one or both of the NFc or the MTLF.

[0255] Example 26 includes: a method of operating a vendor repository function (VRF) to register MTLFs from the same or different vendors.

[0256] Example 27 includes: a method as described in Example 26 and / or some other examples herein, wherein the VRF is a logical function and is co-located with the NRF.

[0257] Example 28 includes: a method as described in Example 26 and / or some other examples herein, wherein the VRF is a new network function separate from the NRF.

[0258] Example 29 includes: a method as described in Example 28 and / or some other examples herein, wherein NF state management is handled by the NRF for the MTLF, but registration of NF profiles is also completed by the VRF.

[0259] Example 30 includes the method of Examples 26-29 and / or some other examples herein, wherein the VRF handles policy and authorization for accessing MTLFs from the same or different vendors.

[0260] Example 31 includes: a method as described in Examples 26-30 and / or some other examples herein, wherein the VRF provides an access token with a scope of access, validity, etc. to the NF consumer (e.g., MTLF).

[0261] Example 32 includes: a method of operating a NWDAF including a MTLF, including an interoperability support parameter by analysis ID in a Nnrf_NFManagement_NFRegister sent to an NRF, the parameter indicating whether the service producer's NWDAF including a MTLF supports ML model interoperability.

[0262] Example 33 includes: a method as described in Example 32 and / or some other examples herein, wherein the NWDAF containing the MTLF calls Nnrf_NfDiscovery_Request, which includes the ML model file serialization format supported for the trained ML model and interoperability support in the ML model filtering information.

[0263] Example 34 includes: a method as described in Example 33 and / or some other examples herein, wherein the NRF determines a set of NWDAF instances including MTLF that matches at least one of the ML model file serialization format and interoperability support in the Nnrf_NFDiscovery_Request.

[0264] Example 35 includes: a method to be performed by a vendor repository function (VRF), wherein the method includes: identifying information related to a vendor's model training logic function (MTLF); and registering the MTLF.

[0265] Example 36 includes: a method comprising: identifying a machine learning (ML) model file attribute included in a network function (NF) configuration file of a network data analysis function (NWDAF) including a model training logic function (MTLF), the attribute indicating a list of supported ML model file serialization formats; and registering with a NF repository function (NRF) based on the ML model file attribute.

[0266] Example 37 includes: a method as described in Example 36 and / or some other examples herein, wherein the method is performed by a NF repository function (NRF).

[0267] Example Z01 may include an apparatus comprising components for performing one or more elements of a method described in or related to any of Examples 1-37, or any other method or process described herein.

[0268] Example Z02 may include one or more non-transitory computer-readable media, including instructions that, when executed by one or more processors of an electronic device, cause the electronic device to perform one or more elements of a method described in or related to any of Examples 1-37, or any other method or process described herein.

[0269] Example Z03 may include an apparatus comprising logic, modules, or circuits for performing one or more elements of a method described in or related to any of Examples 1-37, or any other method or process described herein.

[0270] Example Z04 may include a method, technique, or process as described in or related to any of Examples 1-37, or a portion or section thereof.

[0271] Example Z05 may include a device comprising: one or more processors and one or more computer-readable media, the media comprising instructions that, when executed by the one or more processors, cause the one or more processors to perform a method, technique, or process described in or related to any one of Examples 1-37 or portions thereof.

[0272] Example Z06 may include signals as described in or related to any of Examples 1-37, or portions or sections thereof.

[0273] Example Z07 may include a datagram, packet, frame, segment, protocol data unit (PDU), or message as described in or related to any of Examples 1-37 or a portion or section thereof or otherwise described in the present disclosure.

[0274] Example Z08 may include a signal encoded with data as described in or related to any of Examples 1-37, or a portion or section thereof, or otherwise described in this disclosure.

[0275] Example Z09 may include a signal encoded with a datagram, packet, frame, segment, protocol data unit (PDU), or message as described in or related to any of Examples 1-37 or a portion or section thereof or otherwise described in the present disclosure.

[0276] Example Z10 may include an electromagnetic signal carrying computer-readable instructions, wherein execution of the computer-readable instructions by one or more processors causes the one or more processors to perform a method, technique, or process described in or related to any one or a portion thereof of Examples 1-37.

[0277] Example Z11 may include a computer program comprising instructions, wherein execution of the program by a processing element causes the processing element to perform a method, technique, or process described in or related to any one of Examples 1-37 or a portion thereof.

[0278] Example Z12 may include a signal in a wireless network as shown and described herein.

[0279] Example Z13 may include a method of communicating in a wireless network as shown and described herein.

[0280] Example Z14 may include a system for providing wireless communications as shown and described herein.

[0281] Example Z15 may include an apparatus for providing wireless communications as shown and described herein.

[0282] Unless explicitly stated otherwise, any of the above examples may be combined with any other example (or combination of examples). The foregoing description of one or more implementations provides illustration and description, but is not intended to be exhaustive or to limit the scope of the embodiments to the precise form disclosed. Modifications and variations are possible in light of the above teachings or may be obtained from the practice of the various embodiments.

[0283] 4. Terminology

[0284] For the purposes of this document, the following terms and definitions apply to the examples and embodiments discussed herein. As used herein, the singular forms "a", "an", and "the" are also intended to include the plural forms, unless the context clearly indicates otherwise. It will be further understood that the terms "comprise" and / or "comprising" when used in this specification specifically refer to the presence of the claimed features, integers, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. The phrase "A and / or B" means (A), (B), or (A and B). For the purposes of this disclosure, the phrase "A, B, and / or C" means (A), (B), (C), (A and B), (A and C), (B and C), or (A, B, and C). The phrase "X" means one or more X or a set of X. The description may use the phrases "in an embodiment", "in some embodiments", "in one implementation", "in some implementations", "in some examples", etc., each of which may refer to one or more of the same or different embodiments, implementations, and / or examples. Furthermore, the terms "comprising," "including," "having," and the like used in connection with the present disclosure are synonymous.

[0285] The terms "master" and "slave" refer, at least in some examples, to a model of asymmetric communication or control, in which one device, process, element or entity ("master") controls one or more other devices, processes, elements or entities ("slaves"). The terms "master" and "slave" are used in this disclosure only for their technical meanings. The terms "master" or "grandmaster" may be replaced with any of the following terms "main", "source", "primary", "initiator", "requestor", "transmitter", "host", "maestro", "controller", "provider", "producer", "client", "source", "mix", "parent", "chief", "mannager", "reference" (e.g., such as "reference clock", etc.), etc. Additionally, the term "slave" may be replaced with any of the following terms: "receiver", "secondary", "subordinate", "replica", "target", "responder", "device", "performer", "agent", "standby", "consumer", "peripheral", "follower", "server", "child", "helper", "worker", "node", etc.

[0286] The terms "coupled", "coupled in a communicative manner", and their derivatives are used herein. The term "coupled" may mean that two or more elements are in direct physical or electrical contact with each other, may mean that two or more elements are in indirect contact with each other but still cooperate or interact with each other, and / or may mean that one or more other elements are coupled or connected between elements referred to as being coupled to each other. The term "directly coupled" may mean that two or more elements are in direct contact with each other. The term "coupled in a communicative manner" may mean that two or more elements may contact each other by communication (including by wired or other interconnection connection, by wireless communication channels or links, etc.).

[0287] The term "establish" or "establishment" refers, at least in some examples (partially or completely), to an action, task, operation, etc. (e.g., opening a device identification or an entity identification) related to actively or passively playing or preparing to play something. Additionally or alternatively, the term "establish" or "establishment" refers, at least in some examples (partially or completely), to an action, task, operation, etc. (e.g., establishing a session, establishing a session, etc.) related to initiating, starting or preheating a communication or initiating, starting or preheating a relationship between two entities or elements. Additionally or alternatively, the term "establish" or "establishment" refers, at least in some examples, to initiating something to enter a working ready state. The term "established" refers, at least in some examples, to an operational or ready-to-use state (e.g., fully established). In addition, any definition of the term "establish" or "establishment" defined in any specification or standard may be used for the purposes of this disclosure, and these definitions are not overturned by the aforementioned definitions.

[0288] The term "obtain" refers, at least in some examples (partially or completely), to an action, task, operation, etc. of intercepting, moving, copying, obtaining, or acquiring (e.g., from memory, an interface, or a buffer) on an original packet stream or on a copy (e.g., a new instance) of a packet stream. Other aspects of obtaining or receiving may involve: instantiating, implementing, or controlling capabilities for obtaining or receiving a packet stream (or the following parameters and templates or template values).

[0289] The term "receive" refers, at least in some examples, to any action (or set of actions) related to receiving or obtaining an object, data, data unit, etc. and / or the fact of receiving an object, data, data unit, etc. The term "receive" refers, at least in some examples, to an object, data, data unit, etc. being pushed to a device, system, element, etc. (e.g., generally referred to as a push model), being pulled by a device, system, element, etc. (e.g., generally referred to as a pull model), etc.

[0290] The term "element" refers, at least in some examples, to a unit that is indivisible and has well-defined boundaries at a given level of abstraction, wherein an element may be any type of entity, including, for example, one or more devices, systems, controllers, network elements, modules, engines, components, etc., or combinations thereof. The term "entity" refers, at least in some examples, to a unique element of a component, architecture, platform, device, and / or system. Additionally or alternatively, the term "entity" refers, at least in some examples, to information transmitted as a payload.

[0291] The term "measurement" refers to the observation and / or quantification of the properties of an object, event or phenomenon in at least some examples. Additionally or alternatively, the term "measurement" refers to an actual instance or set of operations performed with an object that determines a measured value or measurement result and / or an operation that causes a measured value in at least some examples. Additionally or alternatively, the term "measurement" refers to data recorded during testing in at least some examples. The term "metric" refers to a quantity generated in the evaluation of the measured value in at least some examples. Additionally or alternatively, the term "metric" refers to data derived from a measurement set in at least some examples. Additionally or alternatively, the term "metric" refers to a set of events that are combined or otherwise grouped into one or more values ​​in at least some examples. Additionally or alternatively, the term "metric" refers to a combination of measurements or a set of collected data points in at least some examples. Additionally or alternatively, the term "metric" refers to a standard definition of a quantity generated in the evaluation of the performance and / or reliability of a network in at least some examples, which has an expected practicality and is carefully specified to convey the exact meaning of the measured value.

[0292] The term "signal" refers, at least in some examples, to an observable change in quality and / or quantity. Additionally or alternatively, the term "signal" refers, at least in some examples, to a function of conveying information about an object, event, or phenomenon. Additionally or alternatively, the term "signal" refers, at least in some examples, to any time-varying voltage, current, or electromagnetic wave that may or may not carry information. The term "digital signal" refers, at least in some examples, to a signal consisting of a collection of discrete waveforms of physical quantities so as to represent a sequence of discrete values.

[0293] The terms "self" (as in, for example, "self device") and "subject" (as in, for example, "data subject"), at least in some examples, refer to an entity, element, device, system, etc. to be or being considered. The terms "neighbor" and "neighboring" (as in, for example, "neighboring device"), at least in some examples, refer to an entity, element, device, system, etc. other than a self device or a subject device.

[0294] The term "identifier" refers, at least in some examples, to a value or set of values ​​that uniquely identifies an identifier within a specific range. Additionally or alternatively, the term "identifier" refers, at least in some examples, to a character sequence that identifies or otherwise indicates an identifier of a unique object, element, or entity or a unique class of an object, element, or entity. Additionally or alternatively, the term "identifier" refers, at least in some examples, to a character sequence used to identify or reference an application, program, session, object, element, entity, variable, data set, etc. The aforementioned "character sequence" refers, at least in some examples, to one or more names, labels, words, numbers, letters, symbols, and / or any combination thereof. Additionally or alternatively, the term "identifier" refers, at least in some examples, to a name, address, label, distinguishing index, and / or attribute. Additionally or alternatively, the term "identifier" refers, at least in some examples, to an instance of an identifier. The term "permanent identifier" refers, at least in some examples, to an identifier that is reused by a device or another device associated with the same person or a group of people for an indefinite period of time. The term "identification" refers, at least in some examples, to the process of identifying an identity that is distinguishable from other identities in a specific scope or context, which may involve processing the identifier to reference the identity in an identity database. The term "application identifier," "application ID," or "app ID" refers, at least in some examples, to an identifier that can be mapped to a specific application, application instance, or application instance. In the context of 3GPP 5G / NR, an "application identifier" refers, at least in some examples, to an identifier that can be mapped to a specific application service detection rule.

[0295] The term "circuit" refers to a circuit or a system of multiple circuits configured to perform a specific function in an electronic device, at least in some examples. A circuit or a system of circuits may be part of or include one or more hardware components (e.g., logic circuits, processors (shared, dedicated or group) and / or memories (shared, dedicated or group), application specific integrated circuits (ASICs), field programmable gate arrays (FPGAs), programmable logic controllers (PLCs), single board computers (SBCs), systems on chips (SoCs), systems in packages (SiPs), multi-chip packages (MCPs), digital signal processors (DSPs), etc.) configured to provide the described functions. In addition, the term "circuit" may also refer to a combination of one or more hardware elements and the program code that perform the functions of a program code. Some types of circuits may execute one or more software or firmware programs to provide at least some of the described functions. This combination of hardware elements and program code may be referred to as a specific type of circuit.

[0296] The term "processor circuitry" refers, at least in some examples, to, is a part of, or includes circuitry that is capable of sequentially and automatically performing a series of arithmetic or logical operations or recording, storing, and / or transmitting digital data. The term "processor circuitry" refers, at least in some examples, to one or more application processors, one or more baseband processors, a physical CPU, a single-core processor, a dual-core processor, a triple-core processor, a quad-core processor, and / or any other device capable of executing or otherwise operating computer-executable instructions (e.g., program code, software modules, and / or functional processes). The terms "application circuitry" and / or "baseband circuitry" may be considered synonymous with and may be referred to as "processor circuitry."

[0297] The terms "memory" and / or "memory circuit" refer, at least in some examples, to one or more hardware devices for storing data, including random access memory (RAM), static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), magnetoresistive RAM (MRAM), conductive bridge random access memory (CB-RAM), spin transfer torque (STT)-MRAM, phase change RAM (PRAM), core memory, read-only memory (ROM), programmable ROM (PROM), erasable PROM (EPROM), electrical EPROM (EEPROM), flash memory, non-volatile RAM (NVRAM), magnetic disk storage media, optical storage media, flash memory devices, or other machine-readable media for storing data. The term "computer-readable medium" includes, but is not limited to, memory, portable or fixed storage devices, optical storage devices, and various other media capable of storing, containing or carrying instructions or data.

[0298] The term "interface circuitry" refers, at least in some examples, to, is a part of, or includes circuitry that enables information to be exchanged between two or more components or devices. The term "interface circuitry for" refers, at least in some examples, to one or more hardware interfaces (e.g., a bus, an I / O interface, a peripheral component interface, a network interface card, etc.).

[0299] The term "infrastructure processing unit" or "IPU" refers, at least in some examples, to an advanced network device with hardened accelerators and network connectivity (e.g., Ethernet, etc.) that accelerate and manage infrastructure functions using tightly coupled, dedicated programmable cores. In some implementations, the IPU provides full infrastructure offload and provides an additional layer of security by acting as a control point for hosts running infrastructure applications. The IPU is able to offload the entire infrastructure stack from the host and can control how the host attaches to the infrastructure. This gives the service provider an additional layer of security and control implemented in hardware by the IPU.

[0300] The term "device" refers, at least in some examples, to a physical entity that is embedded internally or attached to another physical entity in its vicinity, with the ability to communicate digital information from or to the physical entity. The term "controller" refers, at least in some examples, to an element or entity that has the ability to affect a physical entity, such as by changing its state or moving the physical entity. The term "scheduler" refers, at least in some examples, to an entity or element that allocates resources (e.g., processor time, network link, memory space, and / or the like) to perform a task. The term "network scheduler" refers, at least in some examples, to a node, element, or entity that manages network packets in the send and / or receive queues of one or more protocol stacks of a network access circuit (e.g., a network interface controller (NIC), a baseband processor, etc.). The term "network scheduler" can be used interchangeably with the terms "packet scheduler," "queuing order," or "qdisc," and / or "queuing algorithm," at least in some examples.

[0301] The term "terminal" refers, at least in some examples, to a point where a conductor from a component, device, or network ends. Additionally or alternatively, the term "terminal" refers, at least in some examples, to an electrical connector that acts as an interface to a conductor and creates a point to which an external circuit can be connected. In some examples, a terminal can include an electrical lead, an electrical connector, an electrical connector, a solder cup or barrel, and the like.

[0302] The term "computing node" or "computing device" refers, at least in some examples, to an identifiable entity that implements aspects of computing operations, whether part of a larger system, a collection of distributed systems, or a stand-alone device. In some examples, a computing node may be referred to as a "computing device," "computing system," etc., whether operating as a client, server, or intermediate entity. Specific implementations of computing nodes may be incorporated into servers, base stations, gateways, roadside units, residential units, user equipment, end consumer devices, appliances, etc. For purposes of this disclosure, the term "node" refers, at least in some examples, to and / or may be interchangeable with the terms "device," "component," "subsystem," etc.

[0303] The term "computer system" refers, at least in some examples, to any type of interconnected electronic device, computer device, or components thereof. Additionally, the terms "computer system" and / or "system" refer, at least in some examples, to various components of a computer that are communicatively coupled to one another. Furthermore, the terms "computer system" and / or "system" refer, at least in some examples, to multiple computer devices and / or multiple computing systems that are communicatively coupled to one another and configured to share computing resources and / or networking resources.

[0304] The term "server" refers to a computing device or system including processing hardware and / or processing space, associated storage media (e.g., storage devices or databases), and suitable applications known in the art in some cases, at least in some examples. The terms "server system" and "server" can be used interchangeably herein, and these terms refer to one or more computing systems that provide access to physical and / or virtual resource pools, at least in some examples. The various servers discussed herein include computer devices having rack-mounted computing architecture components, tower computing architecture components, blade computing architecture components, etc. A server can represent a server cluster, a server farm, a cloud computing service, or other server group or pool that can be located in one or more data centers. The server can also be connected to one or more data storage devices (not shown) or otherwise associated therewith. In addition, the server includes an operating system (OS), which provides executable program instructions for general management and operation of a separate server computer device, and includes a computer-readable medium storing instructions, which, when executed by a processor of the server, can allow the server to perform its intended function. Suitable implementations of the OS and general functions of the server are known or commercially available, and can be easily implemented by a person of ordinary skill in the art.

[0305] The term "platform" refers, at least in some examples, to an environment in which instructions, program codes, software elements, and the like can be executed or otherwise operated, examples of which include an architecture (e.g., a motherboard, a computing system, etc.), one or more hardware elements (e.g., an embedded system, etc.), a cluster of computing nodes, a distributed collection of computing nodes or a network, an operating system, a virtual machine (VM), a virtualization container, a software framework, a client application (e.g., a web browser, etc.) and associated application programming interfaces, cloud computing services (e.g., platform as a service (PaaS)), or other underlying software that executes instructions, program codes, software elements, and the like.

[0306] The term "architecture" refers, at least in some examples, to computer architecture or network architecture. "Computer architecture" refers, at least in some examples, to the physical and logical design or arrangement of software and / or hardware elements in a computing system or platform, including technical standards for interaction therebetween. The term "network architecture" refers, at least in some examples, to the physical and logical design or arrangement of software and / or hardware elements in a network, including communication protocols, interfaces, and media transports.

[0307] The terms "appliance", "computer appliance", etc. refer, at least in some examples, to a computer device or computer system with program code (e.g., software or firmware) specifically designed to provide specific computing resources. The term "virtual device" refers, at least in some examples, to a virtual machine image to be implemented by a device equipped with a hypervisor that virtualizes or emulates a computer appliance or is otherwise dedicated to providing specific computing resources. The terms "security appliance", "firewall", etc. refer, at least in some examples, to a computer appliance designed to protect a computer network from unwanted business and / or malicious attacks. The term "policy appliance" refers, at least in some examples, to a technical control and logging mechanism for implementing or coordinating policy rules (information usage rules) and ensuring accountability in information systems. The term "gateway" refers, at least in some examples, to a network appliance that allows data to flow from one network to another or a computing system or application configured to perform these tasks. Examples of gateways include IP gateways, Internet to Rail (I2O) gateways, IoT gateways, cloud storage gateways, etc.

[0308] The term "user equipment" or "UE" in some examples refers to a device with radio communication capabilities and can describe a remote user of network resources in a communication network. The term "user equipment" or "UE" can be considered synonymous with and can be referred to as a client, mobile, mobile device, mobile terminal, user terminal, mobile unit, station, mobile station, mobile user, subscriber, user, remote station, access agent, user agent, receiver, radio device, reconfigurable radio device, reconfigurable mobile device, etc. In addition, the term "user equipment" or "UE" includes any type of wireless / wired device or any computing device that includes a wireless communication interface. Examples of UE, client device, etc. include desktop computers, workstations, laptop computers, mobile data terminals, smart phones, tablet computers, wearable devices, machine-to-machine (M2M) devices, machine type communication (MTC) devices, Internet of Things (IoT) devices, embedded systems, sensors, self-driving cars, drones, robots, in-vehicle infotainment systems, dashboards, onboard diagnostic devices, dashboard mobile devices, electronic engine management systems, electronic / engine control units / modules, microcontrollers, control modules, server devices, network appliances, head-up display (HUD) devices, helmet-mounted display devices, augmented reality (AR) devices, virtual reality (VR) devices, mixed reality (MR) devices, and / or other similar systems or devices. The term "station" or "STA" refers, at least in some examples, to a logical entity that is a separately addressable instance of a medium access control (MAC) and physical layer (PHY) interface to a wireless medium (WM). The term "wireless medium" or "WM" refers, at least in some examples, to the medium used to enable transmission of protocol data units (PDUs) between peer physical layer (PHY) entities of a wireless local area network (LAN).

[0309] The term "network element" refers to a physical or virtualized device and / or infrastructure used to provide wired or wireless communication network services in at least some examples. The term "network element" may be considered synonymous with and / or referred to as a networked computer, networking hardware, network equipment, network node, router, switch, hub, bridge, radio network controller, network access node (NAN), base station, access point (AP), RAN equipment, RAN node, gateway, server, network appliance, network function (NF), virtualized NF (VNF), etc. The term "network controller" refers to a functional block that centralizes some or all control and management functions of a network domain and can provide an abstract view of the network domain to other functional blocks via an interface in at least some examples. The term "network access node" or "NAN" refers to a network element in a radio access network (RAN) that is responsible for sending and receiving radio signals to or from a UE or station in one or more cells or coverage areas. A "network access node" or "NAN" may have an integrated antenna or may be connected to an antenna array via a feeder cable. Additionally or alternatively, a "network access node" or "NAN" includes dedicated digital signal processing, network function hardware and / or computing hardware to operate as a computing node. In some examples, for flexibility, cost and performance, a "network access node" or "NAN" can be split into multiple functional blocks that operate in software. In some examples, a "network access node" or "NAN" can be a base station (e.g., an evolved Node B (eNB) or a next generation Node B (gNB)), an access point and / or a wireless network access point, a router, a switch, a hub, a radio unit or a remote radio head, a transmit receive point (TRxP), a gateway device (e.g., a residential gateway, a wired 5G access network, a wired 5G cable access network, a wired BBF access network, etc.), a network appliance and / or some other network access hardware. The term "access point" or "AP" refers to an entity that includes a station (STA) and provides access to distribution services to associated STAs via a wireless medium (WM), at least in some examples. The AP includes a STA and a distribution system access function (DSAF).

[0310] The term "cell" refers, at least in some examples, to a radio network object that can be uniquely identified by a UE according to an identifier (e.g., a cell ID) broadcast over a geographic area from a network access node (NAN). Additionally or alternatively, the term "cell" refers, at least in some examples, to a geographic area covered by a NAN. The term "serving cell" refers, at least in some examples, to a UE in a connected mode or state (e.g., RRC_CONNECTED) and not configured with carrier aggregation (CA) and / or dual connectivity (DC), to a primary cell (PCell). Additionally or alternatively, the term "serving cell" refers, at least in some examples, to a UE in a connected mode or state (e.g., RRC_CONNECTED) and configured with CA, to a cell set including zero or more special cells and one or more secondary cells. The term "primary cell" or "PCell" refers, at least in some examples, to a primary cell group (MCG) cell operating on a primary frequency, in which the UE performs an initial connection establishment procedure or initiates a connection reestablishment procedure. The term "secondary cell" or "SCell" refers, at least in some examples, to a UE configured with CA, to a cell that provides additional radio resources in addition to a special cell (SpCell). The term "special cell" or "SpCell" refers, at least in some examples, to a PCell for non-DC operation, or to a PCell of an MCG or a PSCell of an SCG for DC operation. The term "primary cell group" or "MCG" refers, at least in some examples, to a group of service cells associated with a "master node" that includes an SpCell (PCell) and optionally one or more SCells. The term "secondary cell group" or "SCG" refers, at least in some examples, to a UE configured with DC, to a subset of service cells that includes a primary SCell (PSCell) and zero or more SCells. The term "primary SCG cell" refers, for DC operation, to an SCG cell to which a UE performs random access when performing a reconfiguration with a synchronization process. The term "handover" refers, at least in some examples, to transferring a user's connection from one radio channel to another (which may be the same or a different cell). Additionally or alternatively, the term "handover" refers, at least in some examples, to a process by which a radio access network changes radio transmitters, radio access modes and / or radio systems used to provide a bearer service while maintaining a defined bearer service QoS.

[0311] The term "primary node" or "MN" refers, at least in some examples, to a NAN that provides a control plane connection to a core network. The term "secondary node" or "SN" refers, at least in some examples, to a NAN that provides resources to a UE in addition to the resources provided by the MN and / or NAN without a control plane connection to the core network. The term "E-UTEAN NodeB", "eNodeB" or "eNB" refers, at least in some examples, to a RAN node that provides E-UTRA user plane (e.g., PDCP, RLC, MAC, PHY) and control plane (e.g., RRC) protocol terminations toward a UE and is connected to an evolved packet core (EPC) via an S1 interface. Two or more eNBs are interconnected with each other (and / or with one or more en-gNBs) by means of an X2 interface. The term "next generation eNB" or "ng-eNB" refers, at least in some examples, to a RAN node that provides E-UTRA user plane and control plane protocol terminations toward a UE and is connected to a 5GC via an NG interface. Two or more ng-eNBs are interconnected with each other (and / or with one or more gNBs) by means of an Xn interface. The term "Next Generation Node B", "gNodeB" or "gNB" refers, at least in some examples, to a RAN node that provides NR user plane and control plane protocol termination towards the UE and is connected to the 5GC via the NG interface. Two or more gNBs are interconnected with each other (and / or with one or more ng-eNBs) by means of an Xn interface. The term "E-UTRA-NR gNB" or "en-gNB" refers, at least in some examples, to a RAN node that provides NR user plane and control plane protocol termination towards the UE and acts as a secondary node in an E-UTRA-NR dual connectivity (EN-DC) scenario (see, for example, 3GPP TS 37.340 v17.0.0 (2022-04-15) ("[TS37340]")). Two or more en-gNBs are interconnected with each other (and / or with one or more eNBs) by means of an X2 interface.

[0312] The term "next generation RAN node" or "NG-RAN node" refers to a gNB or ng-eNB in ​​at least some examples. The term "IAB node" refers to a RAN node that supports a new radio (NR) access link to a user equipment (UE) and an NR backhaul link to a parent node and a child node in at least some examples. The term "IAB donor" refers to a RAN node (e.g., a gNB) that provides network access to a UE via a network of backhaul and access links in at least some examples. The term "transmit receive point" or "TRxP" refers to an antenna array having one or more antenna elements that can be used by a network located at a specific geographic location in a specific area in at least some examples. The term "central unit" or "CU" refers to a logical node of the RRC and PDCP protocols of an en-gNB that hosts the radio resource control (RRC), service data adaptation protocol (SDAP) and / or packet data convergence protocol (PDCP) protocol / layers of an NG-RAN node or controls the operation of one or more DUs; the CU terminates the F1 interface connected to the DU and can be connected to multiple DUs. The term "distributed unit" or "DU" refers, at least in some examples, to a logical node hosting the Backhaul Adaptation Protocol (BAP), F1 Application Protocol (F1AP), Radio Link Control (RLC), Medium Access Control (MAC), and Physical (PHY) layers of an NG-RAN node or en-gNB, whose operations are partially controlled by the CU; one DU supports one or more cells, and one cell is supported by only one DU; and the DU terminates the F1 interface connected to the CU. The term "radio unit" or "RU" refers, at least in some examples, to a logical node hosting the PHY layer or Low-PHY layer and the radio frequency (RF) processing based on the split of lower layer functions. The term "split architecture" refers, at least in some examples, to an architecture in which the CU, DU, and / or RU are physically separated from each other. Additionally or alternatively, the term "split architecture" refers, at least in some examples, to a RAN architecture such as the architecture discussed in 3GPP TS 38.401 v17.4.0 (2023-04-03) ("[TS38401]"), 3GPP TS 38.410 v 17.1.0 (2022-06-23), and 3GPP TS 38.473 v17.4.1 (2023-04-05) ("[TS38473]"), the contents of each of which are hereby incorporated by reference in their entirety. The term "integrated architecture" refers, at least in some examples, to an architecture in which the RU and DU are implemented on one platform and / or an architecture in which the DU and CU are implemented on one platform.

[0313] The term "residential gateway" or "RG" refers, at least in some examples, to a device that provides, for example, voice, data, broadcast video, video on demand, etc. to other devices in a customer's premises. The term "wired 5G access network" or "W-5GAN" refers, at least in some examples, to a wired AN connected to the 5GC via the N2 and N3 reference points. The W-5GAN may be a W-5GBAN or a W-5GCAN. The term "wired 5G cable access network" or "W-5GCAN" refers, at least in some examples, to an access network defined in / by CableLabs. The term "wired BBF access network" or "W-5GBAN" refers, at least in some examples, to an access network defined in / by the Broadband Forum (BBF). The term "wired access gateway function" or "W-AGF" refers, at least in some examples, to a network function in the W-5GAN that provides connectivity to the 3GPP 5G core network (5GC) to the 5G-RG and / or FN-RG. The term "5G-RG" refers, at least in some examples, to an RG capable of connecting to a 5GC that plays the role of a user equipment with respect to the 5GC; it supports security elements and exchanges N1 signaling with the 5GC. The 5G-RG may be a 5G-BRG or a 5G-CRG.

[0314] The term "SMTC" refers to the SSB-based measurement timing configuration configured by SSB-MeasurementTimingConfiguration. The term "SSB" refers to the SS / PBCH block.

[0315] The term "primary cell" refers to an MCG cell operating on the primary frequency, in which the UE performs an initial connection establishment procedure or initiates a connection reestablishment procedure. The term "primary SCG cell" refers to an SCG cell to which the UE performs random access when performing a reconfiguration with synchronization procedure for DC operation. The term "secondary cell" refers to a cell that provides additional radio resources in addition to special cells for a UE configured with CA. The term "secondary cell group" refers to a subset of service cells including a PSCell and zero or more secondary cells for a UE configured with DC. The term "serving cell" refers to a primary cell for a UE in RRC_CONNECTED that is not configured with CA / DC, and there is only one serving cell including the primary cell. The term "serving cell" or "each serving cell" refers to a set of cells including a special cell and all secondary cells for a UE in RRC_CONNECTED that is configured with CA. The term "special cell" refers to a PCell of an MCG or a PSCell of an SCG for DC operation; otherwise, the term "special cell" refers to a Pcell.

[0316] The term "edge computing" refers, at least in some examples, to the implementation or arrangement of distributed computing elements that move processing activities and resources (e.g., computing resources, storage resources, acceleration resources, and / or network resources) toward the "edge" of a network to reduce latency and increase throughput for endpoint users (client devices, user devices, etc.). Additionally or alternatively, the term "edge computing" refers, at least in some examples, to a collection of services hosted at an access point of attachment to a network relatively close to a client / UE to achieve relatively efficient service delivery by reducing end-to-end latency and / or load on a transport network. In some examples, edge computing implementations involve providing services and / or resources in cloud-like systems, functions, applications, and subsystems from one or more locations accessible via a wireless network. Additionally or alternatively, the term "edge computing" refers, at least in some examples, to the concept described in [TS23501]: enabling operator and third-party services to be hosted near a UE's access point of attachment to achieve efficient service delivery by reducing end-to-end latency and load on a transport network. The term "edge computing node" or "edge computing device" refers, at least in some examples, to an identifiable entity that implements aspects of edge computing operations, whether part of a larger system, a distributed system collection, or a stand-alone device. In some examples, a computing node may be referred to as an "edge node," "edge device," "edge system," whether operating as a client, server, or intermediate entity. Additionally or alternatively, the term "edge computing node" refers, at least in some examples, to a real-world, logical, or virtualized implementation of a computing-capable element in the form of a device, gateway, bridge, system or subsystem, component, whether operating in server, client, endpoint, or peer-to-peer mode, and whether located at the "edge" of a network or at a more distant connected location within the network. However, references to an "edge computing system" generally refer to a distributed architecture, organization, or collection of multiple nodes and devices, and which are organized to complete or provide some aspects of services or resources in an edge computing environment. The term "edge computing platform" or "edge platform" refers, at least in some examples, to a set of functions for instantiating, executing, or running edge applications on a particular edge computing node (e.g., a virtualized infrastructure, etc.) so that such edge applications can provide and / or consume edge services, and / or otherwise provide one or more edge services. The term "edge application" or "edge app" refers, at least in some examples, to an application that can be instantiated on or executed by an edge computing node within an edge computing network, system, or framework and can potentially provide and / or consume edge computing services. The term "edge service" refers, at least in some examples, to a service provided either by the edge platform itself and / or by an edge application via an edge computing node and / or edge platform.

[0317] The term "cloud computing" or "cloud" refers, at least in some examples, to a paradigm for enabling network access to a scalable, elastic pool of shareable computing resources through on-demand self-service provisioning and management, and without the need for active management by the user. Cloud computing provides cloud computing services (or cloud services), which are one or more capabilities provided by cloud computing via calls using defined interfaces (e.g., APIs, etc.).

[0318] The term "network function" or "NF" refers, at least in some examples, to a functional block of a network infrastructure having one or more external interfaces and defined functional behaviors. The term "network instance" refers, at least in some examples, to information identifying a domain; in some examples, a UPF uses a network instance for service detection and routing. The term "network service" or "NS" refers, at least in some examples, to a composition or collection of NFs and / or network services defined by their functional and behavioral specifications. The term "NF service instance" refers, at least in some examples, to an identifiable instance of a NF service. The term "NF instance" refers, at least in some examples, to an identifiable instance of a NF. The term "NF service" refers, at least in some examples, to a function exposed by a NF through a service-based interface and consumed by other authorized NFs. The term "NF service operation" refers, at least in some examples, to a basic unit that composes a NF service. The term "NF service set" refers, at least in some examples, to a group of interchangeable NF service instances of the same service type within a NF instance; in some examples, NF service instances in the same NF service set may have access to the same context data. The term "NF set" refers, at least in some examples, to a group of interchangeable NF instances of the same type that support the same service and the same network slice; in some examples, NF instances in the same NF set can be geographically distributed but have access to the same context data.

[0319] The term "RAN function" or "RANF" refers, at least in some examples, to a functional block within the RAN architecture having one or more external interfaces and defined behaviors related to the operation of the RAN or RAN node. Additionally or alternatively, the term "RAN function" or "RANF" refers, at least in some examples, to a set of functions and / or NFs that are part of the RAN. The term "application function" or "AF" refers, at least in some examples, to an element or entity that interacts with the 3GPP core network in order to provide services. Additionally or alternatively, the term "application function" or "AF" refers, at least in some examples, to an edge computing node or ECT framework from the perspective of the 5G core network. The term "management function" refers, at least in some examples, to a logical entity that plays the role of a service consumer and / or service producer. The term "management service" refers, at least in some examples, to a set of management capabilities provided. The term "network function virtualization" or "NFV" refers, at least in some examples, to the principle of separating network functions from the hardware on which they run by using virtualization techniques and / or virtualization technologies. The term "virtualized network function" or "VNF" refers, at least in some examples, to an implementation of a NF that can be deployed on a network function virtualization infrastructure (NFVI). The term "network function virtualization infrastructure manager" or "NFVI" refers, at least in some examples, to the sum of all hardware and software components that build an environment in which a VNF is deployed. The term "virtualized infrastructure manager" or "VIM" refers, at least in some examples, to a functional block that is typically responsible for controlling and managing NFVI computing, storage, and network resources within an operator's infrastructure domain. The term "virtualization container," "execution container," or "container" refers, at least in some examples, to a partition of a computing node that provides an isolated virtualized computing environment. The term "operating system container" refers, at least in some examples, to a virtualized container that utilizes a shared operating system (OS) kernel of its host, wherein the host that provides the shared OS kernel can be a physical computing node or another virtualized container. Additionally or alternatively, the term "container" refers, at least in some examples, to a standard software unit (or package) that includes code and its associated dependencies and / or an abstraction at the application layer that encapsulates code and dependencies together. Additionally or alternatively, the term "container" or "container image" refers, at least in some examples, to a lightweight, standalone executable software package that includes everything needed to run an application (e.g., code, runtime environment, system tools, system libraries, and settings). The term "virtual machine" or "VM" refers, at least in some examples, to a virtualized computing environment that behaves in the same or similar manner as a physical computer and / or server. The term "hypervisor" refers, at least in some examples, to a software element that partitions the underlying physical resources of a computing node, creates VMs, manages resources for VMs, and isolates individual VMs from each other.

[0320] The term "data network" or "DN" refers, at least in some examples, to a network that hosts a data center service (e.g., a carrier service, the Internet, a third-party service, or an enterprise network). Additionally or alternatively, the DN refers, at least in some examples, to a service network belonging to a carrier or a third party that provides a client or user equipment (UE) as a service. The DN is sometimes referred to as a "packet data network" or "PDN". The term "local area data network" or "LADN" refers, at least in some examples, to a DN that is accessible only to UEs in a specific location, which provides connectivity to a specific DNN, and whose availability is provided to the UE.

[0321] The term "Internet of Things" or "IoT" refers, at least in some examples, to a system of interrelated computing devices, mechanical and digital machines capable of communicating data with little or no human interaction, and may involve technologies such as real-time analytics, machine learning and / or AI, embedded systems, wireless sensor networks, control systems, automation (e.g., smart home, smart building, and / or smart city technologies), etc. IoT devices are typically low-power devices without heavy computing or storage capabilities.

[0322] The term "protocol" refers, at least in some examples, to a predefined process or method for performing one or more operations. Additionally or alternatively, the term "protocol" refers, at least in some examples, to a common means (sometimes also referred to as an interface) for unrelated objects to communicate with each other. The term "communication protocol" refers, at least in some examples, to a standardized set of rules or instructions implemented by a communication device and / or system to communicate with other devices and / or systems, including instructions for encapsulating / decapsulating data, modulating / demodulating signals, implementation of a protocol stack, etc. In various implementations, "protocol" and / or "communication protocol" can be represented using a protocol stack, a finite state machine (FSM), and / or any other suitable data structure. The term "standard protocol" refers, at least in some examples, to a protocol whose specifications are published and known to the public and controlled by a standards body. The term "protocol stack" or "network stack" refers, at least in some examples, to an implementation of a protocol suite or protocol family. In various implementations, a protocol stack includes a set of protocol layers, wherein the lowest protocol handles low-level interactions with hardware and / or communication interfaces, and each higher layer adds additional capabilities. Additionally or alternatively, the term "protocol" refers, at least in some examples, to a formal set of procedures applicable to ensuring communication between two or more functions within the same layer of a hierarchy of functions.

[0323] The term "application layer" refers, at least in some examples, to an abstraction layer that specifies shared communication protocols and interfaces used by hosts in a communication network. Additionally or alternatively, the term "application layer" refers, at least in some examples, to an abstraction layer that interacts with software applications that implement communication components and includes identifying communication partners, determining resource availability, and synchronizing communications. Examples of application layer protocols include HTTP, HTTPs, File Transfer Protocol (FTP), Dynamic Host Configuration Protocol (DHCP), Internet Message Access Protocol (IMAP), Lightweight Directory Access Protocol (LDAP), MQTT (MQ Telemetry Transport), Remote Authentication Dial-In User Service (RADIUS), Diameter Protocol, Extensible Authentication Protocol (EAP), RDMA over Converged Ethernet Version 2 (RoCEv2), Real-time Transport Protocol (RTP), RTP Control Protocol (RTCP), Real-time Streaming Protocol (RTSP), SBMV Protocol, Thin Client Control Protocol (SCCP), Session Initiation Protocol (SIP), Session Description Protocol (SDP), Simple Mail Transfer Protocol (SMTP), Simple Network Management Protocol (SNMP), Simple Service Discovery Protocol (SSDP), Small Computer System Interface (SCSI), Internet SCSI (iSCSI), iSCSI Extensions for RDMA (iSER), Transport Layer Security (TLS), Voice over IP (VoIP), Virtual Private Network (VPN), Extensible Messaging and Presence Protocol (XMPP), etc.

[0324] The term "session layer" refers, at least in some examples, to controlling conversations and / or connections between entities or elements, and may include an abstraction layer that establishes, manages, and terminates connections between entities or elements.

[0325] The term "transport layer" refers, at least in some examples, to a protocol layer that provides end-to-end (e2e) communication services (e.g., connection-oriented communication, reliability, flow control, and multiplexing). Examples of transport layer protocols include Datagram Congestion Control Protocol (DCCP), Fibre Channel Protocol (FBC), Generic Routing Encapsulation (GRE), GPRS Tunneling (GTP), Micro Transport Protocol (μTP), Multipath TCP (MPTCP), Multipath QUIC (MPQUIC), Multipath UDP (MPUDP), Quick UDP Internet Connections (QUIC), Remote Direct Memory Access (RDMA), Resource Reservation Protocol (RSVP), Stream Control Transmission Protocol (SCTP), Transmission Control Protocol (TCP), User Datagram Protocol (UDP), etc.

[0326] The term "network layer" refers, at least in some examples, to a protocol layer including means for transmitting network packets from a source to a destination via one or more networks. Additionally or alternatively, the term "network layer" refers, at least in some examples, to a protocol layer responsible for packet forwarding and / or routing through intermediate nodes. Additionally or alternatively, the term "network layer" or "internet layer" refers, at least in some examples, to a protocol layer including intercommunication methods, protocols, and specifications for transmitting network packets across networks. For example, network layer protocols include Internet Protocol (IP), IP Security (IPsec), Internet Control Message Protocol (ICMP), Internet Group Management Protocol (IGMP), Open Shortest Path First Protocol (OSPF), Routing Information Protocol (RIP), RDMA over Converged Ethernet Version 2 (RoCEv2), Subnet Access Protocol (SNAP), and / or some other Internet or network protocol layers.

[0327] The term "link layer" or "data link layer" refers, at least in some examples, to a protocol layer that transfers data between nodes on a network segment across the physical layer. Examples of link layer protocols include Logical Link Control (LLC), Media Access Control (MAC), Ethernet, RDMA over Converged Ethernet Version 1 (RoCEv1), etc.

[0328] The terms "radio resource control", "RRC layer" or "RRC" refer, at least in some examples, to a protocol layer or sublayer that performs: system information processing; paging; establishment, maintenance and release of RRC connections; security functions; establishment, configuration, maintenance and release of signaling radio bearers (SRBs) and data radio bearers (DRBs); mobility functions / services; QoS management; and some sidelink specific services and functions over the Uu interface (see, for example, 3GPP TS 36.331 v17.4.0 (2023-03-30) ("[TS36331]") and / or 3GPP TS 38.331 v17.4.0 (2023-03-30) ("[TS38331]")).

[0329] The terms "Service Data Adaptation Protocol", "SDAP layer", or "SDAP" refer, at least in some examples, to a protocol layer or sublayer that performs mapping between QoS flows and data radio bearers (DRBs) and marks QoS flow IDs (QFIs) in both DL and UL packets (see, for example, 3GPP TS 37.324 v17.0.0 (2022-04-13) ("[TS37324]").

[0330] The terms "Packet Data Convergence Protocol", "PDCP layer", or "PDCP" refer, at least in some examples, to a protocol layer or sublayer that performs: transmitting user plane or control plane data; maintaining PDCP sequence numbers (SNs); header compression and decompression using Robust Header Compression (ROHC) and / or Ethernet Header Compression (EHC) protocols; encryption and decryption; integrity protection and integrity verification; providing timer-based SDU discard; routing for split bearers; duplication and duplicate discard; reordering and in-order delivery; and / or out-of-order delivery (see, for example, 3GPP TS 36.323v17.2.0 (2023-01-13) and / or 3GPP TS38.323v17.4.0 (2023-03-28) ("[TS38323]")).

[0331] The terms "radio link control layer", "RLC layer" or "RLC" refer, at least in some examples, to a protocol layer or sublayer that performs delivery of upper layer PDUs; sequence numbering independent of sequence numbering in PDCP; error correction through ARQ; segmentation and / or resegmentation of RLC SDUs; reassembly of SDUs; duplicate detection; RLC SDU discard; RLC reconstruction; and / or protocol error detection (see, for example, 3GPP TS 36.322 v17.0.0 (2022-04-15) and 3GPP TS 38.322 v17.2.0 (2023-01-13) ("[TS38322]")).

[0332] The term "medium access control protocol", "MAC protocol" or "MAC" refers, at least in some examples, to a protocol that manages access to a transmission medium in a network to enable the exchange of data between stations in the network. Additionally or alternatively, the term "medium access control layer", "MAC layer" or "MAC" refers, at least in some examples, to a protocol layer or sublayer that performs functions to provide frame-based connectionless mode (e.g., datagram style) data transfer between stations or devices. Additionally or alternatively, the term "medium access control layer", "MAC layer" or "MAC", at least in some examples, refers to a protocol layer or sublayer that: performs mapping between logical channels and transport channels; multiplexing / demultiplexing MAC SDUs belonging to one or different logical channels into / from transport blocks (TBs) delivered to / from the physical layer on the transport channel; scheduling information reporting; error correction through HARQ (one HARQ entity per cell in the case of CA); priority handling between UEs by means of dynamic scheduling; priority handling between logical channels of a UE by means of logical channel prioritization; priority handling between overlapping resources of a UE; and / or padding (see, for example, 3GPP TS 36.321v17.3.0 (2023-01-13) and 3GPP TS 38.321v17.4.0 (2023-03-29) ("[TS38321]").

[0333] The terms "physical layer", "PHY layer" or "PHY" refer, at least in some examples, to a protocol layer or sublayer that includes the capability to send and receive modulated signals for communication in a communication network (see, for example, 3GPP TS 36.201 v17.0.0 (2022-03-31) and 3GPP TS 38.201 v17.0.0 (2022-01-05) ("[TS38201]").

[0334] The term "access technology" refers, at least in some examples, to the technology used for the underlying physical connection to a communication network. The term "radio access technology" or "RAT" refers, at least in some examples, to the technology used for the underlying physical connection to a radio-based communication network. The term "radio technology" refers, at least in some examples, to the technology used for wireless transmission and / or reception of electromagnetic radiation for information transmission. The term "RAT type" may identify, at least in some examples, the transmission technology and / or communication protocol used in the access network. Examples of access technologies include wireless access technology / RAT, wired, wired cable, wired broadband forum (wireline-BBF), Ethernet (see, e.g., IEEE Standard for Ethernet, IEEE Std802.3-2018 (August 31, 2018) ("[IEEE8023]")) and its variants, fiber optic networks (e.g., ITU-T G.651, ITU-T G.652, Optical Transport Network (OTN), Synchronous Optical Networking (SONET) and Synchronous Digital Hierarchy (SDH), Digital Subscriber Line (DSL) and its variants, Data over Cable Service Interface Specification (DOCSIS) technology, Hybrid Fiber Coaxial (HFC) technology, etc. Examples of RATs (or RAT types) and / or communication protocols include Advanced Mobile Phone System (AMPS) technologies (e.g., Digital AMPS (D-AMPS), Total Access Communication System (TACS) and its variants (e.g., Extended TACS (ETACS)), etc.); Global System for Mobile Communications (GSM) technologies (e.g., Circuit Switched Data (CSD), High Speed ​​CSD (HSCSD), General Packet Radio Service (GPRS), and Enhanced Data Rates for GSM Evolution (EDGE)); Third Generation Partnership Project (3GPP) technologies (e.g., Universal Mobile Telecommunications System (UMTS) and its variants (e.g., UMTS Terrestrial Radio Access (UTRA), Wideband Code Division Multiple Access (W-CDMA), Freedom of Multi-Media Access (FOMA), Time Division-Code Division Multiple Access (TD-CDMA), Time Division-Synchronous Code Division Multiple Access (TD-SCDMA), etc.), Generic Access Network (GAN) / Unlicensed Mobile Access (UMA), High Speed ​​Packet Access (HSPA) and its variants (e.g., HSPA Plus (HSPA+)), Long Term Evolution (LTE) and its variants (e.g., LTE-Advanced (LTE-A), Evolved UTRA (E-UTRA), LTE Extra, LTE-A Pro, LTE LAA, MuLTEfire, etc.), fifth generation (5G) or new radio (NR), narrowband IoT (NB-IOT), 3GPP proximity services (ProSe), etc.); ETSI RAT (e.g., High Performance Radio Metropolitan Area Network (HiperMAN), Intelligent Transportation System (ITS) (e.g., ITS-G5, ITS-G5B, ITS-G5C, etc.), etc.); Institute of Electrical and Electronics Engineers (IEEE) technology and / or WiFi (e.g., IEEE Standard for Local and Metropolitan Area Networks: Overview and Architecture, IEEE Std 802-2014, pp.1-74 (June 30, 2014) ("[IEEE802]"), IEEEStandard for Information Technology--Telecommunications and InformationExchange between Systems-Local and Metropolitan Area Networks--SpecificRequirements-Part 11: Wireless LAN Medium Access Control(MAC) and PhysicalLayer(PHY)Specifications,IEEE Std 802.11-2020, pp.1-4379 (February 26, 2021) ("[IEEE80211]"), IEEE 802.15 technology (e.g., IEEE Standard for Low-Rate Wireless Networks, IEEE Std 802.15.4-2020, pp.1-800 (July 23, 2020) ("[IEEE802154]") and its variants (e.g., ZigBee, WirelessHART, MiWi, ISA100.11a, Thread, IPv6 over low-power WPAN (6LoWPAN)), IEEE Standard for Local and metropolitan area networks-Part15.6:Wireless Body Area Networks, IEEE Std 802.15.6-2012, pp.1-271 (February 29, 2012)), WLAN V2X RAT (e.g., IEEE Standard for Information technology--Local and metropolitan area networks--Specific requirements--Part 11:Wireless LAN Medium Access Control (MAC) and Physical Layer (PHY) Specifications Amendment 6:Wireless Access in Vehicular Environments, IEEE Std 802.11p-2010,pp.1-51 (July 15, 2010) ("[IEEE80211p]") (which is now part of [IEEE80211]), IEEE Guide for Wireless Access in Vehicular Environments (WAVE) Architecture, IEEE Standards Association, IEEE 1609.0-2019 (April 10, 2019) ("[IEEE16090]"), IEEE 802.11bd, Dedicated Short Range Communications (DSRC) and others), Worldwide Interoperability for Microwave Access (WiMAX) (e.g., IEEE Standard for Air Interface for Broadband Wireless Access Systems, IEEE Std 802.16-2017, pp. 1-2726 (March 2, 2018) ("[WiMAX]")), Mobile Broadband Wireless Access (MBWA) / iBurst (e.g., IEEE 802.20 and its variants), Wireless Gigabit Alliance (WiGig) standards (e.g., IEEE 802.11ad, IEEE 802.11ay, etc.); integrated digital enhanced network (iDEN) and its variants (e.g., broadband integrated digital enhanced network (WiDEN)); millimeter wave (mmWave) technology / standards (e.g., wireless systems operating above 10-300 GHz and 3GPP 5G); short-range and / or wireless personal area network (WPAN) technology / standards (e.g., IEEE 802.15 technology (e.g., as described above); Bluetooth and its variants (e.g., Bluetooth 5.3, Bluetooth Low Energy (BLE), etc.), WiFi-direct, Miracast, ANT / ANT+, Z-Wave, Universal Plug and Play (UPnP), low power wide area network (LPWAN), long range wide area network (LoRA or LoRaWAN. TM) etc.); optical and / or visible light communication (VLC) technologies / standards (e.g., IEEE Standard for Local and metropolitan area networks--Part 15.7: Short-Range Optical Wireless Communications, IEEE Std 802.15.7-2018, pp. 1-407 (April 23, 2019) etc.); Sigfox; Mobitex; 3GPP2 technologies (e.g., cdmaOne (2G), Code Division Multiple Access 2000 (CDMA) 2000) and Evolution Data Optimized or Evolution Data Only (EV-DO); Push-to-Talk (PTT), Mobile Telephone System (MTS) and its variants (e.g., Improved MTS (IMTS), Advanced MTS (AMTS), etc.); Personal Digital Cellular (PDC); Personal Handyphone System (PHS), Cellular Digital Packet Data (CDPD); Cellular Digital Packet Data (CDPD); DataTAC; Digital Enhanced Cordless Telecommunications (DECT) and its variants (e.g., DECT Ultra Low Energy (DECT ULE), DECT-2020, DECT-5G, etc.); Ultra High Frequency (UHF) communications; Very High Frequency (VHF) communications; and / or any other suitable RAT or protocol. In addition to the above RATs / standards, for the purposes of this disclosure, any number of satellite uplink technologies may be used, including, for example, radios that comply with standards published by the International Telecommunication Union (ITU), or ETSI, etc. The examples provided herein should therefore be understood to be applicable to a variety of other communication technologies, both existing and yet to be developed.

[0335] The term "channel" refers, at least in some examples, to any tangible or intangible transmission medium for transmitting data or data streams. The term "channel" may be synonymous with and / or equivalent to "communication channel", "data communication channel", "transmission channel", "data transmission channel", "access channel", "data access channel", "link", "data link", "carrier", "radio frequency carrier" and / or any other similar term representing a path or medium through which data is transmitted. Additionally, the term "link" refers, at least in some examples, to a connection between two devices over a RAT for the purpose of sending and receiving information.

[0336] The term "carrier" refers, at least in some examples, to a modulated waveform that conveys one or more physical channels (e.g., 5G / NR, E-UTRA, UTRA, and / or GSM / EDGE physical channels). The term "carrier frequency" refers, at least in some examples, to the center frequency of a cell.

[0337] The term "bearer" refers, at least in some examples, to an information transmission path of defined capacity, delay, bit error rate, etc. The term "radio bearer" refers, at least in some examples, to a service provided by Layer 2 (L2) for the transmission of user data between a user equipment (UE) and a radio access network (RAN). The term "radio access bearer" refers, at least in some examples, to a service provided by an access stratum to a non-access stratum for the transmission of user data between a UE and a CN.

[0338] The terms "beamforming" and "beam steering" refer, at least in some examples, to a spatial filtering mechanism used at a transmitter (Tx) to improve received signal power, signal-to-noise ratio (SNR), or some other signaling metric at an intended receiver (Rx). The term "beamformer" refers, at least in some examples, to a STA that transmits a physical layer PDU (PPDU) using a beamforming steering matrix. The term "beamforming steering matrix" refers, at least in some examples, to a matrix determined using knowledge of the channel between Tx and intended Rx that maps from space-time streams to transmit antennas with the goal of improving signal power, SNR, and / or some other signaling metric at intended Rx.

[0339] The term "subframe" refers, at least in some examples, to a time interval during which a signal is transmitted. In some implementations, a subframe is equal to 1 millisecond (ms). The term "time slot" refers, at least in some examples, to an integer multiple of consecutive subframes. The term "superframe" refers, at least in some examples, to a time interval comprising two time slots.

[0340] The term "channel coding" refers, at least in some examples, to a process and / or technique for adding redundancy to a message or packet so as to make it more robust against noise, channel interference, limited channel bandwidth, and / or other errors. For the purposes of this disclosure, the term "channel coding" may be used interchangeably with the following terms: "forward error correction" or "FEC"; "error correction coding", "error correction code" or "ECC"; and / or "network coding" or "NC". The term "network coding" refers, at least in some examples, to a process and / or technique for encoding and decoding transmitted data to improve network performance. The term "code rate" refers, at least in some examples, to the proportion of useful or non-redundant data streams or flows (e.g., for a code rate of k / n, for every k bits of useful information, the encoder generates a total of n bits of data, where n–k is redundant). The term "systematic code" refers, at least in some examples, to any error correction code in which the input data is embedded in the encoded output. The term "non-systematic code" refers, at least in some examples, to any error correction code in which the input data is not embedded in the encoded output. The term "interleaving" refers, at least in some examples, to the process of rearranging code symbols so that a burst of errors is spread over multiple code words that can be corrected by ECC. The terms "code word" or "codeword" refer, at least in some examples, to the elements of a code or protocol that are assembled according to the specific rules of the code or protocol.

[0341] The term "network address" refers, at least in some examples, to an identifier for a node or host in a computer network, and may be a unique identifier across a network and / or may be unique to a locally governed portion of a network. Examples of identifiers and / or network addresses may include an am application identifier, a Bluetooth hardware device address (BD_ADDR), a cellular network address (e.g., an access point name (APN), an AMF name and / or an AMF identifier (ID), an AF service identifier, a closed access group identifier (CAG-ID), an edge application server (EAS) ID, a data network access identifier (DNAI), a data network name (DNN), an EPS bearer identifier (EBI), an equipment identity register (EIR) and / or a 5G-EIR, an extended unique identifier (EUI), a network selection group ID (GIN), a general public subscription identifier (GPSI), a globally unique AMF identifier (GUAMI), a globally unique temporary identifier (GUTI) and / or a 5G-GUTI, a gNB identifier (gNB ID), a global gNB ID, International Mobile Equipment Identity (IMEI), IMEI Type Assignment Code (IMEA / TAC), International Mobile Subscriber Identity (IMSI), IMSI Software Version (IMSISV), Permanent Equipment Identifier (PEI), Local Area Data Network (LADN) DNN, Local NG-RAN Node Identifier, Mobile Subscriber Identity Number (MSIN), Mobile Subscriber / Station ISDN Number (MSISDN), Network Identifier (NID), NR Cell Global Identifier (NCGI), Network Slice Instance (NSI) ID, Network Slice AS Group (NSAG), Permanent Equipment Identifier (PEI), Public Land Mobile Network (PLMN) ID, Physical Cell Identifier (PCI), QoS Flow ID (QFI) and / or 5G QoS Identifier (5QI), RAN ID, Routing Indicator, Radio Network Temporary Identifier (RNTI) and its variants (e.g., 3GPP TS 38.300v17.4.0 (2023-03-28) (“anything discussed in clause 8 of [TS38300]) and / or previously discussed NCR-RNTI), SMS Function (SMSF) ID, Standalone Non-Public Network (SNPN) ID, Single Network Slice Selection Assistance Information (S-NSSAI), Sidelink Identifiers (e.g., Source Layer 2 ID, Target Layer 2 ID, PC5 Link Identifier, etc.), Subscription Hidden Identifier (SUCI), Subscription Permanent Identifier (SUPI), Temporary Mobile Subscriber Identity (TMSI) and its variants, Tracking Area Identity (TAI), UE Access Category and Identity and / or other cellular network related identifiers), CA G-ID, driver's license number, Global Trade Item Number (GTIN) (e.g., Australian Product Number (APN), EPC, European Article Number (EAN), Universal Product Code (UPC), etc.), email address, Enterprise Application Server (EAS) ID, endpoint address, Electronic Product Code (EPC) defined by the EPCglobal Tag Data Standard, Fully Qualified Domain Name (FQDN), flow ID and / or flow hash, hash, index, Internet Protocol (IP) address in an IP network (e.g., IP version 4 (IPv4), IP version 6 (IPv6), etc.), Internet Packet Exchange (IPX) address, LAN ID, MAC address, personal area network (PAN) ID, port number (e.g., TCP port number, UDP port number, etc.), price lookup code (PLC), product key, QUIC connection ID, RFID tag, serial number, service set identifier (SSID) and its variants, screen name, serial number, stock keeping unit (SKU), socket address, social security number (SSN), telephone number (e.g., in public switched telephone network (PTSN)), unique identifier (UID) (e.g., including global UID, universal unique identifier (UUID) (e.g., as specified in ISO / IEC 11578:1996), etc.), uniform resource locator (URL) and / or uniform resource identifier (URI), username (e.g., ID used to log into a service provider platform (e.g., a social network and / or some other service)), vehicle identification number (VIN), virtual LAN (VLAN) ID, X.21 address, X.25 address, . ID, The device network ID and / or any other suitable network address and components thereof.

[0342] The term "endpoint address" refers, at least in some examples, to an address (e.g., a URI and / or any other network address (e.g., an address discussed herein)) used to determine the host / authority portion of a target network address, where the target network address (e.g., a URI and / or any other network address (e.g., an address discussed herein)) is used to access a NF service of a NF service producer (e.g., to invoke a service operation) or for notification to a NF service consumer. The term "port" in the context of a computer network refers, at least in some examples, to a communication endpoint, a virtual point where a virtual data connection and / or network connection between two or more entities begins and ends. Additionally or alternatively, a "port" is associated with a specific process or service, at least in some examples. Additionally or alternatively, the term "port" refers, at least in some examples, to a specific interface of a specified device (apparatus) with an electromagnetic environment (e.g., any connection point on a device intended for a cable connection to or from the device is considered a port).

[0343] The term "delay" refers, at least in some examples, to the time interval between two events. Additionally or alternatively, the term "delay" refers, at least in some examples, to the time interval between the propagation of a signal and its reception. The term "delay bound" refers, at least in some examples, to a predetermined or configured amount of acceptable delay. The term "per-packet delay bound" refers, at least in some examples, to a predetermined or configured amount of acceptable packet delay, wherein packets that are not processed and / or sent within the delay bound are considered delivery failures and are discarded or abandoned. The term "effective throughput (goodput)" refers, at least in some examples, to the number of useful information bits delivered by a network to a specific destination per unit of time. The term "jitter" refers, at least in some examples, to the deviation of a predefined ("true") periodicity of a presumed periodic signal relative to a reference clock signal. The term "latency" refers, at least in some examples, to the amount of time it takes to transmit the first / initial data unit in a data burst from one point to another. Additionally or alternatively, the term "latency" refers, at least in some examples, to the delay experienced by a data unit (e.g., a frame) during its propagation between two points in a network, measured from the time a known reference point in the frame passes through a first point to the time a reference point in the data unit passes through a second point. The term "network delay" refers, at least in some examples, to the delay of a data unit within a network (e.g., an IP packet within an IP network). The term "packet delay" refers, at least in some examples, to the time it takes to transmit any packet from one point to another. Additionally or alternatively, the term "packet delay" or "per packet delay" refers, at least in some examples, to the difference between a packet reception time and a packet transmission time. Additionally or alternatively, "packet delay" or "per packet delay" can be measured by subtracting a packet transmission time from a packet reception time, where the transmitter and receiver are synchronized at least to some extent. The term "packet drop rate" refers, at least in some examples, to the share of packets that are not sent to a target due to high traffic load or traffic management and should be considered as part of the packet loss rate. The term "packet loss rate" refers, at least in some examples, to the share of packets that a target cannot receive, including discarded packets, packets lost in transmission, and packets received in an incorrect format. The term "performance indicator" refers, at least in some examples, to performance data aggregated over a group of network functions (NFs) derived from performance measurements collected from NFs belonging to a group according to an aggregation method identified in a performance indicator definition. The term "physical rate" or "PHY rate" refers, at least in some examples, to the speed at which one or more bits are actually sent over a transmission medium. Additionally or alternatively, the term "physical rate" or "PHY rate" refers, at least in some examples, to the speed at which data can move across a wireless link between a transmitter and a receiver. The term "processing delay" refers, at least in some examples, to the amount of time it takes to process a packet in a network node.The term "propagation delay" refers, at least in some examples, to the amount of time it takes for the head of a signal to travel from a transmitter to a receiver. The term "queuing delay" refers, at least in some examples, to the amount of time a job waits in a queue until it can be executed. Additionally or alternatively, the term "queuing delay" refers, at least in some examples, to the amount of time a packet waits in a queue until it can be processed and / or sent. The term "throughput" or "network throughput" refers, at least in some examples, to the rate at which something is combed or processed by production. Additionally or alternatively, the term "throughput" or "network throughput" refers, at least in some examples, to the rate at which successful message (dates) are delivered through a communication channel. The term "transmission delay" refers, at least in some examples, to the amount of time required (or necessary) to push a packet (or all bits of a packet) into a transmission medium.

[0344] The term "application" or "app" refers, at least in some examples, to a computer program designed to perform specific tasks other than those related to the operation of the computer itself. Additionally or alternatively, the term "application" or "app" refers, at least in some examples, to a complete and deployable software package, environment for achieving specific functions in an operating environment. The term "process" refers, at least in some examples, to an instance of a computer program executed by one or more threads. In some implementations, a process may consist of multiple threads of execution executing instructions concurrently. The term "algorithm" refers, at least in some examples, to an unambiguous specification of how to solve a problem or a class of problems by performing calculations, input / output operations, data processing, automated reasoning tasks, and the like.

[0345] The term "application programming interface" or "API" refers, at least in some examples, to a collection of subroutine definitions, communication protocols, and tools for building software. Additionally or alternatively, the term "application programming interface" or "API" refers, at least in some examples, to a collection of well-defined methods for communication between various components. In some examples, an API may be defined or otherwise used for a Web-based system, an operating system, a database system, computer hardware, a software library, and the like.

[0346] The terms "instantiate," "instantiation," and the like refer, at least in some examples, to the creation of an instance. In some examples, an "instance" refers to a specific occurrence of an object that may occur, for example, during execution of program code.

[0347] The term "reference point" refers, at least in some examples, to a conceptual point at which two non-overlapping functional groups, elements, or entities are joined. The term "service-based interface" refers, at least in some examples, to a representation of how a particular NF provides and / or exposes a set of services.

[0348] The term "use case" refers, at least in some examples, to a description of a system from the user's perspective. Use cases sometimes treat the system as a black box, with interactions with the system (including system responses) viewed as coming from outside the system. Use cases typically avoid technical jargon, preferring instead the language of end users or domain experts.

[0349] The term "user" refers, at least in some examples, to an abstract representation of any entity that issues commands, requests, and / or data to a computing node or system and / or otherwise consumes or uses services. Additionally or alternatively, the term "user" refers, at least in some examples, to an entity that uses 3GPP system services without being part of the 3GPP system (e.g., a person using a 3GPP system mobile station as a portable phone). The term "user profile" refers, at least in some examples, to a collection of information that provides a consistent, personalized service environment for a user, regardless of the user's location or the terminal used (within the limitations of the terminal and the service network).

[0350] The term "service consumer" or "consumer" refers, at least in some examples, to an entity that consumes one or more services. The term "service producer" or "producer" refers, at least in some examples, to an entity that supplies, services, or otherwise provides one or more services. The term "service provider" or "provider" refers, at least in some examples, to an organization or entity that provides one or more services to at least one service consumer. For the purposes of this disclosure, the terms "service provider" and "service producer" may be used interchangeably, even though these terms may refer to different concepts. Examples of service providers include cloud service providers (CSPs), network service providers (NSPs), application service providers (ASPs) (e.g., application software service providers (ASSPs) in a service-oriented architecture), Internet service providers (ISPs), telecommunication service providers (TSPs), online service providers (OSPs), payment service providers (PSPs), managed service providers (MSPs), storage service providers (SSPs), SAML service providers, and the like.

[0351] The terms "configuration," "policy," "rule set," and / or "operating parameters" refer, at least in some examples, to machine-readable information objects that contain instructions, conditions, parameters, guidelines, data, metadata, and / or other information related to components, devices, systems, networks, service producers, service consumers, and / or other elements / entities.

[0352] The term "datagram" refers, at least in some examples, to the basic transmission unit associated with a packet-switched network; a datagram may be structured to have a header segment and a payload segment. The term "datagram" may be synonymous with any of the following terms, at least in some examples, even though they may refer to different aspects: "data unit", "protocol data unit" or "PDU", "service data unit" or "SDU", "frame", "packet", "network packet", "segment", "block", "cell", "chunk", "type-length-value" or "TLV", etc. Examples of datagrams, network packets, etc. include Internet Protocol (IP) packets, Internet Control Message Protocol (ICMP) packets, UDP packets, TCP packets, SCTP packets, ICMP packets, Ethernet frames, RRC messages / packets, SDAP PDUs, SDAP SDUs, PDCP PDUs, PDCP SDUs, MAC PDUs, MAC SDUs, BAP PDUs. BAP SDU, RLC PDU, RLC SDU, WiFi frames discussed in IEEE 802 protocols / standards (e.g., [IEEE80211], etc.), type-length-value (TLV), and / or other similar data structures. The term "packet" refers, at least in some examples, to a unit of information identified by a tag at layer 3 of the OSI reference model. In some examples, a "packet" may also be referred to as a "network protocol data unit" or "NPDU". The term "protocol data unit" refers, at least in some examples, to a unit of data specified in a (N) protocol layer, and includes (N) protocol control information and possibly (N) user data.

[0353] The term "information element" or "IE" refers to a structural element containing one or more fields in at least some examples. Additionally or alternatively, the term "information element" or "IE" refers to a field or field set defined in a standard or specification for conveying data and / or protocol information in at least some examples. The term "field" refers to a separate content of an information element or a data element containing content in at least some examples. The term "data frame", "data field" or "DF" refers to a data type containing more than one data element in a predetermined order in at least some examples. The term "data element" or "DE" refers to a data type containing a single data in at least some examples. Additionally or alternatively, the term "data element" refers to an atomic state of a specific object having at least one specific property at a specific point in time, and may include a data element name or identifier, a data element definition, one or more representation items, an enumerated value or code (e.g., metadata) and / or a list of synonyms of data elements in other metadata registers. Additionally or alternatively, a "data element" refers to a data type containing a single data in at least some examples. A data element can store data that can be referred to as the content (or "content item") of a data element. Content items may include textual content, attributes, properties, and / or other elements referred to as "sub-elements". Additionally or alternatively, a data element may include zero or more properties and / or zero or more attributes, each of which may be defined as a database object (e.g., a field, a record, etc.), an object instance, and / or other data elements. "Attribute" refers, at least in some examples, to a markup construct comprising a name-value pair present within a start tag or an empty element tag. Attributes contain data associated with their element and / or control the behavior of the element.

[0354] The term "reference," at least in some examples, refers to data that can be used to locate other data, and can be implemented in various ways (eg, pointer, index, handle, key, identifier, hyperlink, etc.).

[0355] The term "data set" or "dataset" refers, at least in some examples, to a collection of data; a "data set" or "dataset" may be formed or arranged in any type of data structure. In some examples, one or more characteristics may define or affect the structure and / or properties of a data set (e.g., the number and type of attributes and / or variables) and various statistical measures (e.g., standard deviation, kurtosis, etc.). The term "data structure" refers, at least in some examples, to a data organization, management, and / or storage format. Additionally or alternatively, the term "data structure" refers, at least in some examples, to a collection of data values, relationships between these data values, and / or functions, operations, tasks, etc. that may be applied to the data. Examples of data structures include primitives (e.g., Boolean values, characters, floating point numbers, fixed point numbers, integers, references or pointers, enumeration types, etc.), composite types (e.g., arrays, records, strings, unions, tagged unions, etc.), abstract data types (e.g., data containers, lists, tuples, associative arrays, maps, dictionaries, sets (or data sets), multisets or bags, stacks, queues, graphs (e.g., trees, heaps, etc.), routing tables, symbol tables, quads, blockchains, purely functional data structures (e.g., stacks, queues, (multi)sets, random access lists, hash consing, zipper data structures, etc.).

[0356] The term "authorization" refers to the provision that a specific behavior must not be prevented, at least in some examples. The term "authentication" refers to the process of proving or verifying identification, at least in some embodiments. Additionally or alternatively, the term "authentication" refers to the mechanism by which a computer system checks or verifies that a user or entity is indeed the user or entity being claimed, at least in some embodiments. Examples of authentication and / or authorization techniques include the use of API keys, basic access authentication ("basic authentication"), open authorization (OAuth), hash-based message authentication code (HMAC), Kerberos protocol, OpenID, WebID and / or other authentication and / or authorization techniques. The term "consistency check" refers to a test or evaluation performed to determine whether data has any internal conflicts, conflicts with other data and / or whether there are any contradictions, at least in some examples. In some examples, a "consistency check" can be operated according to a "consistency model", which at least in some examples refers to a set of rules or policies for performing consistency checks and / or determining whether data is consistent (or predictable). The term "integrity" refers to a mechanism to ensure that data has not been changed in an unapproved manner, at least in some examples. Examples of cryptographic mechanisms that may be used for integrity protection include digital signatures, message authentication codes (MACs), and secure hashes.The term "verify" refers, at least in some examples, to a process, method, function, or any other means of confirming the correctness of information or data.

[0357] The term "certificate" or "digital certificate" refers to an information object (e.g., an electronic document or other data structure) used to prove the validity of a piece of data (e.g., a public key in a public key infrastructure (PKI) system) in at least some examples. Examples of digital certificates include X.509 format and / or some other suitable formats, and can be signed using any suitable encryption mechanism (e.g., elliptic curve encryption digital signature algorithm (ECDSA) or some other suitable algorithm (e.g., any algorithm discussed herein)). Additionally or alternatively, the digital certificates discussed herein may include various certificates issued by other entities defined by the issuer, verification subject, notified subject, certificate authority (CA) (e.g., root CA, etc.), registration authority (EA), certification authority (AA) and / or relevant certificate authority security committee (CASC) standards, common computing security standards forum (CCSF) standards, CA / browser forum standards, GSMA standards, ETSI standards, GlobalPlatform standards and / or some other suitable standards. The term "certificate revocation list" or "CRL" refers to a signature list of a collection of certificates that the certificate issuer no longer considers valid, at least in some examples. The terms "signature" or "digital signature" refer, at least in some examples, to a mathematical scheme, process, or method for verifying the authenticity of a digital message or information object (eg, an electronic document or other data structure).

[0358] The term "confidential data" refers, at least in some examples, to any form of information that a person or entity is legally or contractually obligated to protect from unauthorized access, use, disclosure, modification, or destruction. Additionally or alternatively, "confidential data" refers, at least in some examples, to any data owned or licensed by a person or entity that is not intended to be shared with the general public or classified by a person or entity with a designation that precludes sharing with the general public.

[0359] The term "encryption mechanism" refers, at least in some examples, to any encryption protocol and / or encryption algorithm. Examples of cryptographic mechanisms include cryptographic hash algorithms, such as functions in the Secure Hash Algorithm (SHA) 2 cryptographic hash algorithm (e.g., SHA-226, SHA-256, SHA-512, etc.) set, SHA 3, etc., or any type of keyed or unkeyed cryptographic hash function and / or any other function discussed herein; elliptic curve cryptography (ECC) algorithms (e.g., elliptic curve cryptographic key agreement algorithm (ECKA) algorithm, elliptic curve cryptographic digital signature algorithm (ECDSA), Lenstra elliptic curve factorization or elliptic curve factorization method (ECM), Menezes-Qu-Vanstone (MQV) or elliptic curve MQV (ECMQV), elliptic curve Diffie-Hellman (ECDH) key agreement, elliptic curve integrated encryption scheme (ECIES) or elliptic curve enhanced encryption scheme, Edwards curve digital signature algorithm (EdDSA)), etc.); Rivest-Shamir-Adleman (RSA) encryption; Merkle signature scheme, Advanced Encryption System (AES) algorithm; Triple Data Encryption Algorithm (3DES); quantum encryption algorithm; etc. Additionally or alternatively, the term "cryptographic protocol" refers, at least in some examples, to a sequence of steps that precisely specifies the actions required by two or more entities to achieve a particular security goal (e.g., a cryptographic protocol for key agreement). Additionally or alternatively, the term "cryptographic algorithm" refers, at least in some examples, to an algorithm that specifies the steps followed by a single entity to achieve a particular security goal (e.g., a cryptographic algorithm for symmetric key encryption). The term "public key encryption" or "asymmetric encryption" refers, at least in some examples, to an encryption system that uses an associated key pair including, for example, a public key for generating ciphertext and a corresponding private key for decrypting the ciphertext to obtain the original message (e.g., plaintext); in some examples, these key pairs are generated using a cryptographic algorithm based on a one-way function.

[0360] The term "cryptographic hash function," "hash function," or "hash") refers, at least in some examples, to a mathematical algorithm that maps data of arbitrary size (sometimes called a "message") to a fixed-size array of bits (sometimes called a "hash," "hash," or "message digest." A cryptographic hash function is typically a one-way function, which is a function that is practically infeasible to invert.

[0361] The term "encryption key" or "key" refers, at least in some examples, to a piece of information, typically a string of numbers or letters stored in a file, which, when processed through an encryption algorithm, can encode or decode encrypted data. The term "symmetric key algorithm" refers, at least in some examples, to an encryption algorithm that uses the same encryption key for both encryption of plain text and decryption of ciphertext; the key can be the same, or there can be a simple transformation between the two keys. The term "anchor key" refers, at least in some examples, to an encryption key used to generate other keys. In some examples, an "anchor key" is used in a key management system to create keys and distribute them to users. In some examples, an "anchor key" is stored in a secure location and is not used directly to encrypt or decrypt data. Examples of anchor keys include master keys, subkeys, and session keys.

[0362] The term "encryption" refers, at least in some examples, to the process of encoding an original representation of information (called "plain text") into an alternate form (called "cipher text"). In some examples, an encryption scheme includes using a pseudo-random encryption key generated by an encryption mechanism or some other algorithm to generate an encryption key that can be used to encrypt and / or decrypt plain text.

[0363] The term "one-time credential" refers, at least in some examples, to a type of authentication that is valid only for a single use. In some examples, one-time credentials are used for two-factor authentication (2FA), which is a security measure that requires two different forms of authentication to access an account. Examples of one-time credentials include a time-based one-time password (TOTP) (e.g., a one-time credential generated by a time-based algorithm that is valid for a short period of time (e.g., 30 seconds, etc.; in some examples, the TOTP is generated by a mobile app or hardware token)) and an out-of-band (OOB) one-time password (OTP) (e.g., a one-time credential sent to a user's phone (via an SMS message), an email address, etc.; in some examples, an OOB OTP is valid for a single use and can only be used once).

[0364] The term "data breach" refers, at least in some examples, to a breach of security that results in the accidental or unlawful destruction, loss, alteration, unauthorized disclosure, or access to transmitted, stored, or otherwise processed data (including personal, sensitive, and / or confidential data).

[0365] The term "information security" or "InfoSec" refers, at least in some examples, to any practices, techniques, and technologies used to protect information by mitigating information risks, and typically involves preventing or reducing the probability of unauthorized / inappropriate access to data, or the unlawful use, disclosure, disruption, deletion, damage, modification, inspection, recording, or devaluation of information; and the information to be protected can take any form, including electronic information, physical or tangible (e.g., computer-readable media storing information, paperwork, etc.) or intangible (e.g., knowledge, intellectual property assets, etc.).

[0366] The term "artificial intelligence" or "AI" refers, at least in some examples, to any intelligence exhibited by a machine, as contrasted to the natural intelligence displayed by humans and other animals. Additionally or alternatively, the term "artificial intelligence" or "AI" refers, at least in some examples, to the study of "intelligent agents" and / or any devices that perceive their environment and take actions to maximize their chances of successfully achieving a goal.

[0367] The term "artificial neural network", "neural network" or "NN" refers to an ML technique that includes a collection of connected artificial neurons or nodes that (loosely) model neurons in a biological brain that can send signals to other arterial neurons or nodes, where the connections (or edges) between artificial neurons or nodes are (loosely) modeled on synapses in a biological brain. Artificial neurons and edges typically have weights that are adjusted as learning proceeds. The weights increase or decrease the strength of the signal at the connection. Neurons can have thresholds so that a signal is sent only if the aggregate signal exceeds the threshold. Artificial neurons can be aggregated or grouped into one or more layers, where different layers can perform different transformations on their inputs. The signal travels from the first layer (input layer) to the last layer (output layer), possibly after traversing the layers multiple times. NNs are typically used for supervised learning, but can also be used for unsupervised learning. Examples of NN include deep NN (DNN), feedforward NN (FFN), deep FNN (DFF), convolutional NN (CNN), deep CNN (DCN), deconvolutional NN (DNN), deep confidence NN, perceptual NN, recurrent NN (RNN) (e.g., including long short-term memory (LSTM) algorithm, gated recurrent unit (GRU), echo state network (ESN), etc.), spiking NN (SNN), deep stacked network (DSN), Markov chain, perceptual NN, generative adversarial network (GAN), transformer, random NN (e.g., Bayesian network (BN), Bayesian belief network (BBN), Bayesian NN (BNN), deep BNN (DBNN), dynamic BN (DBN), probabilistic graphical model (PGM), Boltzmann machine, restricted Boltzmann machine (RBM), Hopfield network or HopfieldNN, convolutional deep belief network (CDBN), etc.), linear dynamic system (LDS), switching LDS (SLDS), optical NN (ONN), NN for reinforcement learning (RL) and / or deep RL (DRL), etc.

[0368] The term "mathematical model" refers, at least in some examples, to a system of assumptions, data, and reasoning proposed as a mathematical description of a state of affairs of an entity or event, including governing equations, assumptions, and constraints. The term "statistical model" refers, at least in some examples, to a mathematical model that embodies a set of statistical assumptions about generating sampled data and / or similar data from a population; in some examples, a "statistical model" refers to a data generating process.

[0369] The term "machine learning" or "ML" refers, at least in some examples, to the use of a computer system to optimize a performance criterion using example (training) data and / or past experience. ML involves the use of algorithms to perform specific tasks without the use of explicit instructions to perform specific tasks and / or relying on patterns, predictions, and / or inferences. ML uses statistics to build ML models (also called "models") to make predictions or decisions based on sampled data (e.g., training data).

[0370] The term "machine learning model" or "ML model" refers, at least in some examples, to an application, program, process, algorithm, and / or function that has the ability to make predictions, inferences, or decisions based on a set of input data and / or has the ability to detect patterns based on a set of input data. In some examples, a "machine learning model" or "ML model" is trained on training data to detect patterns and / or make predictions, inferences, and / or decisions. In some examples, a "machine learning model" or "ML model" is based on mathematical and / or statistical models. For the purposes of this disclosure, the terms "ML model", "AI model", "AI / ML model", etc. may be used interchangeably. For the purposes of this disclosure, the term "ML model" may be used interchangeably with the terms "AI / ML model" and "model".

[0371] The term "machine learning algorithm" or "ML algorithm" refers, at least in some examples, to an application, program, process, algorithm, and / or function that builds or estimates an ML model based on sampled data or training data. Additionally or alternatively, the term "machine learning algorithm" or "ML algorithm" refers, at least in some examples, to a program, process, algorithm, and / or function that learns from experience about some tasks and some performance measurements / metrics, and the ML model is an object or data structure created after training the ML algorithm with the training data. For the purposes of this disclosure, the terms "ML algorithm", "AI algorithm", "AI / ML algorithm", etc. may be used interchangeably. Additionally or alternatively, although the term "ML algorithm" may refer to a different concept from the term "ML model", for the purposes of this disclosure, these terms may be used interchangeably.

[0372] The term "machine learning application" or "ML application" refers, at least in some examples, to an application, program, process, algorithm, and / or functionality that includes some AI / ML models and application-level descriptions. Additionally or alternatively, the term "machine learning application" or "ML application" refers, at least in some examples, to a complete and deployable application and / or package that includes at least one ML model and / or other data that has the ability to achieve a specific function and / or perform a set of operations or tasks in an operating environment. For the purposes of this disclosure, the terms "ML application", "AI application", "AI / ML application", etc. may be used interchangeably.

[0373] The terms "model parameters" and / or "parameters" in the context of ML refer, at least in some examples, to values, features, and / or properties that are learned during training. Additionally or alternatively, "model parameters" and / or "parameters" in the context of ML refer, at least in some examples, to configuration variables that are internal to a model and whose values ​​can be estimated from given data. Models typically require model parameters when making predictions, whose values ​​define the skill of the model on a particular problem. Examples of such model parameters / parameters include weights (e.g., in ANNs); constraints; support vectors in support vector machines (SVMs); coefficients in linear regression and / or logistic regression; word frequency, sentence length, noun or verb distribution per sentence, number of specific character n-grams per word, vocabulary diversity, etc. for natural language processing (NLP) and / or natural language understanding (NLU); etc. The term "hyperparameter" refers, at least in some examples, to features, properties, and / or parameters for an ML process that cannot be learned during training. Hyperparameters are typically set before training occurs and can be used during the process to help estimate model parameters. Examples of hyperparameters include model size (e.g., in terms of memory space, bytes, number of layers, etc.); training data shuffling (e.g., whether to do so and by how much); number of evaluation instances, iterations, cycles (e.g., number of iterations or passes over the training data), or rounds; number of passes over the training data; regularization; learning rate (e.g., how quickly the algorithm reaches (converges to) optimal weights); learning rate decay (or weight decay); momentum; number of hidden layers; size of individual hidden layers; weight initialization scheme; dropout and gradient clipping thresholds; C and sigma values ​​for SVMs; k in k-nearest neighbors; number of branches in a decision tree; number of clusters in a clustering algorithm; vector size; word vector size for NLP and NLU; etc.

[0374] While many of the previous examples are provided by using specific cellular / mobile network terminology, including by using 4G / 5G 3GPP network components (or anticipated terahertz-based 6G / 6G+ technologies), it will be understood that these examples can be applied to many other deployments of wide-area and local wireless networks as well as integration of wired networks (including optical networks and associated optical fibers, transceivers, etc.). In addition, various standards (e.g., 3GPP, ETSI, etc.) may define various message formats, PDUs, containers, frames, etc. as including a series of optional or mandatory data elements (DEs), data frames (DFs), information elements (IEs), etc. However, it should be understood that the requirements of any particular standard should not limit the examples discussed herein, and thus, any combination of containers, frames, DFs, DEs, IEs, values, actions, and / or features are possible in various examples, including any combination of containers, DFs, DEs, values, actions, and / or features that are strictly required to be followed in order to comply with these standards, or any combination of containers, frames, DFs, DEs, IEs, values, actions, and / or features that are strongly recommended and / or used with or without the presence / absence of optional elements.

[0375] Aspects of the subject matter of the present invention may be referred to herein individually and / or jointly, for convenience only, and if more than one aspect is actually disclosed, it is not intended to voluntarily limit the scope of the present application to any single aspect or inventive concept. Therefore, although specific aspects have been shown and described herein, it should be understood that any arrangement calculated to achieve the same purpose may be replaced for the specific aspects shown. The present disclosure is intended to cover any and all adaptations or variations of various aspects. When browsing the above description, the combination of the above aspects and other aspects not specifically described herein will be apparent to those skilled in the art.

Claims

1. A method of operating a network function service producer (NFp), the method comprising: sending a storage request to an analytical data repository function (ADRF), the storage request comprising machine learning (ML) model information to be stored, wherein the ML model information comprises data related to a set of trained ML models; receiving a storage response from the ADRF, the storage response indicating that the ML model information has been stored; receiving an ML model provisioning request from a NF service consumer (NFc), the ML model provisioning request including an access token for accessing stored ML model information from the ADRF; and When the access token is correctly verified, an ML model provisioning response including an encryption key is sent to the NFc, wherein the encryption key is to be used by the NFc to decrypt the ML model information.

2. The method of claim 1, wherein: The ML model information includes the set of trained ML models, and The storage request causes the ADRF to store the set of trained ML models in the storage request.

3. The method of claim 1, wherein: The ML model information includes a set of references to respective storage locations from which corresponding trained ML models in a set of trained ML models can be accessed, and The storage request causes the ADRF to access the storage location to obtain the set of trained ML models.

4. The method according to claim 2-3, wherein: For each ML model in the set of trained ML models, the ML model information includes an ML model identifier (ID), one or more analysis IDs, a vendor ID, a binary signature of an ML application, an execution environment for the ML model, and a reference to acquisition information for generating authentication credentials.

5. The method according to claims 1 to 4, wherein: The method comprises: generating a security context for protecting the ML model information using a logic function or a named network function; encrypting the ML model information using one or more symmetric keys generated using the security context; and The storage request is generated to include the encrypted ML model information.

6. The method of claim 5, wherein: The security context includes an encryption key, an integrity key, and a corresponding security algorithm used to generate the encryption key and the integrity key.

7. The method according to claims 1 to 6, wherein: The ML model provisioning request includes one or more analysis IDs and an ADRFID associated with the stored ML information.

8. The method of claims 1-7, wherein: The ML model provisioning response includes a one-time credential for accessing the ML model information from the ADRF.

9. The method of claim 8, wherein: The one-time credential includes at least one of: a nonce, a message authentication code (MAC), a binary hash, a random number, a signature key, and a credential generated using the signature key.

10. The method of claim 9, wherein: The one-time credential is to be used by the NFc to obtain the ML model information from the ADRF.

11. The method of claims 1 to 10, wherein: The NFp is a network data analysis function (NWDAF) including a model training logic function (MTLF), and the NFc is a NWDAF including an analysis logic function (ANLF).

12. The method of claims 1 to 11, wherein: The method comprises: A VRF registration request is sent to a Vendor Repository Function (VRF) indicating a NF profile associated with the NFp.

13. The method of claim 12, wherein: The VRF registration request includes one or more ML model provisioning services supported by the NFp.

14. The method according to claims 12-13, wherein: The VRF registration request includes a set of analysis identifiers (IDs) corresponding to the set of trained ML models or ML model filtering information for the set of trained ML models by analysis ID.

15. The method of claims 12-14, wherein: The VRF registration request includes an ML model file format for the set of trained ML models and a model execution environment for the set of trained ML models.

16. The method of claims 12-15, wherein: The VRF registration request causes the VRF to send a network repository function (NRF) registration request to the NRF to inform the NRF of the NF profile of the VRF.

17. The method of claim 16, wherein: The NRF registration request includes a list of supported vendors, a list of ML models for each vendor in the vendor list, ML model filtering information for the set of trained ML models by analysis ID, and an ML model serialization format for the set of trained ML models.

18. The method of claims 12-17, wherein: The method comprises: receiving an ML model provisioning subscription request from the NFc, the ML model provisioning subscription request indicating a subset of ML models in the set of trained ML models that the NFc wants to access; and Based on the ML model provisioning subscription request, an ML model provisioning notification is sent to the NFc.

19. The method of claim 18, wherein: The ML model provisioning subscription request includes an access token generated by the NRF.

20. The method of claims 18-19, wherein: The ML model provisioning subscription request includes a set of analysis IDs, an ML model file serialization format for the ML model subset, and supported hardware platforms for the ML model subset.

21. The method of claim 20, wherein: For each ML model in the ML model subset, the ML model provisioning notification includes an information pair including a unique ML model ID and ML model information associated with each analysis ID requested by the NFc.

22. The method of claims 12-21, wherein: The method comprises: sending an ML model discovery request including a list of analysis IDs corresponding to the requested set of ML models to the VRF; and A list of candidate NWDAFs including MTLFs including a corresponding ML model in the requested set of ML models is received from the VRF.

23. The method of claims 12-22, wherein: The VRF is a logical function of the NWDAF, or the VRF is co-located with the NWDAF.

24. The method of claims 12-22, wherein: The VRF registration request includes a set of supported ML model provisioning services.

25. The method of claim 24, wherein: The VRF is a logical function of the NRF, or the VRF is co-located with the NRF.

26. The method of claims 12-25, wherein: The method comprises: Sending an NRF registration request indicating the NF profile associated with the NFp to an NRF.

27. The method of claim 26, wherein: The NF configuration file associated with the NFp includes supported ML model file serialization formats for the set of trained ML models available at the NFp.

28. The method of claims 26-27, wherein: The NF profile associated with the NFp is used to provide discovery information to the NFc.

29. The method of claims 12-28, wherein: The NFc and the NFp belong to the same supplier, or the NFc and the NFp belong to different suppliers.

30. The method of claims 12-29, wherein: The NFp is a first NWDAF containing MTLF, and the NFc is a second NWDAF containing MTLF or a NWDAF containing ANLF.

31. One or more computer-readable media comprising instructions, wherein the instructions are executable by a processor circuit to cause the processor circuit to perform the method of claims 1-30.

32. A computer program comprising instructions as claimed in claim 31.

33. An application programming interface defining functions, methods, variables, data structures and / or protocols for a computer program as claimed in claim 32.

34. An apparatus comprising circuitry loaded with the instructions of claim 31.

35. An apparatus comprising circuitry operable to execute the instructions of claim 31.

36. An integrated circuit comprising the processor circuit of claim 31 and one or more of the one or more computer-readable media of claim 31.

37. A computing system comprising one or more computer readable media as claimed in claim 31 and a processor circuit.

38. An apparatus comprising means for executing the instructions of claim 31.

39. A signal generated as a result of executing the instructions of claim 31.

40. A data unit generated as a result of executing the instructions of claim 31.

41. The data unit of claim 40, the data unit being a datagram, a network packet, a data frame, a data segment, a protocol data unit (PDU), a service data unit (SDU), a message, or a database object.

42. A signal encoded with a data unit as claimed in claim 40 or 41.

43. An electromagnetic signal carrying the instructions of claim 31.

44. An apparatus comprising means for performing the method of claims 1-30.

Citation Information

Cited By

  • Security authentication method, device, equipment, medium and program product

    CN119584126A

  • Analytics services built on ai / ML foundational service

    WO2026143409A1