EP4674096A1

EP4674096A1 - Life cycle management using ml model identification and ml functionality identification

Info

Publication number: EP4674096A1
Application number: EP24701607.4A
Authority: EP
Inventors: Sakira HASSAN; István Zsolt KOVÁCS; Keeth Saliya Jayasinghe LADDU
Original assignee: Nokia Technologies Oy
Current assignee: Nokia Technologies Oy
Priority date: 2023-02-27
Filing date: 2024-01-22
Publication date: 2026-01-07
Also published as: WO2024179749A1; KR20250152655A; CN120660331A; MX2025009949A; JP2026509772A

Abstract

A method comprising, requiring to use a machine learning, ML, feature based on using a ML Model appropriate for the ML feature; receiving an activation message in relation to at least one of an activation of one ML Model available and appropriate for the ML feature, the one ML Model associated with at least one Functionality, or an activation of one Functionality of at least one Functionality associated with one ML Model available and appropriate for the ML feature, wherein the activation message indicates that the one ML Model and/or the one Functionality is activated; based on the received activation message, starting to use the one ML Model; and reporting at least one of ML Model performance key performance indicators, KPIs, of the one ML Model, or Functionality performance KPIs of the one Functionality associated with the one ML Model.

Description

LIFE CYCLE MANAGEMENT USING ML MODEL IDENTIFICATION AND ML

FUNCTIONALITY IDENTIFICATION

DESCRIPTION

Technical Field

The present disclosure relates to a method and an apparatus for life cycle management (LCM) using machine learning (ML) model identification and ML functionality identification.

Background Art

The following description of background art may include insights, discoveries, understandings or disclosures, or associations, together with disclosures not known to the relevant prior art, to at least some examples of embodiments of the present disclosure but provided by the disclosure. Some of such contributions of the disclosure may be specifically pointed out below, whereas other of such contributions of the disclosure will be apparent from the related context.

In the last years, an increasing extension of communication networks, e.g. of wire based communication networks, such as the Integrated Services Digital Network (ISDN), Digital Subscriber Line (DSL), or wireless communication networks, such as the cdma2000 (code division multiple access) system, cellular 3^rd generation (3G) like the Universal Mobile Telecommunications System (UMTS), fourth generation (4G) communication networks or enhanced communication networks based e.g. on Long Term Evolution (LTE) or Long Term Evolution-Advanced (LTE-A), fifth generation (5G) communication networks, sixth generation (6G) communication networks, cellular 2^nd generation (2G) communication networks like the Global System for Mobile communications (GSM), the General Packet Radio System (GPRS), the Enhanced Data Rates for Global Evolution (EDGE), or other wireless communication system, such as the Wireless Local Area Network (WLAN), Bluetooth or Worldwide Interoperability for Microwave Access (WiMAX), took place all over the world. Various organizations, such as the European Telecommunications Standards Institute (ETSI), the 3^rd Generation Partnership Project (3GPP), Telecoms & Internet converged Services & Protocols for Advanced Networks (TISPAN), the International Telecommunication Union (ITU), 3^rd Generation Partnership Project 2 (3GPP2), Internet Engineering Task Force (IETF), the IEEE (Institute of Electrical and Electronics Engineers), the WiMAX Forum and the like are working on standards or specifications for telecommunication network and access environments.

In such context, in machine learning paradigm, the life cycle management (LCM) of a model encompasses around data processing, training, deployment, and monitoring. The LCM ensures the effectiveness and robustness of a model being used for complex tasks. Although machine learning solutions have already been adapted in 5G NR Air interface, the LCM interactions and signalling are yet to be discovered. As agreed in RAN1#110, the different LCM components are listed as follows.

For interoperability across different components of LCM and models in different nodes (UE, NW), 3GPP proposed two different LCM based identification: a) Model identification based LCM, b) Functionality identification based LCM. In RAN1#111 , the working assumption and the agreement for Model identification and Functionality identification was defined. Figures 1 and 2 illustrate an illustration of Model ID based LCM and Functionality (ID) based LCM, respectively, as proposed by Nokia in 3GPP RAN1#112. Apart from the LCM of an AI/ML model, 3GPP also identified three use cases as AI/ML application. Figure 3 represents an example of how these two LCM mechanisms can be applicable to one of the use cases. Rel-18 NR air interface use cases are: CSI feedback enhancement, Beam management, and Positioning enhancement. In addition to this, RAN1 also had agreement on LIE-NW collaboration depending on the signalling over air interface and model transfer, which is mentioned as follows.

In general, LCM procedures ensure the continuous integration and effectiveness of an AI/ML model throughout its full life cycle. Although in machine learning domain, LCM of the model is well defined, in 3GPP NR air interface, the architectural framework is still under discussion. During the discussion two different LCM mechanisms are being identified: model based LCM mechanism, and functionality based LCM mechanism. The former is to allow continuous development of a model and to ensure a smooth integration with other underlying operations enclosing the model. The later allows the continuous integration of an ML functionality within the UE and NW communication. However, it is yet to be discovered a) How to organize ML model and ML functionality to ensure smooth collaboration between UE(s) and NW? b) How to define the interaction between model based LCM and functionality based LCM? c) How to ensure the different collaboration levels (level x, level y and level z) using these two LCM mechanisms? d) How to provide a scalable structure to accommodate an increasing number of ML enabled features?

The example of ML-enabled Feature definition in Figure 3, shows only one of the several different ways to organise and define the use of the ML Models and ML Functionalities, within the overall ML Feature.

For example, another way is to define ML Functionalities such that one Functionality ID corresponds to one ML Feature (use case), and the different ML Models (identified by IDs) are defined for the different sub-use cases, scenario configurations (across the columns in Figure 3).

Another approach would be when there is one-to-one mapping between ML Functionality and ML Model, potentially both are identified with one single ID (one ML model in each column in Figure 3).

Variations are further possible by combining the above approaches.

These options lead to increased complexity for both UE specifications (UE capabilities, features, testability, requirements, etc.), for actual implementation solutions, and ultimately results in increased LCM procedures complexity.

Thus, there is need for improvement. Particularly, there is need for LCM using ML model identification and ML functionality identification.

LCM using ML model identification and ML functionality identification as disclosed herein may solve such problems.

It is therefore an object of the present disclosure to improve the prior art.

The following meanings for the abbreviations used in this specification apply:

2G Second Generation

3G Third Generation

3GPP 3^rd Generation Partnership Project

3GGP2 3^rd Generation Partnership Project 2

4G Fourth Generation

5G Fifth Generation

6G Sixth Generation

Al Artificial Intelligence

AP Access Point

BS Base Station

CDMA Code Division Multiple Access

CSI Channel State Information

DSL Digital Subscriber Line EDGE Enhanced Data Rates for Global Evolution

EEPROM Electrically Erasable Programmable Read-only Memory eNB Evolved Node B

ETSI European Telecommunications Standards Institute

FID Functionality ID gNB Next Generation Node B

GPRS General Packet Radio System

GSM Global System for Mobile communications

ID Identification

IEEE Institute of Electrical and Electronics Engineers

ISDN Integrated Services Digital Network

ITU International Telecommunication Union

KPI Key Performance Indicator

LCM Life Cycle Management

LTE Long Term Evolution

LTE-A Long Term Evolution-Advanced

MANETs Mobile Ad-Hoc Networks

MAC CE MAC Control Element

MID Model ID

ML Machine Learning

NB Node B

NW Network

RAM Random Access Memory

RAN Radio Access Network

ROM Read Only Memory

TISPAN Telecoms & Internet converged Services & Protocols for Advanced Networks

UE User Equipment

UMTS Universal Mobile Telecommunications System

UUID Universally Unique Identifier

UWB Ultra- Wideband

WCDMA Wideband Code Division Multiple Access

WiMAX Worldwide Interoperability for Microwave Access

WLAN Wireless Local Area Network

SUMMARY It is an objective of various examples of embodiments of the present disclosure to improve the prior art. Hence, at least some examples of embodiments of the present disclosure aim at addressing at least part of the above issues and/or problems and drawbacks.

Various aspects of examples of embodiments of the present disclosure are set out in the appended claims and relate to methods, apparatuses and computer program products relating to LCM using ML model identification and ML functionality identification.

The objective is achieved by the methods, apparatuses and non-transitory storage media as specified in the appended claims. Advantageous further developments are set out in respective dependent claims.

Any one of the aspects mentioned according to the appended claims enables LCM using ML model identification and ML functionality identification, thereby allowing to solve at least part of the problems and drawbacks as identified/derivable from above.

Thus, improvement is achieved by methods, apparatuses and computer program products enabling LCM using ML model identification and ML functionality identification.

In more detail, the present specification discloses LCM using ML model identification and ML functionality identification, which is advantageous over a proprietary solution, since, among other advantageous technical effects, there is achieved a reduced complexity for UE specifications and actual implementation solutions, as well as a reduced LCM procedures complexity.

Further advantages become apparent from the following detailed description.

BRIEF DESCRIPTION OF THE DRAWINGS

Some embodiments of the present disclosure are described below, by way of example only, with reference to the accompanying drawings, in which: Figure 1 shows proposed Model identification based LCM modules for ML model training and inference procedures [3GPP RAN1#112 ];

Figure 2 shows proposed Functionality identification based LCM modules for ML Functionality with ML inference mode procedures [3GPP RAN1#112];

Figure 3 shows an example of ML-enabled Feature definition and the usage of Functionality ID and Model ID with associated information;

Figure 4 shows different illustrations of a function (FID) or a collection of functions (FID1 , FID2, ,.) to be correlated with a model (MID) or a collection of models (MIDI , MID2);

Figure 5 (parts 1/2 and 2/2) shows Framework C: call flow of Functionality (ID) based and ML Model ID based LCM with Functionality and/or ML model switching in collaboration level z;

Figure 6 (parts 1/2 and 2/2) shows Framework B: call flow of Functionality (ID) based LCM monitoring and switching at UE with Framework B in collaboration level y;

Figure 7 (parts 1/2 and 2/2) shows Framework B: call flow of Functionality (ID) based LCM monitoring and switching at UE with Framework B in collaboration level z;

Figure 8 shows a flowchart illustrating steps corresponding to a method according to various examples of embodiments;

Figure 9 shows a flowchart illustrating steps corresponding to a method according to various examples of embodiments;

Figure 10 shows a block diagram illustrating an apparatus according to various examples of embodiments; and

Figure 11 shows a block diagram illustrating an apparatus according to various examples of embodiments.

DESCRIPTION OF EMBODIMENTS Basically, for properly establishing and handling a communication between two or more end points (e.g. communication stations or elements or functions, such as terminal devices, user equipments (UEs), or other communication network elements, a database, a server, host etc.), one or more network elements or functions (e.g. virtualized network functions), such as communication network control elements or functions, for example access network elements like access points (APs), radio base stations (BSs), relay stations, eNBs, gNBs etc., and core network elements or functions, for example control nodes, support nodes, service nodes, gateways, user plane functions, access and mobility functions etc., may be involved, which may belong to one communication network system or different communication network systems.

In the following, different exemplifying embodiments will be described using, as an example of a communication network to which examples of embodiments may be applied, a communication network architecture based on 3GPP standards for a communication network, such as a 5G/NR (or 6G/NR), without restricting the embodiments to such an architecture, however. It is obvious for a person skilled in the art that the embodiments may also be applied to other kinds of communication networks like 4G and/or LTE (or 5G, 6G or further “XG”) where mobile communication principles are integrated, e.g. Wi-Fi, worldwide interoperability for microwave access (WiMAX), Bluetooth®, personal communications services (PCS), ZigBee®, wideband code division multiple access (WCDMA), systems using ultra-wideband (UWB) technology, mobile ad-hoc networks (MANETs), wired access, etc.. Furthermore, without loss of generality, the description of some examples of embodiments is related to a mobile communication network, but principles of the disclosure can be extended and applied to any other type of communication network, such as a wired communication network or datacenter networking.

The following examples and embodiments are to be understood only as illustrative examples. Although the specification may refer to “an”, “one”, or “some” example(s) or embodiment(s) in several locations, this does not necessarily mean that each such reference is related to the same example(s) or embodiment(s), or that the feature only applies to a single example or embodiment. Single features of different embodiments may also be combined to provide other embodiments. Furthermore, terms like “comprising” and “including” should be understood as not limiting the described embodiments to consist of only those features that have been mentioned; such examples and embodiments may also contain features, structures, units, modules etc. that have not been specifically mentioned. A basic system architecture of a (tele)communication network including a mobile communication system where some examples of embodiments are applicable may include an architecture of one or more communication networks including wireless access network subsystem(s) and core network(s). Such an architecture may include one or more communication network control elements or functions, access network elements, radio access network elements, access service network gateways or base transceiver stations, such as a base station (BS), an access point (AP), a NodeB (NB), an eNB or a gNB, a distributed or a centralized unit (CU), which controls a respective coverage area or cell(s) and with which one or more communication stations such as communication elements or functions, like user devices (e.g. customer devices), mobile devices, or terminal devices, like a UE, or another device having a similar function, such as a modem chipset, a chip, a module etc., which can also be part of a station, an element, a function or an application capable of conducting a communication, such as a UE, an element or function usable in a machine-to-machine communication architecture, or attached as a separate element to such an element, function or application capable of conducting a communication, or the like, are capable to communicate via one or more channels via one or more communication beams for transmitting several types of data in a plurality of access domains. Furthermore, (core) network elements or network functions ((core) network control elements or network functions, (core) network management elements or network functions), such as gateway network elements/functions, mobility management entities, a mobile switching center, servers, databases and the like may be included.

The general functions and interconnections of the described elements and functions, which also depend on the actual network type, are known to those skilled in the art and described in corresponding specifications, so that a detailed description thereof is omitted herein. However, it is to be noted that several additional network elements and signaling links may be employed for a communication to or from an element, function or application, like a communication endpoint, a communication network control element, such as a server, a gateway, a radio network controller, and other elements of the same or other communication networks besides those described in detail herein below.

A communication network architecture as being considered in examples of embodiments may also be able to communicate with other networks, such as a public switched telephone network or the Internet. The communication network may also be able to support the usage of cloud services for virtual network elements or functions thereof, wherein it is to be noted that the virtual network part of the telecommunication network can also be provided by non-cloud resources, e.g. an internal network or the like. It should be appreciated that network elements of an access system, of a core network etc., and/or respective functionalities may be implemented by using any node, host, server, access node or entity etc. being suitable for such a usage. Generally, a network function can be implemented either as a network element on a dedicated hardware, as a software instance running on a dedicated hardware, or as a virtualized function instantiated on an appropriate platform, e.g., a cloud infrastructure.

Furthermore, a network element, such as communication elements, like a UE, a mobile device, a terminal device, control elements or functions, such as access network elements, like a base station (BS), an eNB/gNB, a radio network controller, a core network control element or function, such as a gateway element, or other network elements or functions, as described herein, (core) network management element or function and any other elements, functions or applications may be implemented by software, e.g. by a computer program product for a computer, and/or by hardware. For executing their respective processing, correspondingly used devices, nodes, functions or network elements may include several means, modules, units, components, etc. (not shown) which are required for control, processing and/or communication/signaling functionality. Such means, modules, units and components may include, for example, one or more processors or processor units including one or more processing portions for executing instructions and/or programs and/or for processing data, storage or memory units or means for storing instructions, programs and/or data, for serving as a work area of the processor or processing portion and the like (e.g. ROM, RAM, EEPROM, and the like), input or interface means for inputting data and instructions by software (e.g. floppy disc, CD-ROM, EEPROM, and the like), a user interface for providing monitor and manipulation possibilities to a user (e.g. a screen, a keyboard and the like), other interface or means for establishing links and/or connections under the control of the processor unit or portion (e.g. wired and wireless interface means, radio interface means including e.g. an antenna unit or the like, means for forming a radio communication part etc.) and the like, wherein respective means forming an interface, such as a radio communication part, can be also located on a remote site (e.g. a radio head or a radio station etc.). It is to be noted that in the present specification processing portions should not be only considered to represent physical portions of one or more processors, but may also be considered as a logical division of the referred processing tasks performed by one or more processors. It should be appreciated that according to some examples, a so-called “liquid” or flexible network concept may be employed where the operations and functionalities of a network element, a network function, or of another entity of the network, may be performed in different entities or functions, such as in a node, host or server, in a flexible manner. In other words, a “division of labor” between involved network elements, functions or entities may vary case by case.

Referring now to Figure 4, Figure 4 shows different illustrations of a function (FID) or a collection of functions (FID1 , FID2, ..) to be correlated with a model (MID) or a collection of models (MIDI , MID2), according to various examples of embodiments.

In Figure 4, Framework A may be regarded to represent prior art. With further regard to Figure 4, a UE is assumed to provide information to the gNB about at least the Functionalities supported (at the UE), and optionally also about the ML Model(s) supported (at the UE), wherein Functionalities and ML Model(s) are associated with a certain ML- enabled Feature, like e.g. depicted in Figure 3.

The ML Models may be identified by a model-ID, and optionally complemented by associated meta-information. The Functionality may be identified with Functionality ID, and optionally complemented by associated meta-information.

According to at least some examples of embodiments, with further regard to Figure 4, there can be several frameworks how FID(s) and MID(s) can be associated. As an example, there are visualized three different frameworks in Figure 4. In the framework A 410, the Functionality (ID) is linked/associated with a collection of ML Models (ID). These associated ML Models support the same Functionality. In framework B 420, a given ML Model (ID) is linked/associated with a collection of different Functionalities (ID). In this case, the associated Functionalities are supported by the same ML Model. In the framework C 430, the ML Models (ID) are independent of Functionalities (ID). In this framework any of the combinations of framework A 410 and B 420 are supported.

With the framework disclosed herein, it is aimed to mitigate above complexity problems, and provide a scalable structure, which can accommodate an increasing number of ML-enabled Features (Release 19, 6G) and of underlying ML Models. Figure 4 Framework B 420 and C 430 summarizes the two identified categories of solutions proposed in this specification. There is disclosed in this specification two possible associations between LCM procedures signalling for ML Model configurations corresponding to Framework B 420 and C 430 in Figure 4. The objective is to enable effective collaboration and signalling among UE(s) and NW throughout the course of ML enabled features. Some of the key concepts are highlighted below.

For Framework B 420 or C 430, according to various examples of embodiments:

A first node (e.g. UE) may provide information to a second node (e.g. gNB) about the association between each of its ML Model(s) and the Functionalities provided by these ML Models.

The first node may receive an activation message from the second node to activate one of the ML Models, wherein the activation may include at least the configuration of the desired Functionality output KPIs to be monitored by the second node.

After receiving the ML Model activation, the first node may start the appropriate ML Model management procedure (model (monitoring, activation, deactivation, switching) for the associated Functionalities.

The first node may report the configured Functionality output KPIs to the second node.

The second node may start the appropriate ML Functionality management procedure (model (monitoring, activation, deactivation, switching) based on the Functionality output KPIs received from the first node.

In this specification, there is introduce the following terminologies:

FID (Functionality ID): During the functionality identification procedure, a function or a set of related functions is uniquely identified by an ID, referred as FID. It is assumed this FID to be a string of bits, which may be a collection of functions (such as, activation, deactivation, switching) for a use case. However, the format of this FID is out of scope of this specification. An example format is shown further below.

MID (Model ID): During the model identification procedure, a model is uniquely identified by an ID, referred to as MID. This ID allows that model to be uniquely identified during the life cycle of the model. The format of this MID is also out of scope of this specification. An example format is shown further below. MID-LCM (MID based LCM): The life cycle management of a ML Model is identified by the MID.

FID-LCM (FID based LCM): The life cycle management of a Functionality is identified by the FID.

Furthermore, according to at least some examples of embodiments, the following is to be considered.

In Framework B 420, Model based LCM and Functionality based LCM are dependent, where a given ML Model is associated with different (a collection of) Functionalities. The interoperability between the Model LCM and Functionality LCM requires certain alignment in their configurations.

In such Framework B 420 according to various examples of embodiments, it is assumed that UE has a set of ML Model(s) and the NW requires a certain Functionality to be used/activated by the UE. The Functionality LCM is controlled by the NW while the ML Model LCM can be controlled by the UE or, alternatively by the NW. When the ML Model LCM and Functionality LCM operations are in different nodes their configurations still need to be aligned, made compatible.

In Framework C 430, Model based LCM and Functionality based LCM are completely independent. Therefore, in order to ensure efficient interoperability between these two LCM requires strict alignment (configuration, monitoring, etc) between the models and functionalities.

Call flow for framework C 430, according to various examples of embodiments.

Referring now to Figure 5 (parts 1/2 and 2/2; to be connected at A-A’ and B-B’), Figure 5 shows Framework C: call flow of Functionality (ID) based and ML Model ID based LCM with Functionality and/or ML model switching in collaboration level z.

The detailed example call flow is given in Figure 5. According to various examples of embodiments, it is assumed that UE has a functionality and requires a model from the NW. Both the model and the functionality will be controlled by the NW. However, LCM operations in both cases are independent to each other though the synchronization and compatibility of the operations need to be handled carefully and incorporated using the configuration and signalling.

With reference to Figure 5, the call flow steps are described below:

Step 0 (not shown): UE 500 indicates to the NW (e.g. represented by gNB 510) that it needs to use a particular ML feature (e.g. Beam management use case).

Step 1 (only for level z collaboration): NW transfers a preferred model with MID and associated meta information if the UE 500 has no appropriate models for the ML feature. Step 2: UE 500 determines the ML Functionality association for the received model.

Option 1

Step 3: If UE 500 accepts the ML Functionality, then UE 500 will sends ACK (e.g. an acknowledge message) to the NW in order to use that received model.

Step 4: UE 500 will configure its functionality for that model and exchange its functionality configuration to NW. Once it is ready, it will request NW to enable the ML feature.

Step 5: NW will enable/activate the ML functionality.

Step 6: UE 500 will start using the model.

Step 7-8: Since the Model based LCM is controlled by the NW, UE 500 will send periodic, aperiodic/triggered MID’s performance reports to NW. If NW decides to switch functionality based on the performance reported, then NW signals to switch models.

Step 9: If switching is needed then, UE 500 can decide to continue from Step 1 or fallback. Otherwise, continue step 10.

Step 10-12: Since the functionality based LCM is controlled by the NW, UE 500 will send periodic, aperiodic/triggered functionality reports to NW. If NW decides to switch functionality based on the performance reported, then NW signals to switch Functionalities. Option 2

Step 13: UE 500 rejects the association with its ML Functionality to the received ML model. Step 14: NW selects another model and continue Step 1 or fallback.

In the following, reference is made to call flow for framework B 420, according to various examples of embodiments.

Furthermore, according to at least some examples of embodiments, the following is to be considered. In Framework B 420, a model is associated/linked with one or more functionalities. Examples can be intermediate KPIs generated from a model. The following procedures can happen.

Framework B 420 can be implemented in either UE or NW.

If Framework B 420 is implemented in NW, then Functionality (ID) based LCM and model (ID) based LCM will be controlled by the NW fully.

If Framework B 420 is in UE, then, depending on the collaboration level, three possible interactions can happen: o During level x collaboration, Functionality and model LCM are transparent to NW. o During level y collaboration, Functionality needs to be indicated in order to enable the requirements, configurations, and KPIs related to a particular use case. Since, in this case, the functionality (a set of functionalities) is inter-linked with the model, therefore, model(s) need

■ to be registered to the NW,

■ activated by the NW, and then functionality (a set of functionalities) is

■ Activated by the NW

■ LCM management by the NW o During level z, UE may not have the model(s) needed for the particular use case and requests specified model(s) from NW. In this case,

■ model LCM is maintained by the NW,

■ functionality LCM is maintained by the NW.

Figures 6 and 7 show plausible call flows for Framework B where it is assumed that model LCM is maintained by UE but functionality based LCM is maintained by gNB. Since the gNB has more knowledge about the surrounding environment, it can help UE to use the model effectively through the functionality based LCM.

Referring now to Figure 6 (parts 1/2 and 2/2; to be connected at A-A’ and B-B’), Figure 6 shows Framework B: call flow of Functionality (ID) based LCM monitoring and switching at UE with Framework B in collaboration level y.

The call flow for collaboration level y in Framework B 420 is shown in Figure 6 and steps are described below, according to various examples of embodiments: Step 1 : UE 600 selects a model for ML enabled feature (eg. Beam management use case) and indicates and registers with NW (e.g. represented by gNB 610) regarding the model’s meta information (identified by MID) and functionality (or a set of functionality) information (identified by FID(s)) associated with the model.

Step 2: NW determines the meta information of functionality (the set of functionalities) FID(s) for the indicated ML enabled feature.

Step 3-4: NW sends and ACK signal (like. e.g. acknowledgement message) for FID(s). UE 600 configures FID options for the model.

Step 5: NW sends an FID activation via MAC CE.

Step 6-7: UE 600 starts using the model with the activated FID(s). LCM operations (such as model (monitoring, update)) related to the model are performed by UE and transparent to NW.

Step 8: UE 600 sends periodically/aperiodically/event-triggered performance KPI reports of FID(s) to the NW.

Step 9-10: NW evaluates the FID’s performance and if the performance is not satisfactory, then NW can send FID deactivation via MAC CE. NW can also send new FID activation via MAC CE.

Step 11 : If there is a functionality switch in UE 600, then step 6 operation continues or fallback.

Referring now to Figure 7 (parts 1/2 and 2/2; to be connected at A-A’ and B-B’), Figure 7 Framework B: call flow of Functionality (ID) based LCM monitoring and switching at UE with Framework B in collaboration level z.

The call flow for collaboration level z in Framework B 420 is shown in Figure 7 and steps are described below, according to at least some examples of embodiments:

Step 0 (not shown): UE 700 indicates NW (e.g. represented by gNB 710) for ML enabled feature (eg. Beam management use case).

Step 1 : NW transfer the model and the functionality (or a set of functionality) associated with the model. Model’s meta information is identified by MID and functionality (or a set of functionality) information is identified by FID(s), which is associated with the model.

Step 2: UE 700 determines the meta information of functionality (the set of functionalities) FID(s) for the indicated ML enabled feature. Step 3-4: UE 700 sends and ACK signal (like acknowledgement message) for both MID and FID(s). This signal ensures that UE 700 is able to use the model and UE 700 has the support for the functionality configuration indicated in the FID(s). UE 700 configures FID options for the model.

Step 5-6: NW sends an FID activation and an MID activation signals via MAC CE to UE.

Step 6-7: UE 700 starts using the model with the activated FID(s). LCM operations related to model (such as model (monitoring, update)) and LCM operations related to functionality (such as functionality monitoring, switching) are performed by NW.

Step 8: UE 700 sends periodically/aperiodically/event-triggered performance KPI reports of both MID and FID(s) to the NW.

Step 11 : If there is a functionality switch in UE 700, then step 7 operation continues or fallback.

Step 12: As NW is also monitoring the model, it may require update of the model. In this case, NW may trigger model switching.

Step 13-14: NW first sends FID(s) deactivation signal via MAC CE, and then sends MID deactivation signal via MAC CE.

Step 15: If the model switch indication is received by UE 700, then it requires to use another model. If the model is not available in UE 700, then step 1 continues.

For completeness there is provided examples of elements in FID and MID, but the contents are not limited to these.

FID can contain at least one of the following: a) Functionality ID: a label/tag/UUlD which can uniquely identify the Functionality and the combination of the items b) - g) below. b) Additional ID/label/tags(s): a. If the model identification is applied, each of additional ID/label corresponds to a Model ID. b. Otherwise, each of additional ID/label corresponds to a Model to be monitored, which can identify performance variations of the Functionality supported by the Model. c) Applicable scenario/configuration/parameters/conditions that the model functionality is enabled for: including system and intermediary KPIs to be used for functionality-based LCM purposes. d) Input data type/source and preparation/pre-processing: including an indication on any delay-sensitive ML-specific data processing to be performed e.g., as an indication of the expected delay budget for such operation e) Non-ML operation(s) (optional): indication of any non-ML operations/algorithms involved in the model functionality e.g., as an indication of the expected delay budget for such operations f) Output data and post-processing (optional): including an indication on any delaysensitive ML-specific output post-processing is performed e.g., as an indication of the expected delay budget for such operation g) Specific control signaling configuration(s) which enable and (partially) control the b) - f) operations and the corresponding Functionality-based LCM

MID can contain at least one of the following: a) Model ID: a label/tag/UUlD which can uniquely identify the ML Model (implementation version, etc.) either within the ML-enabled Feature (across several potential Functionalities) or only within a specified Functionality, and b) The associated information related to the AI/ML model i. When proprietary ML model format is used, the associated information may include:

• Potential additional meta information required by the Functionality-based LCM ii. When open ML model format is used, the associated information may include:

• Model input data (dimensions, features) and preparation/pre-processing including indication on any feature extraction, feature selection or any other delay sensitive ML-specific data processing is performed e.g., as an indication of the expected delay budget for such operation

• Model output data (dimensions, features) and post-processing (optional) including indication on any delay sensitive ML-specific output postprocessing is performed e.g., as an indication of the expected delay budget for such operation

• Specific control signaling configuration(s) which enable and (partially) control the i) - ii) operations and the corresponding Model ID based LCM In the following, further examples of embodiments are described in relation to the above described methods and/or apparatuses.

Referring now to Figure 8, there is shown a flowchart illustrating steps corresponding to a method according to various examples of embodiments. Such method steps as illustrated in Figure 8 may represent at least part of such method/processing steps as outlined above with reference to Figures 4 to 7. Further, such method as illustrated in Figure 8 may be applied at such UE 500, 600 and/or 700 as outlined above with reference to Figures 5 to 7.

In particular, according to Figure 8, in S810, the method comprises requiring to use a ML feature based on using a ML Model appropriate for the ML feature.

It shall be noted that such requiring may represent such need to use a particular ML feature as outlined above with reference to Figures 5 to 7, e.g. Steps 0 and/or 1. Accordingly, such ML feature may e.g. be a beam management use case. Moreover, a ML Model may represent such transferred/received and/or selected (preferred) ML Model as outlined above with reference to Figures 5 to 7, e.g. Step 1. The ML Model may be represented and/or identified by such MID as outlined above with reference to Figures 4 to 7.

Further, in S820, the method comprises receiving an activation message in relation to at least one of:

- an activation of one ML Model available and appropriate for the ML feature, the one ML Model associated with at least one Functionality, or

- an activation of one Functionality of at least one Functionality associated with one ML Model available and appropriate for the ML feature.

The activation message indicates that the one ML Model and/or the one Functionality is activated.

It shall be noted that such receiving of an activation message may represent at least part of such enabling/activating as outlined above with reference to Figures 5 to 7, e.g. Steps 5 and 6. Hence, the activation message may be provided by an access network element, like e.g. such gNB 510, 610 and/or 710 as outlined above with reference to Figures 5 to 7. Moreover, such one ML Model and/or such one Functionality to be activated may represent such transferred/received and/or selected (preferred) ML Model and/or (ML) Functionality as outlined above with reference to Figures 4 to 7, e.g. Steps 5 and 6. Furthermore, the term “available” may be understood in that a ML Model and/or Functionality is available and ready to be used at an apparatus, like e.g. an endpoint terminal, which may be represented by such UE 500, 600 and/or 700 as outlined above with reference to Figures 5 to 7. Additionally, the expression “at least one Functionality” may e.g. comprise such FID1 and FID2 as outlined above with reference to Figure 4 and/or such (ML) Functionalities as outlined above with reference to Figures 5 to 7. Accordingly, the one ML Model may correspond to such MID as outlined above with reference to Figures 4 to 7.

Additionally, in S830, the method comprises starting to use the one ML Model, based on the received activation message.

It shall be noted that such starting may represent at least part of such starting as outlined above with reference to Figures 5 to 7, e.g. Steps 6 and 7.

Further, in S840, the method comprises reporting at least one of ML Model performance key performance indicators, KPIs, of the one ML Model, or Functionality performance KPIs of the one Functionality associated with the one ML Model.

It shall be noted that such reporting may represent at least part of such reporting/sending of (performance) reports as outlined above with reference to Figures 5 to 7, e.g. Steps 7 and 10 in Figure 5, Step 8 in Figure 6, and Step 8 in Figure 7.

Moreover, according to at least some examples of embodiments, the method may further comprise configuring the one Functionality of the at least one Functionality provided by the one ML Model; and providing a Functionality configuration resulting from the configuring. The Functionality configuration is indicative of at least one of the Functionality performance KPIs to be reported, or the ML Model performance KPIs to be reported. The method may further comprise requesting at least one of the activation of the one ML Model, or the activation of the one Functionality; and receiving the activation message in response to the requesting.

It shall be noted that such configuring may represent such configuring as outlined above with reference to Figure 5, Step 4. Further, such requesting may represent such requesting as outlined above with reference to Figure 5, Step 4. Furthermore, according to various examples of embodiments, the method may further comprise receiving a ML Model appropriate for the ML feature; and determining a Functionality association for the received ML Model, the Functionality association representing an association between the received ML Model and at least one Functionality available. If accepting the determined Functionality association, the method may further comprise providing an acknowledgement message indicating that the received ML model is available. If rejecting the determined Functionality association, the method may further comprise providing a rejection message indicating that the received ML model is not available.

It shall be noted that such receiving may represent such receiving as outlined above with reference to Figure 5, Step 1. Further, such determining may represent such determining as outlined above with reference to Figure 5, Step 2. Such accepting may represent such accepting as outlined above with reference to Figure 5, Step 3. Such rejecting may represent such rejecting as outlined above with reference to Figure 5, Step 13.

Additionally, according to various examples of embodiments, the method may further comprise, if the determined Functionality association being accepted, configuring one Functionality of the at least one Functionality accepted to be associated with the received ML Model; and providing a Functionality configuration resulting from the configuring. The Functionality configuration is indicative of at least one of the Functionality performance KPIs to be reported, or the ML Model performance KPIs to be reported. The method may further comprise requesting at least one of the activation of the received ML Model representing the one ML Model, or the activation of the one Functionality of the at least one Functionality accepted to be associated with the received ML Model representing the one Functionality associated with the one ML Model. The method may further comprise receiving the activation message in response to the requesting.

It shall be noted that such configuring may represent such configuring as outlined above with reference to Figure 5, Step 4. Further, such requesting may represent such requesting as outlined above with reference to Figure 5, Step 4.

Optionally, according to at least some examples of embodiments, the method may further comprise at least one of, in response to the reporting, receiving a signalling to switch the one ML model; and deciding whether to use another ML Model available and appropriate for the ML feature, to await reception of a new ML Model appropriate for the ML feature, or to fallback, or receiving a signalling to switch the one Functionality; and based on the determined Functionality association, configuring another Functionality.

It shall be noted that such receiving may represent such receiving as outlined above with reference to Figure 5, Steps 7 to 8 and Steps 10 to 12. Further, such deciding may represent such deciding as outlined above with reference to Figure 5, Step 9.

Further, according to various examples of embodiments, the method may further comprise selecting a ML Model available and appropriate for the ML feature; indicating the selected ML Model; and registering with regard to information in relation to at least the selected ML Model and at least one Functionality associated with the selected ML Model. The method may further comprise receiving an acknowledgement signal for the at least one Functionality associated with the selected ML Model; configuring one Functionality of the at least one Functionality associated with the selected ML Model; wherein the configuring comprises configuration of at least one of the Functionality performance KPIs to be reported, or the ML Model performance KPIs to be reported; and receiving the activation message, wherein the selected ML Model represents to one ML Model.

It shall be noted that such selecting, indicating and registering may represent such selecting, indicating and registering as outlined above with reference to Figure 6, Step 1. Further, such receiving may represent such receiving as outlined above with reference to Figure 6, Steps 3 to 5. Furthermore, such configuring may represent such configuring as outlined above with reference to Figure 6, Steps 3 to 4.

Moreover, according to at least some examples of embodiments, wherein the method may further comprise controlling life cycle management, LCM, operations related to the selected ML Model.

It shall be noted that such controlling may represent such controlling as outlined above with reference to Figure 6, Steps 6 to 7.

Furthermore, according to various examples of embodiments, the method may further comprise, in response to the reporting, receiving a deactivation for the Functionality associated with the selected ML Model, or in response to the reporting, receiving a deactivation for the Functionality associated with the selected ML Model and receiving an activation for another Functionality of the at least one Functionality associated with the selected ML Model.

It shall be noted that such receiving may represent such receiving as outlined above with reference to Figure 6, Steps 9 to 10.

Additionally, according to various examples of embodiments, the method may further comprise receiving a ML Model appropriate for the ML feature and at least one Functionality associated with the received ML Model; and determining information of the received at least one Functionality in relation to the ML feature. The method may further comprise providing an acknowledgement signal for both the received ML Model and the received at least one Functionality, wherein the acknowledgement signal indicates that both the received ML Model and the received at least one Functionality is available. The method may further comprise configuring one Functionality of the received at least one Functionality; wherein the configuring comprises configuration of at least one of the Functionality performance KPIs to be reported, or the ML Model performance KPIs to be reported; and receiving the activation message, wherein the received ML Model represents the one ML Model and wherein the received one Functionality represents the one Functionality.

It shall be noted that such receiving may represent such receiving as outlined above with reference to Figure 7, Step 1. In addition, such determining may represent such determining as outlined above with reference to Figure 7, Step 2. Further, such providing may represent such providing as outlined above with reference to Figure 7, Steps 3 to 4. Moreover, such configuring may represent such configuring as outlined above with reference to Figure 7, Steps 3 to 4. Further, such receiving may represent such receiving as outlined above with reference to Figure 7, Step 5.

Optionally, according to at least some examples of embodiments, the method may further comprise in response to the reporting, receiving a deactivation for the received Functionality, or in response to the reporting, receiving a deactivation for the received Functionality and receiving an activation for another Functionality of the received at least one Functionality.

It shall be noted that such receiving may represent such receiving as outlined above with reference to Figure 7, Steps 9 to 10. Further, according to various examples of embodiments, the method may further comprise, in response to the reporting, receiving an indication that an update of the received ML Model is required; receiving a deactivation for the received Functionality; receiving a deactivation for the received ML Model; and performing one of using another ML Model available and appropriate for the ML feature, or awaiting reception of a new ML Model appropriate for the ML feature and of at least one Functionality associated with the new ML Model.

It shall be noted that such receiving may represent such receiving as outlined above with reference to Figure 7, Steps 12 to 14. Further, such performing may represent such performing as outlined above with reference to Figure 7, Step 15.

Further, according to various examples of embodiments, the method may further comprise providing association information about an association between at least one ML Model available and at least one Functionality provided by the at least one ML Model. Accordingly, in case such association information are obtained at/for an endpoint terminal and provided to e.g. an access network element, like e.g. a gNB, the access network element (i.e. the network) may be enabled to identify the ML Model(s) and associated at least one Functionality available, i.e. ready to be used, at such endpoint terminal. Further accordingly, the network may use such association information determine, whether or not to perform actions at such endpoint terminal, like e.g. updating at least one of available ML Models or available Functionalities, or like e.g. transferring/providing additional ML Models and/or Functionalities to be available at the endpoint terminal, in case e.g. need may be for using a particular ML feature.

Further, according to various examples of embodiments, the method may further comprise that controlling the LCM operations (in relation to the ML Model and/or the Functionality) may comprise at least one of monitoring the ML Model and/or the Functionality, activating the ML Model and/or the Functionality, deactivating the ML Model and/or the Functionality, or switching the ML Model and/or the Functionality.

The above-outlined solution allow for LCM using ML model identification and ML functionality identification. Therefore, the above-outlined solution is advantageous in that it enables for more efficient and/or more secure and/or more robust and/or failure resistant and/or flexible and/or complexity reduced LCM using ML model identification and ML functionality identification

Referring now to Figure 9, Figure 9 shows a flowchart illustrating steps corresponding to a method according to various examples of embodiments. Such method steps as illustrated in Figure 9 may represent at least part of such method/processing steps as outlined above with reference to Figures 4 to 7. Further, such method as illustrated in Figure 9 may be applied at an access network element, like e.g. such gNB 510, 610 and/or 710 as outlined above with reference to Figures 5 to 7.

It shall be noted that similar terms/expressions used in Figures 8 and 9 may be understood similarly and a repetitive explanation thereof may thus be omitted.

In particular, according to Figure 9, in relation to a machine learning, ML, feature required to be used based on usage of a ML Model appropriate for the ML feature, in S910, the method comprises providing an activation message in relation to at least one of an activation of one ML Model available and appropriate for the ML feature, the one ML Model associated with at least one Functionality, or an activation of one Functionality of at least one Functionality associated with one ML Model available and appropriate for the ML feature.

It shall be noted that such providing may represent such providing/transfer as outlined above with reference to Figures 5 to 7, e.g. Steps 5 to 6. Furthermore, the term “available” may be understood in that a ML Model and/or Functionality is available and ready to be used at an apparatus, like e.g. an endpoint terminal, which may be represented by such UE 500, 600 and/or 700 as outlined above with reference to Figures 5 to 7.

Further, in S920, the method comprises activating the one ML Model and/or the one Functionality.

It shall be noted that such activating may represent such enabling/activating as outlined above with reference to Figures 5 to 7, e.g. Steps 5 to 6.

Additionally, in S930, the method comprises receiving a report indicative of at least ML Model performance key performance indicators, KPIs, of the one ML Model, or Functionality performance KPIs of the one Functionality associated with the one ML Model. It shall be noted that such receiving may represent at least part of such receiving as outlined above with reference to Figures 5 to 7, e.g. Steps 7 to 8 and 10 to 12.

Further, in S940, the method performing action in relation to the one ML Model and/or the one Functionality, comprises based on the received report.

It shall be that such performing action may represent at least part of such performing action as derivable from above with reference to Figures 5 to 7, e.g. Steps 7 to 14, like e.g. actions to be performed by the NW as recited in Figures 5 to 7

Moreover, according to at least some examples of embodiments, the method may further comprise receiving a Functionality configuration indicative of at least one of the Functionality performance KPIs included in the report to be received, or the ML Model performance KPIs included in the report to be received; receiving a request about at least one of the activation of the one ML Model, or the activation of the one Functionality; and providing the activation message in response to the received request.

It shall be noted that such receiving may represent such receiving as outlined above with reference to Figure 5, Step 4. Further, such providing may represent such providing as outlined above with reference to Figure 5, Step 4.

Furthermore, according to various examples of embodiments, the method may further comprise providing a ML Model appropriate for the ML feature; and if an association between the provided ML Model and at least one Functionality available is accepted, receiving an acknowledgement message indicating that the provided ML model is available, wherein if the association is rejected, receiving a rejection message indicating that the provided ML model is not available.

It shall be noted that such providing may represent such providing as outlined above with reference to Figure 5, Step 1. Further, such receiving may represent such receiving as outlined above with reference to Figure 5, Step 3 and 13.

Additionally, according to various examples of embodiments, the method may further comprise, if the association being accepted, receiving a Functionality configuration resulting from a configuration of one Functionality of the at least one Functionality accepted to be associated with the provided ML Model , wherein the Functionality configuration is indicative of at least one of the Functionality performance KPIs included in the report to be received, or the ML Model performance KPIs included in the report to be received; receiving a request about at least one of the activation of the provided ML Model representing the one ML Model, or the activation of the one Functionality of the at least one Functionality accepted to be associated with the provided ML Model representing the one Functionality associated with the one ML Model; and providing the activation message in response to the received request.

It shall be noted that such receiving may represent such receiving as outlined above with reference to Figure 5, Step 4.

Optionally, according to at least some examples of embodiments, the method may further comprise at least one of

- controlling a life cycle management, LCM, operation in relation to the one ML model, or controlling a LCM operation in relation to the one Functionality;

- wherein the performing action may further comprise at least one of, based on the controlled LCM operation in relation to the one ML model, deciding whether or not to switch the one Functionality; and if deciding to switch the one Functionality, providing a signalling to switch the one ML model, or based on the controlled LCM operation in relation to the one Functionality, deciding whether or not to switch the one Functionality; and if deciding to switch the one Functionality, providing a signalling to switch the one Functionality.

It shall be noted that such LCM controlling may represent such LCM controlling as outlined above with reference to Figure 5, Steps 7 to 8 and Steps 10 to 12. Further, such deciding may represent such deciding as outlined above with reference to Figure 5, Steps 7 to 8.

Further, according to various examples of embodiments, the method may further receiving an indication in relation to a selected ML Model available and appropriate for the ML feature; receiving a registration with regard to information in relation to at least the selected ML Model and at least one Functionality associated with the selected ML Model; determining the information for which the registration was received; based on the determining, providing an acknowledgement signal for the at least one Functionality associated with the selected ML Model; and providing the activation message, wherein the selected ML Model represents to one ML Model.

It shall be noted that such receiving may represent such receiving as outlined above with reference to Figure 6, Step 1. Further, such determining may represent such determining as outlined above with reference to Figure 6, Step 2. Further, such providing may represent such providing as outlined above with reference to Figure 6, Steps 3 to 5.

Moreover, according to at least some examples of embodiments, wherein the performing action may further comprise at least one of providing a deactivation for the Functionality associated with the selected ML Model, or providing a deactivation for the Functionality associated with the selected ML Model and providing an activation for another Functionality of the at least one Functionality associated with the selected ML Model.

It shall be noted that such providing may represent such providing as outlined above with reference to Figure 6, Steps 9 to 10.

Furthermore, according to various examples of embodiments, the method may further comprise providing a ML Model appropriate for the ML feature and at least one Functionality associated with the provided ML Model; receiving an acknowledgement signal for both the provided ML Model and the provided at least one Functionality, wherein the acknowledgement signal indicates that both the provided ML Model and the provided at least one Functionality is available; and providing the activation message, wherein the provided ML Model represents the one ML Model and wherein the provided one Functionality represents the one Functionality.

It shall be noted that such providing and receiving may represent such providing and receiving as outlined above with reference to Figure 7, Steps 1 to 5.

Additionally, according to various examples of embodiments, the method may further comprise controlling LCM operations at least related to the provided ML Model and the provided one Functionality; and wherein the performing action comprises at least one of, based on the controlled LCM operations, providing a deactivation for the provided Functionality, or providing a deactivation for the provided Functionality and providing an activation for another Functionality of the provided at least one Functionality. It shall be noted that such providing may represent such providing as outlined above with reference to Figure 7, Steps 9 to 10.

Optionally, according to at least some examples of embodiments, the method may further controlling LCM operations at least related to the provided ML Model and the provided one Functionality; and wherein the performing action comprises at least one of, based on the controlled LCM operations, triggering to switch the one ML Model by providing an indication that an update of the provided ML Model is required; providing a deactivation for the provided Functionality; and providing a deactivation for the provided ML Model.

It shall be noted that such triggering and providing may represent such triggering and providing as outlined above with reference to Figure 7, Steps 12 to 14.

Further, according to various examples of embodiments, the method may further comprise receiving such association information as outlined above with reference to Figure 8. Accordingly, in case such association information are obtained at/for an endpoint terminal and received to e.g. an access network element, like e.g. a gNB, the method may further comprise identifying the ML Model(s) and associated at least one Functionality available, i.e. ready to be used, at such endpoint terminal. Further accordingly, the method may further comprise using such association information to determine, whether or not to perform actions at such endpoint terminal, like e.g. updating at least one of available ML Models or available Functionalities, or like e.g. transferring/providing additional ML Models and/or Functionalities to be available at the endpoint terminal, in case e.g. need may be for using a particular ML feature.

Accordingly, the method may further comprise determining, based on the received association information, that there is no ML Model and/or Functionality available for using a particular ML feature; and providing an appropriate ML Model and/or Functionality for using the particular ML feature.

Further, according to various examples of embodiments, the method may further comprise that controlling the LCM operations (in relation to the ML Model and/or the Functionality) may comprise at least one of monitoring the ML Model and/or the Functionality, activating the ML Model and/or the Functionality, deactivating the ML Model and/or the Functionality, or switching the ML Model and/or the Functionality. The above-outlined solution allow for LCM using ML model identification and ML functionality identification. Therefore, the above-outlined solution is advantageous in that it enables for more efficient and/or more secure and/or more robust and/or failure resistant and/or flexible and/or complexity reduced LCM using ML model identification and ML functionality identification

Referring now to Figure 10, Figure 10 shows a block diagram illustrating an apparatus according to various examples of embodiments.

Specifically, Figure 10 shows a block diagram illustrating an apparatus 1000, which may represent an endpoint terminal, like e.g. such UE as outlined above with reference to Figures 5 to 7, according to various examples of embodiments, which may participate in LCM using ML model identification and ML functionality identification. Furthermore, even though reference is made to an endpoint terminal, the endpoint terminal may be also another device or function having a similar task, such as a chipset, a chip, a module, an application etc., which can also be part of a network element or attached as a separate element to a network element, or the like. It should be understood that each block and any combination thereof may be implemented by various means or their combinations, such as hardware, software, firmware, one or more processors and/or circuitry.

The apparatus 1000 shown in Figure 10 may include a processing circuitry, a processing function, a control unit or a processor 1010, such as a CPU or the like, which is suitable to enable LCM using ML model identification and ML functionality identification. The processor 1010 may include one or more processing portions or functions dedicated to specific processing as described below, or the processing may be run in a single processor or processing function. Portions for executing such specific processing may be also provided as discrete elements or within one or more further processors, processing functions or processing portions, such as in one physical processor like a CPU or in one or more physical or virtual entities, for example. Reference signs 1031 and 1032 denote input/output (I/O) units or functions (interfaces) connected to the processor or processing function 1010. The I/O units 1031 and 1032 may be a combined unit including communication equipment towards several entities/elements, or may include a distributed structure with a plurality of different interfaces for different entities/elements. Reference sign 1020 denotes a memory usable, for example, for storing data and programs to be executed by the processor or processing function 1010 and/or as a working storage of the processor or processing function 1010. It is to be noted that the memory 1020 may be implemented by using one or more memory portions of the same or different type of memory, but may also represent an external memory, e.g. an external database provided on a cloud server.

The processor or processing function 1010 is configured to execute processing related to the above described processing. In particular, the processor or processing circuitry or function 1010 includes one or more of the following sub-portions. Sub-portion 1011 is a requiring portion, which is usable as a portion for using a ML Model. The portion 1011 may be configured to perform processing according to S810 of Figure 8. Further, subportion 1012 is a receiving portion, which is usable as a portion for receiving an activation message. The portion 1012 may be configured to perform processing according to S820 of Figure 8. Moreover, sub-portion 1013 is a starting portion, which is usable as a portion for starting to use a ML Model. The portion 1013 may be configured to perform processing according to S830 of Figure 8. Moreover, sub-portion 1014 is a reporting portion, which is usable as a portion for reporting KPIs. The portion 1014 may be configured to perform processing according to S840 of Figure 8.

According to various examples of embodiments, the following may further be considered with regard to the apparatus 1000 according to Figure 10.

According to various examples of embodiments, the apparatus 1000 may comprise at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus 1000 at least to require to use a machine learning, ML, feature based on using a ML Model appropriate for the ML feature; receive an activation message in relation to at least one of an activation of one ML Model available at the apparatus 1000 and appropriate for the ML feature, the one ML Model associated with at least one Functionality, or an activation of one Functionality of at least one Functionality associated with one ML Model available at the apparatus and appropriate for the ML feature, wherein the activation message indicates that the one ML Model and/or the one Functionality is activated; based on the received activation message, start to use the one ML Model; and report at least one of

ML Model performance key performance indicators, KPIs, of the one ML

Model, or Functionality performance KPIs of the one Functionality associated with the one ML Model.

According to various examples of embodiments, the apparatus 1000 may further be caused to configure the one Functionality of the at least one Functionality provided by the one ML Model; provide a Functionality configuration resulting from the configuring, wherein the Functionality configuration is indicative of at least one of the Functionality performance KPIs to be reported, or the ML Model performance KPIs to be reported; request at least one of the activation of the one ML Model, or the activation of the one Functionality; and receive the activation message in response to the request.

According to various examples of embodiments, the apparatus 1000 may further be caused to receive a ML Model appropriate for the ML feature; determine a Functionality association for the received ML Model, the Functionality association representing an association between the received ML Model and at least one Functionality available; and if the apparatus 1000 caused to determine the Functionality association results in the apparatus 1000 accepting the determined Functionality association, provide an acknowledgement message indicating that the received ML model is available, wherein if the apparatus 1000 caused to determine the Functionality association results in the apparatus rejecting 1000 the determined Functionality association, provide a rejection message indicating that the received ML model is not available.

According to various examples of embodiments, the apparatus 1000 may further be caused to, if the apparatus 1000 accepting the determined Functionality association, configure one Functionality of the at least one Functionality accepted to be associated with the received ML Model; provide a Functionality configuration resulting from the configuring, wherein the Functionality configuration is indicative of at least one of the Functionality performance KPIs to be reported, or the ML Model performance KPIs to be reported; request at least one of the activation of the received ML Model representing the one ML Model, or the activation of the one Functionality of the at least one Functionality accepted to be associated with the received ML Model representing the one Functionality associated with the one ML Model; and receive the activation message in response to the request.

According to various examples of embodiments, the apparatus 1000 may further be caused to at least one of, in response to the report, receive a signalling to switch the one ML model; and decide whether to use another ML Model available at the apparatus and appropriate for the ML feature, to await reception of a new ML Model appropriate for the ML feature, or to fallback, or receive a signalling to switch the one Functionality; and based on the determined Functionality association, configure another Functionality.

According to various examples of embodiments, the apparatus 1000 may further be caused to select a ML Model available at the apparatus and appropriate for the ML feature; indicate the selected ML Model; register with regard to information in relation to at least the selected ML Model and at least one Functionality associated with the selected ML Model; receive an acknowledgement signal for the at least one Functionality associated with the selected ML Model; configure one Functionality of the at least one Functionality associated with the selected ML Model; wherein the apparatus 1000 being caused to configure the one Functionality comprises configuring of at least one of the Functionality performance KPIs to be reported, or the ML Model performance KPIs to be reported; and receive the activation message, wherein the selected ML Model represents to one ML Model.

According to various examples of embodiments, the apparatus 1000 may further be caused to control life cycle management, LCM, operations related to the selected ML Model.

According to various examples of embodiments, the apparatus 1000 may further be caused to, in response to the apparatus 1000 being caused to report, receive a deactivation for the Functionality associated with the selected ML Model, or receive a deactivation for the Functionality associated with the selected ML Model and receive an activation for another Functionality of the at least one Functionality associated with the selected ML Model.

According to various examples of embodiments, the apparatus 1000 may further be caused to receive a ML Model appropriate for the ML feature and at least one Functionality associated with the received ML Model; determine information of the received at least one Functionality in relation to the ML feature; provide an acknowledgement signal for both the received ML Model and the received at least one Functionality, wherein the acknowledgement signal indicates that both the received ML Model and the received at least one Functionality is available at the apparatus; configure one Functionality of the received at least one Functionality; wherein the apparatus being caused to configure comprises configuring of at least one of the Functionality performance KPIs to be reported, or the ML Model performance KPIs to be reported; and receive the activation message, wherein the received ML Model represents the one ML Model and wherein the received one Functionality represents the one Functionality.

According to various examples of embodiments, the apparatus 1000 may further be caused to, in response to the apparatus 1000 being caused to report, receive a deactivation for the received Functionality, or receive a deactivation for the received Functionality and receive an activation for another Functionality of the received at least one Functionality.

According to various examples of embodiments, the apparatus 1000 may further be caused to, in response to the apparatus 1000 being caused to report, receive an indication that an update of the received ML Model is required; receive a deactivation for the received Functionality; receive a deactivation for the received ML Model; and perform one of to use another ML Model available at the apparatus and appropriate for the ML feature, or to await reception of a new ML Model appropriate for the ML feature and of at least one Functionality associated with the new ML Model.

Referring now to Figure 11 , Figure 11 shows a block diagram illustrating an apparatus according to various examples of embodiments.

Specifically, Figure 11 shows a block diagram illustrating an apparatus, which may represent an access network element, like e.g. such gNB as outlined above with reference to Figures 5 to 7, according to various examples of embodiments, which may participate in LCM using ML model identification and ML functionality identification. Furthermore, even though reference is made to an access network element, the access network element may be also another device or function having a similar task, such as a chipset, a chip, a module, an application etc., which can also be part of a network element or attached as a separate element to a network element, or the like. It should be understood that each block and any combination thereof may be implemented by various means or their combinations, such as hardware, software, firmware, one or more processors and/or circuitry.

The apparatus 1100 shown in Figure 11 may include a processing circuitry, a processing function, a control unit or a processor 1110, such as a CPU or the like, which is suitable to enable LCM using ML model identification and ML functionality identification. The processor 1110 may include one or more processing portions or functions dedicated to specific processing as described below, or the processing may be run in a single processor or processing function. Portions for executing such specific processing may be also provided as discrete elements or within one or more further processors, processing functions or processing portions, such as in one physical processor like a CPU or in one or more physical or virtual entities, for example. Reference signs 1131 and 1132 denote input/output (I/O) units or functions (interfaces) connected to the processor or processing function 1110. The I/O units 1131 and 1132 may be a combined unit including communication equipment towards several entities/elements, or may include a distributed structure with a plurality of different interfaces for different entities/elements. Reference sign 1120 denotes a memory usable, for example, for storing data and programs to be executed by the processor or processing function 1110 and/or as a working storage of the processor or processing function 1110. It is to be noted that the memory 1120 may be implemented by using one or more memory portions of the same or different type of memory, but may also represent an external memory, e.g. an external database provided on a cloud server.

The processor or processing function 1110 is configured to execute processing related to the above described processing. In particular, the processor or processing circuitry or function 1110 includes one or more of the following sub-portions. Sub-portion 1111 is a providing portion, which is usable as a portion for providing an activation message. The portion 1111 may be configured to perform processing according to S910 of Figure 9. Further, sub-portion 1112 is an activating portion, which is usable as a portion for activating a ML Model and/or a Functionality. The portion 1112 may be configured to perform processing according to S920 of Figure 9. Moreover, sub-portion 1113 is a receiving portion, which is usable as a portion for receiving a report. The portion 1113 may be configured to perform processing according to S930 of Figure 9. Moreover, sub-portion 1114 is a performing action portion, which is usable as a portion for performing action in relation to the ML Model and/or the Functionality. The portion 1114 may be configured to perform processing according to S940 of Figure 9.

According to various examples of embodiments, the following may further be considered with regard to the apparatus 1100 according to Figure 11.

According to various examples of embodiments, the apparatus 1100 may comprise at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus 1100 at least to, in relation to a machine learning, ML, feature required to be used by another apparatus based on usage of a ML Model appropriate for the ML feature, provide an activation message in relation to at least one of an activation of one ML Model available at the another apparatus and appropriate for the ML feature, the one ML Model associated with at least one Functionality, or an activation of one Functionality of at least one Functionality associated with one ML Model available at the another apparatus and appropriate for the ML feature; activate the one ML Model and/or the one Functionality; receive a report indicative of at least ML Model performance key performance indicators, KPIs, of the one ML Model, or Functionality performance KPIs of the one Functionality associated with the one ML Model; and based on the received report, perform action in relation to the one ML Model and/or the one Functionality.

According to various examples of embodiments, the apparatus 1100 may further be caused to receive a Functionality configuration indicative of at least one of the Functionality performance KPIs included in the report to be received, or the ML Model performance KPIs included in the report to be received; receive a request about at least one of the activation of the one ML Model, or the activation of the one Functionality; and provide the activation message in response to the received request.

According to various examples of embodiments, the apparatus 1100 may further be caused to provide a ML Model appropriate for the ML feature; and if an association between the provided ML Model and at least one Functionality available is accepted, receive an acknowledgement message indicating that the provided ML model is available at the another apparatus, wherein if the association is rejected, receive a rejection message indicating that the provided ML model is not available at the another apparatus.

According to various examples of embodiments, the apparatus 1100 may further be caused to, if the association being accepted, receive a Functionality configuration resulting from a configuration of one Functionality of the at least one Functionality accepted to be associated with the provided ML Model, wherein the Functionality configuration is indicative of at least one of the Functionality performance KPIs included in the report to be received, or the ML Model performance KPIs included in the report to be received; receive a request about at least one of the activation of the provided ML Model representing the one ML Model, or the activation of the one Functionality of the at least one Functionality accepted to be associated with the provided ML Model representing the one Functionality associated with the one ML Model; and provide the activation message in response to the received request. According to various examples of embodiments, the apparatus 1100 may further be caused to at least one of control a life cycle management, LCM, operation in relation to the one ML model, or control a LCM operation in relation to the one Functionality; wherein the apparatus 1100 being caused to perform action may further comprise the apparatus 1100 being caused to at least one of based on the controlled LCM operation in relation to the one ML model, decide whether or not to switch the one Functionality; and if deciding to switch the one Functionality, provide a signalling to switch the one ML model, or based on the controlled LCM operation in relation to the one Functionality, decide whether or not to switch the one Functionality; and if deciding to switch the one Functionality, provide a signalling to switch the one Functionality.

According to various examples of embodiments, the apparatus 1100 may further be caused to receive an indication in relation to a selected ML Model available at the another apparatus and appropriate for the ML feature; receive a registration with regard to information in relation to at least the selected ML Model and at least one Functionality associated with the selected ML Model; determine the information for which the registration was received; based on the information determined, provide an acknowledgement signal for the at least one Functionality associated with the selected ML Model; and provide the activation message, wherein the selected ML Model represents to one ML Model.

According to various examples of embodiments, the apparatus 1100 being caused to perform action may further comprise the apparatus 1100 being caused to at least one of provide a deactivation for the Functionality associated with the selected ML Model, or provide a deactivation for the Functionality associated with the selected ML Model and provide an activation for another Functionality of the at least one Functionality associated with the selected ML Model.

According to various examples of embodiments, the apparatus 1100 may further be caused to provide a ML Model appropriate for the ML feature and at least one Functionality associated with the provided ML Model; receive an acknowledgement signal for both the provided ML Model and the provided at least one Functionality, wherein the acknowledgement signal indicates that both the provided ML Model and the provided at least one Functionality is available at the another apparatus; and provide the activation message, wherein the provided ML Model represents the one ML Model and wherein the provided one Functionality represents the one Functionality. According to various examples of embodiments, the apparatus 1100 may further be caused to control LCM operations at least related to the provided ML Model and the provided one Functionality; and wherein the apparatus 1100 being caused to perform action may further comprise the apparatus 1100 being caused to at least one of, based on the controlled LCM operations, provide a deactivation for the provided Functionality, or provide a deactivation for the provided Functionality and provide an activation for another Functionality of the provided at least one Functionality.

According to various examples of embodiments, the apparatus 1100 may further be caused to control LCM operations at least related to the provided ML Model and the provided one Functionality; and wherein the apparatus 1100 being caused to perform action may further comprise the apparatus 1100 being caused to at least one of, based on the controlled LCM operations, trigger to switch the one ML Model by providing an indication that an update of the provided ML Model is required; provide a deactivation for the provided Functionality; and provide a deactivation for the provided ML Model.

It shall be noted that the apparatuses 1000 and 1100 as outlined above with reference to Figures 10 and 11 may comprise further/additional sub-portions, which may allow the apparatuses 1000 and 1100 to perform such methods/method steps as outlined above with reference to Figures 5 to 7 and/ or Figures 8 and 9.

Further, according to various examples of embodiments, there may be provided a computer program product for a computer, including software code portions for performing the steps of any of appended claims 1 to 9, or any of appended claims 10 to 15, when said product is run on the computer, wherein, optionally, the computer program product includes a computer-readable medium on which said software code portions are stored, and/or the computer program product is directly loadable into the internal memory of the computer and/or transmittable via a network by means of at least one of upload, download and push procedures.

It should be appreciated that

- an access technology via which traffic is transferred to and from an entity in the communication network may be any suitable present or future technology, such as WLAN (Wireless Local Access Network), WiMAX (Worldwide Interoperability for Microwave Access), LTE, LTE-A, 5G, Bluetooth, Infrared, and the like may be used; additionally, embodiments may also apply wired technologies, e.g. IP based access technologies like cable networks or fixed lines.

- embodiments suitable to be implemented as software code or portions of it and being run using a processor or processing function are software code independent and can be specified using any known or future developed programming language, such as a high-level programming language, such as objective-C, C, C++, C#, Java, Python, Javascript, other scripting languages etc., or a low-level programming language, such as a machine language, or an assembler.

- implementation of embodiments is hardware independent and may be implemented using any known or future developed hardware technology or any hybrids of these, such as a microprocessor or CPU (Central Processing Unit), MOS (Metal Oxide Semiconductor), CMOS (Complementary MOS), BiMOS (Bipolar MOS), BiCMOS (Bipolar CMOS), ECL (Emitter Coupled Logic), and/or TTL (Transistor-Transistor Logic).

- embodiments may be implemented as individual devices, apparatuses, units, means or functions, or in a distributed fashion, for example, one or more processors or processing functions may be used or shared in the processing, or one or more processing sections or processing portions may be used and shared in the processing, wherein one physical processor or more than one physical processor may be used for implementing one or more processing portions dedicated to specific processing as described,

- an apparatus may be implemented by a semiconductor chip, a chipset, or a (hardware) module including such chip or chipset;

- embodiments may also be implemented as any combination of hardware and software, such as ASIC (Application Specific IC (Integrated Circuit)) components, FPGA (Field- programmable Gate Arrays) or CPLD (Complex Programmable Logic Device) components or DSP (Digital Signal Processor) components.

- embodiments may also be implemented as computer program products, including a computer usable medium having a computer readable program code embodied therein, the computer readable program code adapted to execute a process as described in embodiments, wherein the computer usable medium may be a non-transitory medium.

Although the present disclosure has been described herein before with reference to particular embodiments thereof, the present disclosure is not limited thereto and various modifications can be made thereto.

Claims

CLAIMS:

1. A method comprising, requiring (S810) to use a machine learning, ML, feature based on using a ML Model appropriate for the ML feature; receiving (S820) an activation message in relation to at least one of an activation of one ML Model available and appropriate for the ML feature, the one ML Model associated with at least one Functionality, or an activation of one Functionality of at least one Functionality associated with one ML Model available and appropriate for the ML feature, wherein the activation message indicates that the one ML Model and/or the one Functionality is activated; based on the received activation message, starting (S830) to use the one ML Model; and reporting (S840) at least one of

ML Model performance key performance indicators, KPIs, of the one ML Model, or

Functionality performance KPIs of the one Functionality associated with the one ML Model.

2. The method according to claim 1 , further comprising, configuring the one Functionality of the at least one Functionality provided by the one ML Model ; providing a Functionality configuration resulting from the configuring, wherein the Functionality configuration is indicative of at least one of the Functionality performance KPIs to be reported, or the ML Model performance KPIs to be reported; requesting at least one of the activation of the one ML Model , or the activation of the one Functionality ; and receiving the activation message in response to the requesting.

3. The method according to claim 1 , further comprising receiving a ML Model appropriate for the ML feature; determining a Functionality association for the received ML Model, the Functionality association representing an association between the received ML Model and at least one Functionality available; and if accepting the determined Functionality association, providing an acknowledgement message indicating that the received ML model is available, wherein if rejecting the determined Functionality association, providing a rejection message indicating that the received ML model is not available.

4. The method according to claim 3, further comprising, if the determined Functionality association being accepted, configuring one Functionality of the at least one Functionality accepted to be associated with the received ML Model; providing a Functionality configuration resulting from the configuring, wherein the Functionality configuration is indicative of at least one of the Functionality performance KPIs to be reported, or the ML Model performance KPIs to be reported; requesting at least one of the activation of the received ML Model representing the one ML Model, or the activation of the one Functionality of the at least one Functionality accepted to be associated with the received ML Model representing the one Functionality associated with the one ML Model; and receiving the activation message in response to the requesting.

5. The method according to any one of claims 1 to 4, further comprising at least one of, in response to the reporting, receiving a signalling to switch the one ML model; and deciding whether to use another ML Model available and appropriate for the ML feature, to await reception of a new ML Model appropriate for the ML feature, or to fallback, or receiving a signalling to switch the one Functionality; and based on the determined Functionality association, configuring another Functionality, or receiving a deactivation for the Functionality associated with the selected ML Model, or receiving a deactivation for the Functionality associated with the selected ML Model and receiving an activation for another Functionality of the at least one Functionality associated with the selected ML Model.

6. The method according to claim 1 , further comprising receiving a ML Model appropriate for the ML feature and at least one Functionality associated with the received ML Model; determining information of the received at least one Functionality in relation to the ML feature; providing an acknowledgement signal for both the received ML Model and the received at least one Functionality, wherein the acknowledgement signal indicates that both the received ML Model and the received at least one Functionality is available; configuring one Functionality of the received at least one Functionality; wherein the configuring comprises configuration of at least one of the Functionality performance KPIs to be reported, or the ML Model performance KPIs to be reported; and receiving the activation message, wherein the received ML Model represents the one ML Model and wherein the received one Functionality represents the one Functionality.

7. The method according to claim 1 or 6, further comprising, in response to the reporting, receiving a deactivation for the received Functionality, or in response to the reporting, receiving a deactivation for the received Functionality and receiving an activation for another Functionality of the received at least one Functionality.

8. The method according to any one of claims 1 , 6 or 7, further comprising, in response to the reporting, receiving an indication that an update of the received ML Model is required; receiving a deactivation for the received Functionality; receiving a deactivation for the received ML Model; and performing one of using another ML Model available and appropriate for the ML feature, or awaiting reception of a new ML Model appropriate for the ML feature and of at least one Functionality associated with the new ML Model.

9. A method comprising, in relation to a machine learning, ML, feature required to be used based on usage of a ML Model appropriate for the ML feature, providing (S910) an activation message in relation to at least one of an activation of one ML Model available and appropriate for the ML feature, the one ML Model associated with at least one Functionality, or an activation of one Functionality of at least one Functionality associated with one ML Model available and appropriate for the ML feature; activating (S920) the one ML Model and/or the one Functionality ; receiving (S930) a report indicative of at least

ML Model performance key performance indicators, KPIs, of the one ML Model, or

Functionality performance KPIs of the one Functionality associated with the one ML Model; and based on the received report, performing action (S940) in relation to the one ML Model and/or the one Functionality .

10. The method according to claim 9, further comprising receiving a Functionality configuration indicative of at least one of the Functionality performance KPIs included in the report to be received, or the ML Model performance KPIs included in the report to be received; receiving a request about at least one of the activation of the one ML Model, or the activation of the one Functionality ; and providing the activation message in response to the received request.

11. The method according to claim 9, further comprising providing a ML Model appropriate for the ML feature; and if an association between the provided ML Model and at least one Functionality available is accepted, receiving an acknowledgement message indicating that the provided ML model is available, wherein if the association is rejected, receiving a rejection message indicating that the provided ML model is not available.

12. The method according to claim 11 , further comprising if the association being accepted, receiving a Functionality configuration resulting from a configuration of one Functionality of the at least one Functionality accepted to be associated with the provided ML Model , wherein the Functionality configuration is indicative of at least one of the Functionality performance KPIs included in the report to be received, or the ML Model performance KPIs included in the report to be received; receiving a request about at least one of the activation of the provided ML Model representing the one ML Model, or the activation of the one Functionality of the at least one Functionality accepted to be associated with the provided ML Model representing the one Functionality associated with the one ML Model; and providing the activation message in response to the received request, wherein the method further comprising at least one of controlling a life cycle management, LCM, operation in relation to the one ML model, or controlling a LCM operation in relation to the one Functionality; wherein the performing action comprises at least one of, based on the controlled LCM operation in relation to the one ML model, deciding whether or not to switch the one Functionality; and if deciding to switch the one Functionality, providing a signalling to switch the one ML model, or based on the controlled LCM operation in relation to the one Functionality, deciding whether or not to switch the one Functionality; and if deciding to switch the one Functionality, providing a signalling to switch the one Functionality.

13. The method according to claim 9, further comprising providing a ML Model appropriate for the ML feature and at least one Functionality associated with the provided ML Model; receiving an acknowledgement signal for both the provided ML Model and the provided at least one Functionality, wherein the acknowledgement signal indicates that both the provided ML Model and the provided at least one Functionality is available; and providing the activation message, wherein the provided ML Model represents the one ML Model and wherein the provided one Functionality represents the one Functionality.

14. The method according to claim 9 or 13, further comprising controlling LCM operations at least related to the provided ML Model and the provided one Functionality; and wherein the performing action comprises at least one of, based on the controlled LCM operations, providing a deactivation for the provided Functionality, or providing a deactivation for the provided Functionality and providing an activation for another Functionality of the provided at least one Functionality.

15. The method according to any one of claims 9, 13 or 14, further comprising controlling LCM operations at least related to the provided ML Model and the provided one Functionality; and wherein the performing action comprises at least one of, based on the controlled LCM operations, triggering to switch the one ML Model by providing an indication that an update of the provided ML Model is required; providing a deactivation for the provided Functionality; and providing a deactivation for the provided ML Model.

16. An apparatus (1000; 1100), comprising at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus (1000) at least to perform the method according to any one of claims 1 to 8, or cause the apparatus (1100) at least to perform the method according to any one of claims 9 to 15.