A user equipment and a network node

Abstract

An apparatus comprising means for: transmitting, to a network node, a multi-level identifier capability, wherein the multi-level identifier capability indicates a capability of the UE to use a model i

Description

Model identification enables a common understanding between a network (NW) and a user equipment (UE) of what machine learning models are available at the UE, and what model can be used, configured, and / or activated at the UE by the NW. BRIEF SUMMARY According to various, but not necessarily all, examples there is provided an apparatus comprising means for: transmitting, to a network node, a multi-level identifier capability, wherein the multi-level identifier capability indicates a capability of a user equipment, UE, to use a model identifier to identify one or more machine learning supported functionalities of the UE associated with one or more model identifier levels, wherein model identifier levels are associated with corresponding different validity domains for machine learning supported functionalities of the UE; receiving, from the network node, a model identification configuration comprising at least a first model identifier identifying one or more first machine learning supported functionalities of the UE and a second model identifier identifying one or more second machine learning supported functionalities of the UE, wherein the one or more first machine learning supported functionalities of the UE are associated with one or more first model identifier levels and the one or more second machine learning supported functionalities of the UE are associated with one or more second model identifier levels; and using one of the one or more first machine learning supported functionalities of the UE based on the received model identification configuration. In some but not necessarily all examples, the apparatus comprises means for: receiving, from the network node, a switching configuration configuring the UE with one or more first reconfiguration conditions for switching from using the first identifier identifying the one or more first machine learning supported functionalities of the UE to using the second identifier identifying the one or more second machine learning supported functionalities of the UE, wherein the second model identifier has a broader validity domain than the first model identifier; and based on at least one of the one or more reconfiguration conditions being met, using one of the one or more second machine learning supported functionalities of the UE based on the second model identifier. In some but not necessarily all examples, the apparatus comprises means for determining that one or more of the one or more first reconfiguration conditions has been met. In some but not necessarily all examples, at least one of the reconfiguration conditions for switching from using the first identifier identifying the one or more first machine learning supported functionalities of the UE to using the second identifier identifying the one or more second machine learning functionalities of the UE is an invalidity event invalidating use of the first identifier of the one or more first machine learning supported functionalities of the UE. In some but not necessarily all examples, the one or more first machine learning supported functionalities of the UE and the one or more second machine learning supported functionalities of the UE perform the same machine learning supported functionality In some but not necessarily all examples, a validity domain is related to an operating context of the UE. In some but not necessarily all examples, the invalidity event is a change in operating context of the UE In some but not necessarily all examples, the validity domain is related to spatial and / or temporal characteristics of the UE and / or the network. In some but not necessarily all examples, a spatial validity domain corresponds to one of: a beam; a cell; a network node; a RAN notification area (RNA); a Registration Area (RA); or a Public Land Mobile Network (PLMN) serving the UE. In some but not necessarily all examples, in the invalidity event is a mobility event of the UE. In some but not necessarily all examples, a temporal validity domain corresponds to one of: a time of day; a day; a week; or a month. In some but not necessarily all examples, the invalidity event is a temporal event related to a temporal change of the corresponding KPIs In some but not necessarily all examples, the validity domain is related to a compatibility of its associated one or more model identifiers with at least one of: the network; a UE vendor; a mobile network operator. In some but not necessarily all examples, the invalidity event causes the UE to be served by a new cell under the same network node or a new cell under a new network node. In some but not necessarily all examples, the model identification configuration indicates a hierarchy of identifier levels associated with corresponding different validity domains within a hierarchy of validity domains In some but not necessarily all examples, the apparatus comprises means for managing implementation of the first ML model and managing validity of the first ML model using associated validity domain. In some but not necessarily all examples, identifiers having the same validity domain have the same format In some but not necessarily all examples, identifiers having a narrower validity domain have a shorter format than identifiers having a broader validity domain. In some but not necessarily all examples, the one or more machine learning supported functionalities of the UE are associated with one or more validity domains, wherein the validity domains of the one or more machine learning supported functionalities of the UE are related to the validity domains of the one or more model identifier levels. In some but not necessarily all examples, a model identifier comprises at least: an indication of a model identifier level with which one or more machine learning supported functionalities of the UE identified by the model identifier is associated; and an indication of an association of the model identifier with one or more operating contexts of the UE. In some but not necessarily all examples, the model identification configuration comprises a third model identifier identifying one or more third machine learning supported functionalities of the UE, and the switching configuration configures the UE with one or more second reconfiguration conditions for switching from using the first identifier identifying the one or more first machine learning supported functionalities of the UE or the second identifier identifying the one or more second machine learning functionalities of the UE to using the third identifier identifying the one or more third machine learning functionalities of the UE, wherein the third model identifier has a broader validity domain than the first model identifier and the second model identifier, and wherein the UE comprises means for: based on one or more of the one or more second reconfiguration conditions being met, using one of the one or more third machine learning supported functionalities of the UE based on the third model identifier. In some but not necessarily all examples, the apparatus comprises means for receiving, from the network node, an indication to use a different model identifier of one or more different machine learning supported functionalities of the UE, wherein the different identifier has a narrower validity domain than an identifier of a current machine learning supported functionality of the UE; and using one of the one or more different machine learning supported functionalities of the UE based on the different model identifier. In some but not necessarily all examples, the apparatus comprises means for, in dependence upon receiving the indication to use the different model identifier of the one or more different machine learning supported functionalities of the UE, determining if the UE has a model identifier with a narrower validity domain than the model identifier of a current machine learning supported functionality of the UE; and in dependence upon a determination that the UE does not have a model ID with a narrower validity domain than the identifier of the current machine learning supported functionality of the UE, continuing to use the current machine learning supported functionality of the UE. According to various, but not necessarily all, examples there is provided an apparatus comprising means for: receiving, from a user equipment, UE, a multi-level identifier capability, wherein the multi-level identifier capability indicates a capability of the UE to use a model identifier to identify one or more machine learning supported functionalities of the UE associated with one or more model identifier levels, wherein model identifier levels are associated with corresponding different validity domains for machine learning supported functionalities of the UE; determining, based on the received multi-level identifier capability, a model identification configuration comprising at least a first model identifier identifying one or more first machine learning supported functionalities of the UE and a second model identifier identifying one or more second machine learning supported functionalities of the UE, wherein the one or more first machine learning supported functionalities of the UE are associated with one or more first model identifier levels and the one or more second machine learning supported functionalities of the UE are associated with one or more second model identifier levels; and transmitting the model identification configuration to the UE. In some but not necessarily all examples, the apparatus comprises means for transmitting, to the UE, a switching configuration configuring the UE with one or more first reconfiguration conditions for switching from using the first identifier identifying the one or more first machine learning supported functionalities of the UE to using the second identifier identifying the one or more second machine learning supported functionality of the UE, wherein the second model identifier has a broader validity domain than the first model identifier. In some but not necessarily all examples, the apparatus comprises means for configuring a plurality of UEs. According to various, but not necessarily all, examples there is provided a user equipment, UE, comprising means for: transmitting, to a network node, a multi-level identifier capability, wherein the multilevel identifier capability indicates a capability of the UE to use a model identifier to identify one or more machine learning supported functionalities of the UE associated with one or more model identifier levels, wherein model identifier levels are associated with corresponding different validity domains for machine learning supported functionalities of the UE; receiving, from the network node, a model identification configuration comprising at least a first model identifier identifying one or more first machine learning supported functionalities of the UE and a second model identifier identifying one or more second machine learning supported functionalities of the UE, wherein the one or more first machine learning supported functionalities of the UE are associated with one or more first model identifier levels and the one or more second machine learning supported functionalities of the UE are associated with one or more second model identifier levels; and using one of the one or more first machine learning supported functionalities of the UE based on the received model identification configuration. According to various, but not necessarily all, examples there is provided a network node comprising means for: receiving, from a user equipment, UE, a multi-level identifier capability, wherein the multi-level identifier capability indicates a capability of the UE to use a model identifier to identify one or more machine learning supported functionalities of the UE associated with one or more model identifier levels, wherein model identifier levels are associated with corresponding different validity domains for machine learning supported functionalities of the UE; determining, based on the received multi-level identifier capability, a model identification configuration comprising at least a first model identifier identifying one or more first machine learning supported functionalities of the UE and a second model identifier identifying one or more second machine learning supported functionalities of the UE, wherein the one or more first machine learning supported functionalities of the UE are associated with one or more first model identifier levels 7 and the one or more second machine learning supported functionalities of the UE are associated with one or more second model identifier levels; and transmitting the model identification configuration to the UE. According to various, but not necessarily all, examples there is provided a method. The method comprises transmitting, to a network node, a multi-level identifier capability. The multi-level identifier capability indicates a capability of a user equipment, UE, to use a model identifier to identify one or more machine learning supported functionalities of the UE associated with one or more model identifier levels. Model identifier levels are associated with corresponding different validity domains for machine learning supported functionalities of the UE. The method further comprises receiving, from the network node, a model identification configuration. The model identification configuration comprises at least a first model identifier identifying one or more first machine learning supported functionalities of the UE and a second model identifier identifying one or more second machine learning supported functionalities of the UE. The one or more first machine learning supported functionalities of the UE are associated with one or more first model identifier levels and the one or more second machine learning supported functionalities of the UE are associated with one or more second model identifier levels. The method further comprises using one of the one or more first machine learning supported functionalities of the UE based on the received model identification configuration. According to various, but not necessarily all, examples there is provided a method. The method comprises receiving, from a user equipment, UE, a multi-level identifier capability. The multi-level identifier capability indicates a capability of the UE to use a model identifier to identify one or more machine learning supported functionalities of the UE associated with one or more model identifier levels. Model identifier levels are associated with corresponding different validity domains for machine learning supported functionalities of the UE. The method further comprises determining, based on the received multi-level identifier capability, a model identification configuration comprising at least a first model identifier identifying one or more first machine learning supported functionalities of the UE and a second model identifier identifying one or more second machine learning supported functionalities of the UE. The one or more first machine learning supported functionalities of the UE are 8 associated with one or more first model identifier levels and the one or more second machine learning supported functionalities of the UE are associated with one or more second model identifier levels. The method further comprises transmitting the model identification configuration to the UE. According to various, but not necessarily all, embodiments there is provided an apparatus comprising at least one processor; and at least one memory including computer program code; the at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus to perform at least a part of one or more methods described herein. According to various, but not necessarily all, embodiments there is provided an apparatus comprising means for performing at least part of one or more methods described herein. The description of a function and / or action should additionally be considered to also disclose any means suitable for performing that function and / or action. Functions and / or actions described herein can be performed in any suitable way using any suitable method. According to various, but not necessarily all, embodiments there is provided examples as claimed in the appended claims. While the above examples of the disclosure and optional features are described separately, it is to be understood that their provision in all possible combinations and permutations is contained within the disclosure. It is to be understood that various examples of the disclosure can comprise any or all the features described in respect of other examples of the disclosure, and vice versa. Also, it is to be appreciated that any one or more or all the features, in any combination, may be implemented by / comprised in / performable by an apparatus, a method, and / or computer program instructions as desired, and as appropriate. The description of a function should additionally be considered to also disclose any means suitable for performing that function BRIEF DESCRIPTION Some examples will now be described with reference to the accompanying drawings in which: FIG. 1 shows an example of the subject matter described herein; FIG. 2 shows another example of the subject matter described herein; FIG. 3 shows another example of the subject matter described herein; FIG. 4 shows another example of the subject matter described herein; FIG. 5 shows another example of the subject matter described herein; FIG. 6 shows another example of the subject matter described herein; FIG. 7 shows another example of the subject matter described herein; FIG. 8 shows another example of the subject matter described herein; FIG. 9 shows another example of the subject matter described herein; FIG. 10 shows another example of the subject matter described herein; FIG. 11 shows another example of the subject matter described herein; FIG. 12 shows another example of the subject matter described herein; FIGs 13A and 13B show another example of the subject matter described herein; FIGs 14A and 14B show another example of the subject matter described herein; FIGs 15A and 15B show another example of the subject matter described herein; FIGs 16A and 16B show another example of the subject matter described herein; FIG. 17 shows another example of the subject matter described herein; FIG. 18 shows another example of the subject matter described herein; FIG. 19 shows another example of the subject matter described herein; and FIGs 20A and 20B show another example of the subject matter described herein. The figures are not necessarily to scale. Certain features and views of the figures can be shown schematically or exaggerated in scale in the interest of clarity and conciseness. For example, the dimensions of some elements in the figures can be exaggerated relative to other elements to aid explication. Similar reference numerals are used in the figures to designate similar features. For clarity, all reference numerals are not necessarily displayed in all figures. DETAILED DESCRIPTION The Figures illustrate a user equipment, UE 110, comprising means for: transmitting, to a network node 120, a multi-level identifier capability, wherein the multi-level identifier capability indicates a capability of the UE 110 to use a model identifier 160 to identify one or more machine learning supported functionalities of the UE 110 associated with one or more model identifier levels 150, wherein model identifier levels 150 are associated with corresponding different validity domains for machine learning supported functionalities of the UE 110; receiving, from the network node 120, a model identification configuration comprising at least a first model identifier 160 identifying one or more first machine learning supported functionalities of the UE 110 and a second model identifier 160 identifying one or more second machine learning supported functionalities of the UE 110, wherein the one or more first machine learning supported functionalities of the UE 110 are associated with one or more first model identifier levels 150 and the one or more second machine learning supported functionalities of the UE 110 are associated with one or more second model identifier levels 150; and using one of the one or more first machine learning supported functionalities of the UE 110 based on the received model identification configuration. This provides the technical effect of providing full control over the usage of UE 110 model identifiers for machine learning supported functionalities of the UE 110. UE 110 model identifiers may be used in more flexible way, taking into account various factors and conditions in the network, which leads to more dynamic / responsive and adaptive use of machine learning supported functionalities. The disclosed way of using multi-level identifier capability further allows to reduce communication latency and signaling / data transfer overhead. The apparatus may be for providing full control over the usage of UE 110 model identifiers for machine learning supported functionalities of the UE 110. Fig. 1 illustrates an example of a network 100 comprising a plurality of network entities including terminal apparatus 110, node apparatus 120 and one or more network apparatus 130. The terminal apparatus 110 and node apparatus 120 communicate 124 with each other. The one or more network apparatus 130 communicate 128 with the node apparatus 120. In some examples the one or more network apparatus 130 communicate with the terminal apparatus 110. The one or more network apparatus 130 can, in some examples, communicate with each other. The one or more node apparatus 120 can, in some examples, communicate 126 with each other. The network 100 can be a cellular network comprising a plurality of cells 122 each served by a node apparatus 120. In this example, the interface between the terminal apparatus 110 and a node apparatus 120 defining a cell 122 is a wireless interface 124. The node apparatus 120 comprises one or more cellular radio transceivers. The terminal apparatus 110 comprises one or more cellular radio transceivers. In the example illustrated the cellular network 100 is a third generation Partnership Project (3GPP) network in which the terminal apparatus 110 are user equipment (UE) and the node apparatus 120 can be access nodes such as base stations. A user equipment comprises a mobile equipment. Where reference is made to user equipment that reference includes and encompasses, wherever possible, a reference to mobile equipment. In some examples, during operation, a user equipment 110 comprises a mobile equipment comprising a smart card for authentication / encryption etc. such as a Subscriber Identity Module (SIM). In some examples, during operation, a user equipment 110 comprises mobile equipment comprising circuitry embedded as part of the user equipment 110 for authentication / encryption such as software SIM. The node apparatus 120 can be any suitable access node such as a base station or transmission reception point. The node apparatus 120 can be a network element responsible for radio transmission and reception in one or more cells 122, to or from the UE 110. The node apparatus 120 can be a network element in a Radio Access Network (RAN), an Open-Radio Access Network (O-RAN) or any other suitable type of network. The network apparatus 130 can be part of a core network. The network apparatus 130 can be configured to manage functions relating to connectivity for the UEs 110. For example, the network apparatus 130 can be configured to manage functions such as connectivity, mobility, authentication, authorization, and / or other suitable functions. In some examples the network apparatus 130 can comprise an Access and Mobility management Function (AMF) and / or a User Plane Function (UPF) or any other suitable entities. In the example of Fig. 1 the network apparatus 130 is shown as a single entity. In some examples the network apparatus 130 could be distributed across a plurality of entities. For example, the network apparatus 130 could be cloud based or distributed in any other suitable manner. The network apparatus 130 can be a core network node 120. The network 100 can be a 4G or 5G network, for example. It can for example be a New Radio (NR) network that uses gNB or eNB as access nodes 120. In such cases the node apparatus 120 can comprise gNodeBs (gNBs) 120 configured to provide user plane and control plane protocol terminations towards the UE 110 and / or to perform any other suitable functions. The gNBs 120 are interconnected with each other by means of an X2 / Xn interface 126. The gNBs are also connected by means of the N2 interface 128 to the network apparatus 130. The gNBs can be connected to an AMF or any other suitable network apparatus 130. Other types of networks and interfaces could be used in other examples. Other types of network could comprise next generation mobile and communication network, for example, a 6G network. The air interface may be augmented with features enabling improved support of artificial intelligence (Al) and machine learning (ML) models for enhanced performance and / or reduced complexity and overheads. AI / ML functionality identification is a method of identifying a UE-side AI / ML functionality to obtain a common understanding between a network (NW) and a user equipment (UE). Information regarding the UE-side AI / ML functionality may be shared during a UE capabilities exchange procedure. Where AI / ML model resides depends on specific use cases and sub-use cases. AI / ML model identification is a method of identifying a UE-side AI / ML model to obtain a common understanding between the NW and the UE 110. Model identification may be, but does not have to be, carried out. Information regarding the UE-side AI / ML model may be shared during model identification procedure. A UE-side AI / ML model may be selected for activation among multiple models for the same AI / ML enabled feature. Model selection may be carried out simultaneously with, or separately from, model activation to enable the AI / ML model for a specific function. Model switching comprises deactivating (disabling) a currently active AI / ML model and activating (enabling) a different AI / ML model for a specific function. Methods of UE-side AI / ML model identification are categorized as type A or type B. In type A model identification, the model is identified to the network (if applicable) and the UE 110 (if applicable) without over-the-air signaling. The model may be assigned with a model identifier 160 during the model identification, which may be referred or used in over-the-air signaling after model identification. In type B model identification, the model is identified via over-the-air signaling. The model may be assigned with a model identifier 160 during the model identification. Type B model identification may be of type B1 or B2. In type B1 model identification, model identification is initiated by the UE 110, and the NW assists the remaining steps (if any) of the model identification. In type B2 model identification, model identification is initiated by the NW, and the UE 110 responds (if applicable) for the remaining steps (if any) of the model identification. In Type B model identification, a model may be identified using at least the below three options: Mi-Option 1: Model identification with data collection related configuration(s) and / or indication(s); Mi-Option 2: Model identification with dataset transfer; Mi-Option 3: Model identification in model transfer from NW to UE 110. Model identification enables a common understanding between the NW and the UE 110 of what models are available at the UE 110, and what model can be used, configured, and / or activated at the UE 110 by the NW. Model identification does not require detailed implementation or algorithm information about the model being identified. In examples, a model may be identified using data defining said model. For example, a model identifier 160 assigned to first data is implicitly assigned to the model defined by the first data. Model identifiers may be considered to be locally unique identifiers or globally unique identifiers. For example, locally unique identifiers may be valid for the RRC connection of a specific UE 110 in a specific cell. Locally unique identifiers may have shorter formats than globally unique identifiers. Using locally unique identifiers may require frequent model identification processes between the UE 110 and the NW, causing potential delays whenever the local identifier becomes invalid. Alternatively, globally unique identifiers may be valid for any UE 110 in any cell. Globally unique identifiers may have longer formats than locally unique identifiers. Using globally unique identifiers for an ML model deployed at the UE 110 can lead to a large overhead due to the large format of the globally unique identifier, especially when the identifier needs to be frequently exchanged between the network and the UE. The below examples describe a signaling mechanism using a hierarchical definition and use of AI / ML model identifiers and relevant signaling to enable the use of hierarchical AI / ML model identifiers. Under the signaling mechanism described, UEs and the NW conduct a hierarchical AI / ML model identification; use a locally unique shorter identifier associated with an AI / ML model as long as the local identifier is valid; and once the local identifier is not valid anymore, UEs (and the NW) are triggered to use an identifier with a higher level of hierarchy and a corresponding longer identifier to identify the / an AI / ML model without requiring a new mobile identification procedure. FIG. 2 illustrates an example of a method. In examples, FIG. 2 can be considered to illustrate a plurality of methods. For example, FIG. 2 illustrates one or more actions at a plurality of actors / entities, and, in examples, FIG. 2 can be considered to illustrate a plurality of methods performed by the individual actors / entities. One or more of the features discussed in relation to FIG. 2 can be found in one or more of the other FIGs. In the example of FIG. 2, a plurality of apparatuses transmit and / or receive one or more signals and / or one or more messages across and / or via and / or using a network. In examples, any suitable form of communication in any suitable network can be used. For example, at least a portion of the network 100 of FIG. 1 can be used. Accordingly, in examples, the plurality of apparatuses in FIG. 2 form at least a portion of network 100 as described in relation to FIG. 1. In the illustrated example, a terminal node 110 and an access node 120 transmit and / or receive one or more signals and / or one or more messages. The access node can comprise a gNodeB (gNB) and the terminal node 110 can comprise a UE. In examples, communications and / or transmissions between elements illustrated in FIG. 2 can proceed via any number of intervening elements, including no intervening elements. Although one terminal node 110 is illustrated in the example of FIG. 2, in examples any suitable number of terminal nodes 110, for example UEs, can be included. Similarly, in examples, any suitable number of access nodes 120 can be included. As described herein, a description of a function and / or action should also be considered to disclose enabling, and / or causing, and / or controlling that function and / or action. For example, a description of transmitting information should also be considered to disclose enabling, and / or causing, and / or controlling transmitting / transmission of information. For example, a description of an apparatus, such as a UE, transmitting information should also be considered to disclose at least one controller of the apparatus enabling, and / or causing, and / or controlling the apparatus to transmit the information. In the illustrated example, the location of blocks indicates the entity performing the functions(s) and / or action(s). Because FIG. 2 illustrates one or more actions / features of transmitting, FIG. 2 illustrates the corresponding receiving / enabling and / or causing receiving action(s) / feature(s). For the further discussion of FIG. 2 it will be considered that the terminal node 110 is a UE 140. From the point of view of the UE, at block 202, the method comprises receiving, from a network node 120, a request for a multi-level identifier capability of the UE. At block 204, the method comprises transmitting, to a network node 120, a multi-level identifier capability. In some, but not necessarily all, examples, transmission of the multi-level identifier capability is carried out in dependence upon receiving the request for a multi-level identifier capability from the NW. Blocks 202 and 204 may thus be referred to as a capability exchange. The multi-level identifier capability indicates a capability of the UE 110 to use a model identifier 160 to identify one or more machine learning supported functionalities of the UE 110 associated with one or more model identifier levels 150. In some, but not necessarily all, examples, the multi-level identifier capability indicates multiple model identifier levels 150 the UE 110 is capable of using. In some, but not necessarily all examples, the multi-level identifier capability indicates a number N of model identifier levels 150 the UE 110 is capable of configuring. In some, but not necessarily all, examples, a model identifier 160 is associated with a UE-side AI / ML model. A model identifier 160 thus identifies one or more machine learning functionalities of the UE. The model identifier 160 has a model identifier level, and the one or more machine learning functionalities of the UE 110 identified by the model identifier 160 are associated with the model identifier level 150 of the model identifier 160. In some, but not necessarily all, examples, a model identifier 160 comprises at least: an indication of a model identifier level 150 with which one or more machine learning supported functionalities of the UE 110 identified by the model identifier 160 is associated; and an indication of an association of the model identifier 160 with one or more operating contexts of the UE. In some, but not necessarily all, examples, different model identifiers are associated with and / or point to the same physical or logical AI / ML model. In other examples, different model identifiers are associated with and / or point to different physical or logical AI / ML models. The mapping between model identifiers and physical or logical AI / ML models is dependent, for the UE 110 and the NW, on specific use cases and is not discussed in this application. Thus, in some, but not necessarily all, examples, the one or more first machine learning supported functionalities of the UE 110 and the one or more second machine learning supported functionalities of the UE 110 perform the same UE functionality for the same UE 110 feature. In some, but not necessarily all, examples, the one or more first machine learning supported functionalities of the UE 110 and the one or more second machine learning supported functionalities of the UE 110 perform different machine learning supported functionalities. In some such examples, the UE 110 may switch between the different machine learning supported functionalities if there are no available AI / ML models suitable for the current machine learning functionality after a change in operating context. Model identifier levels 150 are associated with corresponding different validity domains for machine learning supported functionalities of the UE. Thus, a model identifier level 150 indicates a validity domain of a machine learning supported functionality identified by a model identifier 160 of the model identifier level 150. In some, but not necessarily all, examples, model identifier levels 150 are defined and configured by the network. In some, but not necessarily all, examples, a validity domain is related to at least one of: an operating context of the UE, for example an operating radio (configuration) context of the UE; spatial characteristics of the UE 110 (spatial validity domain), for example spatial radio network characteristics of the UE; temporal characteristics of the UE 110 and / or the NW (temporal validity domain), for example temporal radio characteristics of the UE; or a compatibility between one or more model identifiers associated with the validity domain and at least one of: the network; a UE vendor; a mobile network operator. In some, but not necessarily all, examples, a validity domain is related to a combination of two or more of the above. An operating context of the UE 110 may comprise one or more of: spatial characteristics of the UE; temporal characteristics of the UE; or a compatibility between one or more model identifiers associated with the validity domain and at least one of: the network; a UE vendor; a mobile network operator. The operating context of the UE 110 may be within a validity domain. For example, if a validity domain relates to an area served by a single gNB, and a position of the UE 110 is within the area served by the single gNB, then the operating context of the UE 110 is within the validity domain. In another example, if a validity domain relates to any times between 00:00 and 06:00 and the UE 110 is operating at 05:00, then the operating context of the UE 110 is within the validity domain. The operating context of the UE 110 may be outside a validity domain. For example, if a validity domain relates to an area served by a single gNB, and a position of the UE 110 is outside the area served by the single gNB, then the operating context of the UE 110 is outside the validity domain. In another example, if a validity domain relates to any times between 00:00 and 06:00 and the UE 110 is operating at 08:00, then the operating context of the UE 110 is outside the validity domain. A UE 110 may move into or out of a validity domain. For example, if a validity domain relates to an area served by a single gNB, and the UE 110 moves from a position outside said area to a position inside said area, the UE 110 moves into the validity domain, and if the UE 110 moves from a position inside said area to a position outside said area, the UE 110 moves out of the validity domain. A UE 110 may move from a first validity domain into a second validity domain. For example, if a first validity domain relates to a first area served by a first gNB and a second validity domain relates to a second area served by a second gNB, a UE 110 moving from a position in the first area to a position in the second area moves from the first validity domain into the second validity domain. In another example, if a first validity domain relates to any times between 00:00 and 06:00 and a second validity domain relates to any times between 06:01 and 08:00, at 06:01 the UE 110 moves from the first validity domain to the second validity domain. In some, but not necessarily all, examples, a model identifier 160 is valid if a current operating context of the UE 110 is within the validity domain of the model identifier 160. In some, but not necessarily all, examples, a model identifier 160 is invalid if a current operating context of the UE 110 is not within the validity domain of the model identifier 160. In some, but not necessarily all, examples, an AI / ML model is valid if a current operating context of the UE 110 is within the validity domain of the AI / ML model. In some, but not necessarily all, examples, an AI / ML model is invalid if a current operating context of the UE 110 is not within the validity domain of the AI / ML model. An invalidity event causes a model identifier 160 to become invalid. In some, but not necessarily all, examples, an invalidity event causes an AI / ML model to become invalid. In some, but not necessarily all, examples, an invalidity event occurs when an operating context of the UE 110 changes, and the new operating context of the UE 110 is not within the validity domain of a current model identifier 160. The UE 110 may stop using the current, invalid, model identifier 160 and start using a different, valid, model identifier 160. The UE 110 may additionally stop using a machine learning model associated with the current, invalid model identifier 160 and start using a machine learning model associated with the different, valid model identifier 160. In some, but not necessarily all, examples, an invalidity event is caused by the UE 110 being served by a new cell under the same network node 120 or a new cell under a new network node 120. In some, but not necessarily all, examples, a spatial validity domain corresponds to one of: a radio beam; a radio cell; a radio network node; a RAN notification area (RNA); a Registration Area (RA); or a Public Land Mobile Network (PLMN) serving the UE. In some examples in which the validity domain is a spatial validity domain, an invalidity event is a mobility event of the UE. For example, the UE 110 may move out of an area covered by the spatial validity domain, such as an area served by a network node 120. In some, but not necessarily all, examples, a temporal validity domain corresponds to one of: a time of day; a day; a week; or a month. In some examples in which the validity domain is a temporal validity domain, the invalidity event is a temporal event relating to a temporal change of corresponding KPIs. In some, but not necessarily all, examples, the invalidity event is a UE-side performance of the functionality falling below a minimum level. In some, but not necessarily all, examples, the UE 110 switches model identifier 160 to a different, valid model identifier 160 (the different model identifier 160 does not have an invalidity event). In some, but not necessarily all, examples, the UE 110 switches model identifier 160 to a different model identifier 160 with a broader validity domain. The validity domains of the one or more machine learning supported functionalities of the UE 110 are related to the validity domains of the one or more model identifier levels 150. In some, but not necessarily all, examples, the one or more machine learning supported functionalities of the UE 110 are associated with one or more validity domains. In some, but not necessarily all, examples, a validity domain of a machine learning supported functionality of the UE 110 does not indicate an accuracy of the machine learning supported functionality of the UE. In some, but not necessarily all, examples, the UE 110 has at least one Model ID corresponding to each of the model identifier levels 150 configured for the UE. For example, in the example of FIG. 3, the is configured with four model identifier levels 150 and therefore has at least four model identifiers available. The at least one model identifier 160 is identified via an online or offline identification procedure In some, but not necessarily all, examples, the configuration of the Model identifier levels 150 is UE 110 specific; in other words, not all levels have to be configured for all UEs communicating with a NW. The NW performs the configuration of the Model ID levels at the time of UE 110 capability exchange. A model identifier 160 is linked to (be valid for) a specific ML-enabled Feature / Feature Group, for example, beam management or CSI feedback. In some, but not necessarily all, examples, a UE 110 may not use all levels due to at least one of: validity / performance / priority constraints or special UE conditions, applications, use cases. The UE 110 support for a particular Model ID levels can be determined at the UE 110 capability exchange and / or during the online model identification procedure. In some, but not necessarily all, examples, the model identification configuration indicates a hierarchy of identifier levels 150 associated with corresponding different validity domains within a hierarchy of validity domains. FIG. 3 illustrates example model identifier levels 150 following a hierarchical arrangement. In FIG. 3, each descending level indicates an increasing granularity of the validity domain of the relevant ML supported functionality. An increased granularity of the validity domain provides higher resolution for the model identifier 160. For example, a broad spatial validity domain (with a low granularity) covers a wide area, and a narrow spatial validity domain (with a high granularity) covers a narrow area. Therefore, a broader validity domain is less specific than a narrow validity domain, but it is less likely that the UE 110 leaves the validity domain (for example, by leaving the area served by a cell). Therefore, there is less need for change of model identifiers with a broad validity domain. In some, but not necessarily all, examples, an nth model identifier level 150 has a broader validity domain than an (n+1)th model identifier level 150. In some, but not necessarily all, examples, an nth model identifier level 150 has a narrower validity domain than an (n-1)th model identifier level 150. Thus, in the example of FIG. 3, level 0 model identifiers may be used for a wide range of conditions. For example, level 0 model identifiers may have a PLMN level validity domain. Continuing the example of FIG. 3, multiple level 1 model identifiers may be used as required, said model identifiers having a narrower validity domain than level 0 model identifiers. For example, level 1 model identifiers may have a RNA level validity domain. Using the same example, level 2 model identifiers may have a gNB (multiple cells) level validity domain and level 3 model identifiers may have a cell (multiple beams) level validity domain. In another example, model identifier levels 150 may be assigned based on their temporal validity. In such an example, Level 0 has a largest temporal validity, and Level 3 has a shortest temporal validity. Temporal validity refers to a time period for which the model identifier 160 can be assumed to be valid. In another example, Level 0 model identifiers are identifier via offline identification, and are based on agreements between UE vendors and NW vendors. In this example, Level 1 is CSP / MNO specific and provides model identifiers which are linked to either UE capability categories and / or spatial and / or temporal characteristics of the network deployment; Level 2 is RA specific; Level 3 is RNA specific; and Level 4 is gNB / cell specific. In some, but not necessarily all, examples, different model identifiers at different model identifier levels 150 are used simultaneously by the UE. In some such examples, the levels are configured such that the model identifiers from all model identifier levels 150 are valid simultaneously for the same UE; for example, the model identifier levels 150 may always be related to the UE CONNECTED mode, at different geographical granularities. In other examples, different model identifiers at different model identifier levels 150 may not be used simultaneously. In some such examples, the different model identifiers are configured for use when the UE 110 is in different RRC states IDLE and / or INACTIVE, and / or other CM states. In some, but not necessarily all, examples, the Model ID can be linked to Standalone non-public network (SNPN), operated by NPN operators, which, in general, combination of PLMN ID and NID and may not be globally unique across PLMN list. In some, but not necessarily all, examples, within a PLMN, Model ID to be linked with restrictive cells using Closed Access Group (CAG) ID. In some, but not necessarily all, examples, Model ID can be linked to network slicing or a slice group using NSAG ID within a tracking area. In some, but not necessarily all, examples, the format or representation (such as bit lengths and / or structures) of model identifiers depends on the model identifier level 150 of the model identifier 160. In some, but not necessarily all, examples, model identifiers having the same model identifier level 150 have the same validity domain. In some, but not necessarily all, examples, model identifiers having the same model identifier level 150 have the same format. In the example of FIG. 3, model identifiers on the same model identifier level 150 (for example, Level 0, Level 1, and so on) have the same format. In some, but not necessarily all, examples, model identifiers having different model identifier levels 150 have different validity domains. In some, but not necessarily all examples, model identifiers having different model identifier levels 150 have different formats. In some, but not necessarily all, examples, a model identifier 160 having a narrower validity domain have a shorter format than a model identifier 160 having a broader validity domain. The model ID format therefore changes between different levels such that lower levels use shorter formats in order to allow faster and more localized management of the corresponding model IDs. A format of a model identifier 160 indicates the model identifier level 150 to which it belongs. FIG. 3 shows an example with listed identifier values 0_1 for level 0, 0_1_1, 0_1_2 for level 1, and so on. The actual model ID format definition (bits / bytes) is not discussed in this application. The Model identifiers indicated in FIG. 3 and the examples given should be understood as ‘placeholders’ only for illustration purposes. In some, but not necessarily all, examples, model identifier levels 150 are preserved via NAS procedures that terminate at the AMF. The NAS procedures may control registration management (RM) and connection management (CM) state machines and procedures with the UE. FIGs 20A and 20B illustrate examples of how online model identification can be triggered during CM states. In the examples of FIGs 20A and 20B, a UE 110 may be preserved or assigned in different connection modes. In such examples, model identification can occur during establishment of a connection between the network and the UE. Model identifiers may be used as follows in IDLE / CONNECTED / INACTIVE modes: Level_0: Identified via online or offline model identification procedure; broadest validity domain; for IDLE mode UEs. Level_1: Identifier via online model identification procedure; second broadest validity domain; for INACTIVE mode UEs. Level_2: Identifier via online model identification procedure; third broadest validity domain; for ACTIVE and INACTIVE mode UEs. Level_3: identifier via online model identification procedure; narrowest validity domain; for ACTIVE mode UEs. When moves into ACTIVE mode follow process for “moving down levels” as set out below. Referring back to FIG. 3, an exemplary configuration and use of hierarchical model identifier levels 150 is described below. For example purposes, it is assumed that a group of UEs and all UEs in the group are configured with four levels of model identifiers. It is further assumed that the UEs have performed a UE capabilities exchange procedure with the network at least once. In this example, the model identifier levels 150 are configured based on their spatial validity. At Level 0, a model identifier 160 indicates applicability conditions valid for all cells in a PLMN or a specified Registration Area (RA), comprising a list of tracking area identities (TAI List) as allocated by the AMF. The model identifiers at Level 0 are identified via an online or offline model identification procedure and are configured such that IDLE mode UEs can start using said model identifiers when starting a new RRC connection, irrespective of gNB characteristics, traffic patterns, etc, where the new connection is established. All gNBs in the RA also have this model identifier information stored for all UEs registered as being in the RA. Each model identifier 160 at Level 0 (0_1, 0_2, ..., 0_n) contains at least two fields: a first field indicating the level to which the model identifier 160 belongs and a field indicating the association (or no association) of the model identifier 160 with the UE configured RA. Additional fields may be included in the model identifier 160 to contain other AI / ML model related information, such as temporal identifiers. At Level 1, a model identifier 160 indicates applicability conditions valid for all cells in a specified RAN Notification Area (RNA). The RNA can be a subset of cells configured in the UE’s Registration Area or all cells configured in the UE’s Registration Area. The model identifiers at level 1 are identified via an online model identification procedure and are configured such that INACTIVE mode UEs can start using said model identifiers when re-starting a new RRC connection, irrespective of gNB characteristics, traffic patterns, etc, where the new connection is re-established. Each model identifier 160 at Level 1 (0_1_1, 0_1_2, ...,0_1_n) contains at least two fields: a first field indicating the level to which the model identifier 160 belongs and a field indicating the association (or no association) of the model identifier 160 with the configured RAN-based Notification Area Code. Additional fields may be included in the model identifier 160 to contain other AI / ML model related information, such as temporal identifiers. At Level 2, a model identifier 160 indicates applicability conditions valid for a gNB and its selected cells, for example, for a specific frequency layer. The model identifiers at level 2 are identified via an online model identification procedure and are configured such that ACTIVE and INACTIVE mode UEs can start using said model identifiers when moving between cells or re-starting a new RRC connection, irrespective of cell characteristics, traffic patterns, etc, where the new connection is re-established. Each model identifier 160 at Level 2 (0_1_1_1, 0_1_1_2, ... 0_1_1_n) contains at least two fields: a first field indicating the level to which the model identifier 160 belongs and a second field indicating the associated list of cell IDs (list of PCIs). Additional fields may be included in the model identifier 160 to contain other AI / ML model related information, such as temporal identifiers. At Level 3, a model identifier 160 indicates applicability conditions valid for a specific cell and its relevant characteristics. The model identifiers at level 3 are identified via an online model identification procedure and are configured such that ACTIVE mode UEs can use the Model ID specified at this level when connected to the respective cell. Each model identifier 160 at Level 3 (0_1_1_1_1, 0_1_1_1_2, ..., 0_1_1_1_n) contains at least two fields: a first field indicating the level to which the model identifier 160 belongs and a second field indicating the associated cell ID (PCI). Additional fields may be included in the model identifier 160 to contain other AI / ML model related information, such as temporal identifiers. In the example illustrated in FIG. 3, only one RA and one RNA are configured for a UE, and therefore there is only one model identifier 160 at each of Level 0 and Level 1. There may be multiple model identifiers at each of Level 2 and Level 3. In some, but not necessarily all, examples, the UE 110 comprises means for managing implementation of the first ML model and managing validity of the first ML model using associated validity domain. Referring back to FIG. 2, the method comprises, at block 208, establishing an RRC connection between the UE 110 and the NW. The method comprises, at block 210, initiating a model identification procedure between the NW and the UE. In dependence upon the initiation of the model identification procedure, the method comprises, at block 212, receiving, from the network node 120, a model identification configuration. In some, but not necessarily all, examples, the model identification configuration configures N model identifier levels 150 comprising at least N model identifiers, where N is the number of model identifier levels 150 the UE 110 is capable of configuring, as indicated by the multi-level identifier capability. The model identification configuration comprises at least a first model identifier 160 identifying one or more first machine learning supported functionalities of the UE 110 and a second model identifier 160 identifying one or more second machine learning supported functionalities of the UE. The one or more first machine learning supported functionalities of the UE 110 are associated with one or more first model identifier levels 150 and the one or more second machine learning supported functionalities of the UE 110 are associated with one or more second model identifier levels 150. The model identification configuration identifies at least one model identifier 160 for each of the configured model identifier levels 150. In some, but not necessarily all, examples, the method comprises receiving N model identification configurations from the NW, where N is the number of configured model identifier levels 150. In some such examples, the N model identification configurations are transmitted separately. In some, but not necessarily all, examples, the method comprises receiving a single model identification configuration from the NW. The single model identification configuration comprises information relating to all of the configured model identifier levels 150. A lowest level (for example, Level 0 in FIG. 3) may be identified via an offline identification procedure. Block 212 may thus be understood as an initial conditional configuration for the usage of the different model identifiers. In some, but not necessarily all, examples, the model identification configuration comprises a switching configuration. The switching configuration configures the UE 110 with one or more first reconfiguration conditions for switching from using the first identifier identifying the one or more first machine learning supported functionalities of the UE 110 to using the second identifier identifying the one or more second machine learning supported functionalities of the UE. In such an example, the second model identifier 160 has a broader validity domain than the first model identifier 160. The switching configuration thus enables the UE 110 to switch from using a current model identifier 160 to a new model identifier 160. For example, the switching configuration may indicate, for example, if the UE 110 leaves a current beam, cell, group of cells or RNA and / or is handed over to another beam, cell, group of cells or RNA, that the UE 110 should switch to using a different model identifier 160 that is one or more levels higher in the hierarchy than the current model identifier 160. In some, but not necessarily all, examples, the switching configuration is transmitted / received with the model identification configuration or the model identification configuration comprises the switching configuration. In other examples, the switching configuration is transmitted / received separately to the model identification configuration, for example the switching configuration may be transmitted / received after a machine learning supported functionality of the UE 110 has been activated. In other examples, no switching configuration is transmitted / received. The method comprises, at block 214, (by the NW) configuring and activating the use of one of the one or more first machine learning functionalities of the UE 110 based on the model identification configuration. At block 216, the method comprises receiving, from the NW, a switching configuration. The switching configuration may be a switching configuration as described above. In some examples, block 216 of the method is not performed. In some such examples, the switching configuration is transmitted / received with the model identification configuration or the model identification configuration comprises the switching configuration. In other such examples, no switching configuration is transmitted / received. At block 218, the method comprises using one of the one or more first machine learning supported functionalities of the UE 110 based on the received model identification configuration. In some, but necessarily all, examples, the machine learning supported functionality that is used is associated with a model identifier 160 at the lowest available configured model identifier level 150. At block 220, the method comprises determining that an invalidity event has occurred. The invalidity event invalidates the use of the first identifier of the one or more first machine learning supported functionalities of the UE. The invalidity event may be an invalidity event as described above. In some, but not necessarily all, examples, the UE 110 is configured to detect an invalidity event. In dependence upon determining that an invalidity event has occurred, the method proceeds to method 700 (illustrated in FIG 7) or method 1000 (illustrated in FIG. 10). Consequently, FIG. 2 illustrates a method comprising: at block 202, receiving, from a network node 120, a request for a multi-level identifier capability of the UE; at block 204, transmitting, to the network node 120, a multi-level identifier capability; at block 208, establishing an RRC connection between the UE 110 and the NW; at block 210, initiating a model identification procedure between the NW and the UE; at block 212, receiving, from the network node 120, a model identification configuration; at block 214, configuring and activating the use of one of the one or more first machine learning functionalities of the UE 110 based on the model identification configuration; at block 216, receiving, from the NW, a switching configuration; and at block 218, using one of the one or more first machine learning supported functionalities of the UE 110 based on the received model identification configuration; and at block 220, determining that an invalidity event has occurred. From the point of view of the network node 120, at block 202, the method comprises transmitting, to a UE, a request for a multi-level identifier capability of the UE. The method comprises, at block 204, receiving, from the UE, a multi-level identifier capability of the UE. The method comprises, at block 206, determining a model identification configuration, based on the received multi-level identifier capability. In some, but not necessarily all, examples, the model identification configuration configures N model identifier levels 150 comprising at least N model identifiers, where N is the number of model identifier levels 150 the UE 110 is capable of configuring, as indicated by the multi-level identifier capability. The method comprises, at block 208, establishing an RRC connection between the UE 110 and the NW. The method comprises, at block 210, initiating a model identification procedure. In dependence upon the initiation of the model identification procedure, the method comprises, at block 212, transmitting, to the UE, the model identification configuration. The method comprises, at block 214, configuring and activating the use of one of the one or more first machine learning functionalities of the UE 110 based on the model identification configuration. The method comprises, at block 216, transmitting, to the UE, a switching configuration. In some examples, block 216 of the method is not performed. In some such examples, the switching configuration is transmitted / received with the model identification configuration or the model identification configuration comprises the switching configuration. In other such examples, no switching configuration is transmitted / received. The method comprises, at block 220, the method comprises determining that an invalidity event has occurred. The invalidity event invalidates the use of the first identifier of the one or more first machine learning supported functionalities of the UE. In some, but not necessarily all, examples, the NW is configured to detect an invalidity event. In dependence upon determining that an invalidity event has occurred, the method proceeds to method 700 (illustrated in FIG 7) or method 1000 (illustrated in FIG 10). Consequently, FIG 2 illustrates a method comprising: at block 202, transmitting, to a UE, a request for a multi-level identifier capability of the UE; at block 204, receiving, from the UE, a multi-level identifier capability of the UE; at block 206, determining a model identification configuration, based on the received multi-level identifier capability; at block 208, establishing an RRC connection between the UE 110 and the NW; at block 210, initiating a model identification procedure; at block 212, transmitting, to the UE, the model identification configuration; at block 214, configuring and activating the use of one of the one or more first machine learning functionalities of the UE 110 based on the model identification configuration; at block 216, transmitting, to the UE, a switching configuration; and at block 220, determining that an invalidity event has occurred. FIG. 4 illustrates a method which is a sub-method of FIG 2. The method comprises blocks 204, 212, and 216 of FIG. 2. Consequently, from the point of view of the UE, FIG 4 illustrates a method comprising: at block 402, transmitting, to the network node 120, a multi-level identifier capability; at block 404, receiving, from the network node 120, a model identification configuration; and at block 406, using one of the one or more first machine learning supported functionalities of the UE 110 based on the received model identification configuration. Consequently, from the point of view of the NW, FIG 4 illustrates a method comprising: at block 402, receiving, from the UE, a multi-level identifier capability; and at block 404, transmitting, to the UE, a model identification configuration. FIG. 5 illustrates an example of a method 500. The method can be performed by any suitable apparatus comprising any suitable means for performing the method, for example an apparatus as described in relation to FIG. 17. In examples, the method can be performed by a terminal node 110, such as a UE 110 140. At block 502, the method comprises receiving, from a network node 120, a request for a multi-level identifier capability of the UE. At block 504, the method comprises transmitting, to the network node 120, a multilevel identifier capability. At block 506, the method comprises establishing an RRC connection between the UE 110 and the NW. At block 508, the method comprises initiating a model identification procedure between the NW and the UE. At block 510, the method comprises receiving, from the network node 120, a model identification configuration. At block 512, the method comprises configuring and activating the use of one of the one or more first machine learning supported functionalities of the UE 110 based on the model identification configuration. At block 514, the method comprises using one of the one or more first machine learning supported functionalities of the UE 110 based on the received model identification configuration. At block 516, the method comprises receiving, from the NW, a switching configuration. At block 518, the method comprises determining that an invalidity event has occurred. Consequently, FIG. 5 illustrates a method comprising: at block 502, receiving, from a network node 120, a request for a multi-level identifier capability of the UE; at block 504, transmitting, to the network node 120, a multi-level identifier capability; at block 506, establishing an RRC connection between the UE 110 and the NW; at block 508, initiating a model identification procedure between the NW and the UE; at block 510, receiving, from the network node 120, a model identification configuration; at block 512, configuring and activating the use of one of the one or more first machine learning supported functionalities of the UE 110 based on the model identification configuration; at block 514, using one of the one or more first machine learning supported functionalities of the UE 110 based on the received model identification configuration; at block 516, receiving, from the NW, a switching configuration; and at block 518, determining that an invalidity event has occurred. FIG. 6 illustrates an example of a method 600. The method can be performed by any suitable apparatus comprising any suitable means for performing the method, for example an apparatus as described in relation to FIG. 17. In examples, the method can be performed by an access node, such as a gNB. The method comprises, at block 602, transmitting, to a UE, a request for a multi-level identifier capability of the UE. The method comprises, at block 604, receiving, from the UE, a multi-level identifier capability of the UE. The method comprises, at block 606, determining a model identification configuration, based on the received multi-level identifier capability. The method comprises, at block 608, establishing an RRC connection between the UE 110 and the NW. The method comprises, at block 610, initiating a model identification procedure. The method comprises, at block 612, transmitting, to the UE, the model identification configuration. The method comprises, at block 614, configuring and activating the use of one of the one or more first machine learning supported functionalities of the UE 110 based on the model identification configuration. The method comprises, at block 616, transmitting, to the UE, a switching configuration. The method comprises, at block 618, determining that an invalidity event has occurred. Consequently, FIG. 6 illustrates a method comprising: at block 602, transmitting, to a UE, a request for a multi-level identifier capability of the UE; at block 604, receiving, from the UE, a multi-level identifier capability of the UE; at block 606, determining a model identification configuration, based on the received multi-level identifier capability; at block 608, establishing an RRC connection between the UE 110 and the NW; at block 610, initiating a model identification procedure; at block 612, transmitting, to the UE, the model identification configuration; at block 614, configuring and activating the use of one of the one or more first machine learning supported functionalities of the UE 110 based on the model identification configuration; at block 616, transmitting, to the UE, a switching configuration; and at block 618, determining that an invalidity event has occurred. FIG. 7 illustrates a method which is optionally performed after method 200. In some, but not necessarily all, examples, the method 700 is performed if no switching configuration is received from the NW. For example, method 700 is performed if block 218 is not performed and / or if the model identification configuration does not comprise a switching configuration. From the point of view of the UE, the method comprises, at block 704, receiving, from the NW, an indication to use a second model identifier 160 identifying one or more second machine learning supported functionalities of the UE. In some, but not necessarily all, examples, the second model identifier 160 is a model identifier 160 received in the model identifier configuration. Some or all of the blocks of method 300 may thus be carried out without a further model identification step. The NW assumes that the UE 110 can use the model identifier level 150 identified at block 212. The method comprises, at block 706, receiving, from the network, a model identifier level indicator. In some, but not necessarily all, examples, block 706 is an alternative to block 304; that is, the method comprises block 704 or block 706 but not both. In other examples, the method comprises block 704 and block 706. The model identifier level indicator indicates that the UE 110 should use a model identifier level 150 that is higher in the hierarchy than the current model identifier level 150. In some such examples, the model identifier level indicator does not comprise an indication of which specific model identifier 160 the UE 110 should use. In some such examples, the UE 110 determines a new model identifier 160 of the indicated model identifier level 150. The new model identifier 160 is a model identifier 160 identified in the model identification configuration. In some, but not necessarily all, examples, such as the example illustrated in FIG [B], the indicator is a two-bit word in which the four model identifier levels 150 are mapped to the available combinations 00, 01,10 and 11. Similar approaches can be used in cases with more or fewer model identifier levels 150 and / or in cases in which more than one model identifier 160 is identified for each model identifier level 150. The method comprises, at block 708, configuring and activating one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier 160. The method comprises, at block 710, using one of the one or more second machine learning supported functionalities of the UE 110 based on the second model identifier 160. In some, but not necessarily all, examples, block 710 further comprises stopping use of the one of the one or more first machine learning supported functionalities of the UE 110 based on the invalidity event. In some, but not necessarily all, examples, a model identifier 160 at a model identifier level 150 that is not the lowest configured model identifier level 150 is a fallback model identifier 160. For example, the second model identifier 160 is a fallback model identifier 160. In some, but not necessarily all, examples, a fallback model identifier indicates a nonmachine learning based solution; in other words, a fallback model identifier may not identify a machine learning supported functionality of the UE. In other examples, a fallback model identifier identifies a machine learning supported functionality of the UE. In some, but not necessarily all, examples, a further invalidity event may invalidate the use of the second model identifier 160. In some such examples, the method described above at blocks 702 - 708 is repeated, using a third model identifier 160 having a third model identifier level 150 with a validity domain in which the operating context of the UE 110 falls. The third model identifier 160 is a model identifier 160 identified by the model identification configuration and identifies one or more third machine learning supported functionalities of the UE. The third model identifier 160 has a broader validity domain than each of the first model identifier 160 and the second model identifier 160. Blocks 714 - 718 are optionally carried out if the current model identifier 160 is a fallback model identifier 160. At block 714, the method comprises receiving, from the NW, an indication to use a fourth model identifier 160. In some, but not necessarily all, examples, the fourth model identifier 160 is a model identifier 160 received in the model identifier configuration, the fourth model identifier 160 having a fourth model identifier level 150 and identifying one or more fourth machine learning supported functionalities of the UE. The fourth model identifier level 150 is lower in the hierarchy than the current model identifier 160. In other examples, the fourth model identifier 160 is a new model identifier 160 not received in the model identifier configuration. In some such examples, the fourth model identifier 160 is determined via a model identification procedure. The method comprises, at block 716, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier 160. The method comprises, at block 718, using one of the one or more fourth machine learning supported functionalities of the UE 110 based on the fourth model identifier 160. In some, but not necessarily all, examples, if the fourth model identifier level 150 is the lowest configured model identifier level, the UE 110 continues to use the one or more fourth machine learning supported functionalities of the UE 110 until a further invalidity event occurs. In some, but not necessarily all, examples, if the fourth model identifier level 150 is not the lowest configured model identifier level, blocks 714-718 are repeated until a model identifier 160 at the lowest configured model identifier level is determined. Consequently, FIG 7 illustrates a method comprising: at block 704, receiving, from the NW, an indication to use a second model identifier 160 identifying one or more second machine learning supported functionalities of the UE; at block 706, receiving, from the network, a model identifier level indicator; at block 708, configuring and activating one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier; at block 710, using one of the one or more second machine learning supported functionalities of the UE 110 based on the second model identifier; at block 714, receiving, from the NW, an indication to use a fourth model identifier; at block 716, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier; and at block 718, using one of the one or more fourth machine learning supported functionalities of the UE 110 based on the fourth model identifier 160. From the point of view of the NW, the method comprises, at block 702, determining a second model identifier 160 having a second model identifier level 150 with a validity domain in which the operating context of the UE 110 falls. The determined second model identifier 160 has a broader validity domain than the current model identifier 160. The second model identifier 160 identifies one or more second machine learning supported functionalities of the UE. In some, but not necessarily all, examples, the determination of the second model identifier 160 is carried out in dependence upon determining that an invalidity event has occurred. In the example of FIG. 7, the second model identifier level 150 is one step higher in the hierarchy than the current model identifier level 150. In an alternative example, the model identifier level 150 that is one step higher in the hierarchy than the current model identifier level 150 does not have a validity domain in which the operating context of the UE 110 falls. In such an example, the second determined model identifier level 150 is two or more steps higher in the hierarchy than the current model identifier level 150. In some such examples, the UE 110 transmits to the NW an indicator indicating that the determined second model identifier level 150 is two or more steps higher in the hierarchy than the current model identifier level 150. At block 704, the method comprises transmitting, to the UE, an indication to use the second model identifier 160 identifying the one or more second machine learning supported functionalities of the UE. In some, but not necessarily all, examples, the second model identifier 160 is a model identifier level 150 received in the model identifier configuration. Some or all of the blocks of method 700 may thus be carried out without a further model identification step. The NW assumes that the UE 110 can use the model identifier level 150 identified at block 212. The method comprises, at block 706, transmitting, to the UE, a model identifier level indicator. In some, but not necessarily all, examples, block 706 is an alternative to block 704; that is, the method performs block 704 or block 706 but not both. In other examples, the method performs block 704 and block 706. The method comprises, at block 708, configuring and activating the use of one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier 160. In some, but not necessarily all, examples, a further invalidity event may invalidate the use of the second model identifier 160. In some such examples, the method described above at blocks 702 - 708 is repeated, using a third model identifier 160 having a third model identifier level 150 with a validity domain in which the operating context of the UE 110 falls. The third model identifier 160 is a model identifier 160 identified by the model identification configuration and identifies one or more third machine learning supported functionalities of the UE. The third model identifier 160 has a broader validity domain than each of the first model identifier 160 and the second model identifier 160. Blocks 712 - 718 are optionally performed if the current model identifier 160 is a fallback model identifier 160. The method comprises, at block 712, determining whether a fourth model identifier 160 level, lower in the hierarchy than the current model identifier 160, has a validity domain in which the operating context of the UE 110 falls. If it is determined that no model identifier 160, lower in the hierarchy than the current model identifier 160, has a validity domain in which the operating context of the UE 110 falls, the method does not proceed. If it is determined that there is a model identifier 160, lower in the hierarchy than the current model identifier 160, having a validity domain in which the operating context of the UE 110 falls, the method proceeds to step 714. At block 714, the method comprises transmitting, to the UE, an indication to use the fourth model identifier 160 In some, but not necessarily all, examples, the fourth model identifier 160 is a model identifier 160 received in the model identifier configuration, the fourth model identifier 160 having a fourth model identifier level 150 and identifying one or more fourth machine learning supported functionalities of the UE. The fourth model identifier level 150 is lower in the hierarchy than the current model identifier 160. In other examples, the fourth model identifier 160 is a new model identifier 160 not received in the model identifier 160 configuration. In some such examples, the fourth model identifier 160 is determined via a model identification procedure. The method comprises, at block 716, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier 160. In some, but not necessarily all, examples, if the fourth model identifier level 150 is not the lowest configured model identifier level, blocks 712-716 are repeated until a model identifier 160 at the lowest configured model identifier level is determined. Consequently, FIG 7 illustrates a method comprising: at block 702, determining a second model identifier 160 having a second model identifier level 150 with a validity domain in which the operating context of the UE 110 falls; at block 704, transmitting, to the UE, an indication to use the second model identifier 160 identifying the one or more second machine learning supported functionalities of the UE; at block 706, transmitting, to the UE, a model identifier level indicator; at block 708, configuring and activating the use of one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier; at block 712, determining whether a configured model identifier level, lower in the hierarchy than the current model identifier 160, has a validity domain in which the operating context of the UE 110 falls; at block 716, transmitting, to the UE, an indication to use the fourth model identifier; and at block 718, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier 160. FIG. 8 illustrates an example of a method 800. The method can be performed by any suitable apparatus comprising any suitable means for performing the method, for example an apparatus as described in relation to FIG. 17. In examples, the method can be performed by a terminal node 110, such as a UE. The method comprises, at block 802, receiving, from the NW, an indication to use a second model identifier 160 identifying one or more second machine learning supported functionalities of the UE. The method comprises, at block 804, receiving, from the network, a model identifier level indicator. The method comprises, at block 806, configuring and activating one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier 160. The method comprises, at block 808, using one of the one or more second machine learning supported functionalities of the UE 110 based on the second model identifier 160. The method comprises, at block 810, receiving, from the NW, an indication to use a fourth model identifier 160. The method comprises, at block 812, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier 160. The method comprises, at block 814, using one of the one or more fourth machine learning supported functionalities of the UE 110 based on the fourth model identifier 160. Consequently, FIG. 8 illustrates a method comprising: at block 802, receiving, from the NW, an indication to use a second model identifier 160 identifying one or more second machine learning supported functionalities of the UE; at block 804, receiving, from the network, a model identifier level indicator; at block 806, configuring and activating one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier; at block 808, using one of the one or more second machine learning supported functionalities of the UE 110 based on the second model identifier; at block 810, receiving, from the NW, an indication to use a fourth model identifier; at block 812, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier; and at block 814, using one of the one or more fourth machine learning supported functionalities of the UE 110 based on the fourth model identifier 160. FIG. 9 illustrates an example of a method 900. The method can be performed by any suitable apparatus comprising any suitable means for performing the method, for example an apparatus as described in relation to FIG. 17. In examples, the method can be performed by an access node, such as a gNB. The method comprises, at block 902, determining a second model identifier 160 having a second model identifier level 150 with a validity domain in which the operating context of the UE 110 falls. The method comprises, at block 904, transmitting, to the UE, an indication to use the second model identifier 160 identifying the one or more second machine learning supported functionalities of the UE. The method comprises, at block 906, transmitting, to the UE, a model identifier level indicator. The method comprises, at block 908, configuring and activating the use of one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier 160. The method comprises, at block 910, determining whether a configured model identifier level, lower in the hierarchy than the current model identifier 160, has a validity domain in which the operating context of the UE 110 falls. The method comprises, at block 912, transmitting, to the UE, an indication to use the fourth model identifier 160. The method comprises, at block 914, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier 160. Consequently, FIG. 9 illustrates a method comprising: at block 902, determining a second model identifier 160 having a second model identifier level 150 with a validity domain in which the operating context of the UE 110 falls; at block 904, transmitting, to the UE, an indication to use the second model identifier 160 identifying the one or more second machine learning supported functionalities of the UE; at block 906, transmitting, to the UE, a model identifier level indicator; at block 908, configuring and activating the use of one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier; at block 910, determining whether a configured model identifier level, lower in the hierarchy than the current model identifier 160, has a validity domain in which the operating context of the UE 110 falls; at block 912, transmitting, to the UE, an indication to use the fourth model identifier; and at block 914, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier 160. FIG 10 illustrates a method which is optionally performed after method 200. In some, but not necessarily all, examples, the method 1000 is performed if a switching configuration is received from the NW. For example, method 1000 is performed if block 218 is performed and / or if the model identification configuration comprises a switching configuration. From the point of view of the UE, the method comprises, at block 1002, determining that at least one of the one or more reconfiguration conditions has been met. In some, but not necessarily all, examples, at least one of the reconfiguration conditions for switching from using the first identifier identifying the one or more first machine learning supported functionalities of the UE 110 to using the second identifier identifying the one or more second machine learning functionalities of the UE 110 is an invalidity event invalidating use of the first identifier of the one or more first machine learning supported functionalities of the UE. The method comprises, at block 1004, in response to a determination that at least one of the one or more reconfiguration conditions has been met, determining a second model identifier 160 having a second model identifier level 150 with a validity domain in which the operating context of the UE 110 falls. The determined second model identifier 160 has a broader validity domain than the current model identifier 160. The second model identifier 160 identifies one or more second machine learning supported functionalities of the UE. In the example of FIG 10, the second model identifier level 150 is one step higher in the hierarchy than the current model identifier level 150. In an alternative example, the model identifier level 150 that is one step higher in the hierarchy than the current model identifier level 150 does not have a validity domain in which the operating context of the UE 110 falls. In such an example, the second determined model identifier level 150 is two or more steps higher in the hierarchy than the current model identifier level 150. In some such examples, the UE 110 transmits to the NW an indicator indicating that the determined second model identifier level 150 is two or more steps higher in the hierarchy than the current model identifier level 150. The method comprises, at block 1006, configuring and activating one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier 160. The method comprises, at block 1008, using one of the one or more second machine learning supported functionalities of the UE 110 based on the second model identifier 160. In some, but not necessarily all, examples, block 1008 further comprises stopping use of the one of the one or more first machine learning supported functionalities of the UE 110 based on the determination that at least one of the one or more reconfiguration conditions has been met. In some, but not necessarily all, examples, a model identifier 160 at a model identifier level 150 that is not the lowest configured model identifier level 150 is a fallback model identifier 160. For example, the second model identifier 160 is a fallback model identifier 160. In some, but not necessarily all, examples, a fallback model identifier 160 indicates a non-machine learning based solution; in other words, a fallback model identifier 160 may not identify a machine learning supported functionality of the UE. In other examples, a fallback model identifier 160 identifies a machine learning supported functionality of the UE. In some, but not necessarily all, examples, a further reconfiguration condition is met. In some such examples, the method described above at blocks 1002 - 1008 is repeated, using a third model identifier 160 having a third model identifier level 150 with a validity domain in which the operating context of the UE 110 falls. The third model identifier 160 is a model identifier 160 identified by the model identification configuration and identifies one or more third machine learning supported functionalities of the UE. The third model identifier 160 has a broader validity domain than each of the first model identifier 160 and the second model identifier 160. Blocks 1010 - 1014 are optionally carried out if the current model identifier 160 is a fallback model identifier 160. The method comprises, at block 1010, determining whether a fourth model identifier level, lower in the hierarchy than the current model identifier 160, has a validity domain in which the operating context of the UE 110 falls. If it is determined that no model identifier 160, lower in the hierarchy than the current model identifier 160, has a validity domain in which the operating context of the UE 110 falls, the method does not proceed. If it is determined that there is a model identifier 160, lower in the hierarchy than the current model identifier 160, having a validity domain in which the operating context of the UE 110 falls, the method proceeds to step 1012. In some, but not necessarily all, examples, the fourth model identifier 160 is a model identifier 160 received in the model identifier configuration, the fourth model identifier 160 having a fourth model identifier level 150 and identifying one or more fourth machine learning supported functionalities of the UE. The fourth model identifier level 150 is lower in the hierarchy than the current model identifier 160. The method comprises, at block 1012, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier 160. The method comprises, at block 1014, using one of the one or more fourth machine learning supported functionalities of the UE 110 based on the fourth model identifier 160. In some, but not necessarily all, examples, if the fourth model identifier level 150 is the lowest configured model identifier level, the UE 110 continues to use the one or more fourth machine learning supported functionalities of the UE 110 until a further invalidity event occurs. In some, but not necessarily all, examples, if the fourth model identifier level 150 is not the lowest configured model identifier level, the UE 110 may repeat blocks 1010 -1014 until a model identifier 160 at the lowest configured model identifier level 150 is used. Consequently, FIG 10 illustrates a method comprising: at block 1002, determining that at least one of the one or more reconfiguration conditions has been met; at block 1004, in response to a determination that at least one of the one or more reconfiguration conditions has been met, determining a second model identifier 160 having a second model identifier level 150 with a validity domain in which the operating context of the UE 110 falls; at block 1006, configuring and activating one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier; at block 1008, using one of the one or more second machine learning supported functionalities of the UE 110 based on the second model identifier; at block 1010, determining whether a fourth model identifier level, lower in the hierarchy than the current model identifier 160, has a validity domain in which the operating context of the UE 110 falls; at block 1012, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier; and at block 1014, using one of the one or more fourth machine learning supported functionalities of the UE 110 based on the fourth model identifier 160. From the point of view of the NW, the method comprises, at block 1006, configuring and activating one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier 160. The method comprises, at block 1012, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier 160. Consequently, FIG 10 illustrates a method comprising: at block 1006, configuring and activating one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier; and at block 1012, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier 160. FIG. 11 illustrates an example of a method 1100. The method can be performed by any suitable apparatus comprising any suitable means for performing the method, for example an apparatus as described in relation to FIG. 17. In examples, the method can be performed by a terminal node 110, such as a UE. The method comprises, at block 1102, determining that at least one of the one or more reconfiguration conditions has been met. The method comprises, at block 1104, in response to a determination that at least one of the one or more reconfiguration conditions has been met, determining a second model identifier 160 having a second model identifier level 150 with a validity domain in which the operating context of the UE 110 falls. The method comprises, at block 1106, configuring and activating one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier 160. The method comprises, at block 1108, using one of the one or more second machine learning supported functionalities of the UE 110 based on the second model identifier 160. The method comprises, at block 1110, determining whether a fourth model identifier level, lower in the hierarchy than the current model identifier 160, has a validity domain in which the operating context of the UE 110 falls. The method comprises, at block 1112, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier 160. The method comprises, at block 1114, using one of the one or more fourth machine learning supported functionalities of the UE 110 based on the fourth model identifier 160. Consequently, FIG. 11 illustrates a method comprising: at block 1102, determining that at least one of the one or more reconfiguration conditions has been met; at block 1104, in response to a determination that at least one of the one or more reconfiguration conditions has been met, determining a second model identifier 160 having a second model identifier level 150 with a validity domain in which the operating context of the UE 110 falls; at block 1106, configuring and activating one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier; at block 1108, using one of the one or more second machine learning supported functionalities of the UE 110 based on the second model identifier; at block 1110, determining whether a fourth model identifier level, lower in the hierarchy than the current model identifier, has a validity domain in which the operating context of the UE 110 falls; at block 1112, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier; and at block 1114, using one of the one or more fourth machine learning supported functionalities of the UE 110 based on the fourth model identifier 160. FIG. 12 illustrates an example of a method 1200. The method can be performed by any suitable apparatus comprising any suitable means for performing the method, for example an apparatus as described in relation to FIG. 17. In examples, the method can be performed by an access node, such as a gNB. The method comprises, at block 1202, configuring and activating one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier 160. The method comprises, at block 1204, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier 160. Consequently, FIG. 12 illustrates a method comprising: at block 1202, configuring and activating one of the one or more second machine learning functionalities of the UE 110 based on the second model identifier; and at block 1204, configuring and activating one of the one or more fourth machine learning functionalities of the UE 110 based on the fourth model identifier 160. Generally, the use of 2-levels might solve some of the frequent model identification problems. However, any of the current use cases, require the UE 110 to have very powerful ML models, generally providing sufficiently good performance across the entire operator network (various geographical areas, various time of the day). This is very difficult to achieve with one or two Models, hence one or two Model ID, due to the complex nature of the network deployments. A 2-level approach would be OK for a localized solution where the UE 110 does not move (much or at all) or when the 2-level identification is applied more frequently in case the UE 110 is moving. “Moving” is an example case for the geographically based model identification, but the same approach can be used for time-of-the-day based identification, or a mixture. Furthermore, the use of more than 2 levels this could significantly reduce the frequency of model identification while UE 110 is mobile. This would also allow fast switching while UE 110 does not have a local model available or cannot validate it for some reason. A further example is described below, with reference to FIG. 3. For example purposes, we assume a group of UEs and all UEs in the group are configured with 4 levels of Model IDs for use in the ML-enabled CSI feedback feature. It is further assumed the UEs have performed at least once the UE 110 capabilities exchange procedure in the network. The Model ID levels are configured based on their spatial validity and are described as follows: Level_0: The Model ID indicates applicability conditions (directly or indirectly via mapping, not part of this invention) valid for all cells in whole PLMN or a specified Registration Area (RA) comprising a list of tracking area identities (TAI List) as allocated by the AMF. The Model IDs at this level (0_1, 0_2, ...) are identified via the (online or offline) model identification procedure and are configured such that IDLE mode UEs can assume to start using the Model ID specified at this level when starting a new RRC connection, irrespective of gNB characteristics, or traffic patterns, etc. where the new connection is established. All gNBs in the RA also have this Model ID information stored for all the UEs registered as being in the RA. How the Level 0 Model IDs are propagated (allocated by the operator) among the gNBs is not in the scope of this invention and for example, it can follow the same procedure as used for gNB ID allocation / planning. The mapping of these IDs to actual UE 110 models is part of the model identification procedure (and mostly implementation specific). The legacy mobility registration procedures are used to manage the Model ID at this level (registration, de-registration) Each Model ID at this level (0_1, 0_2, ...) contains at least two fields (mandatory): a) one field indicating the Level to which this Model ID belongs to (e.g. 3 bits for a total of 7 levels + ‘No Level used’ indication), and b) one field indicating the association or no association of the Model ID with the UE 110 configured RA (1 bit) Additional fields can be included in the Model ID which contain other ML Model related information (e.g. temporal IDs.). Level_1 The Model ID indicates applicability conditions valid for all cells in a specified RAN Notification Area (RNA). The RAN Notification area can be a subset of cells configured in UE's Registration Area or all cells configured in the UE's Registration Area [TS23.501], The Model IDs at this level (0_1_1, 0_1_2, ...) are identified via the online model identification procedure and are configured such that INACTIVE mode UEs can assume to start using the Model ID specified at this level when re-starting a new RRC connection, irrespective of gNB characteristics, or traffic patterns, etc. where the new connection is re-established. The usual RNA related procedures are used to manage also the Model ID at this level. Each RNA can be configured for a UE 110 with same or different applicable Model IDs. Each Model ID at this level (0_1_1, 0_1_2, ...) contains at least two fields (mandatory): a) one field indicating the Level to which this Model ID belongs to (e.g. 3 bits for a total of 7 levels + ‘No Level used’ indication), and b) one field indicating the association or no association of the Model ID with the configured RAN-based Notification Area Code (1 bit) Additional fields can be included in the Model ID which contain other ML Model related information (e.g. temporal IDs). Level_2 The Model ID indicates applicability conditions valid for a gNB and its selected cells, e.g. for a specific frequency layer The Model I Ds at this level (0_ 1 _ 1 _ 1, 0_ 1 _ 1 _2, ...) are identified via the online model identification procedure and are configured such that ACTIVE and INACTIVE mode UEs can assume to start using the Model ID specified at this level when moving between cells or re-starting a new RRC connection, irrespective of cell characteristics, or traffic patterns, etc. where the new connection is re-established. The legacy cell level mobility related procedures are used to manage the Model ID at this level. Each Model ID at this level (0_1_1_1, 0_1_1_2,...) contains at least two fields (mandatory): one field indicating the Level to which this Model ID belongs to (e.g. 3 bits for a total of 7 levels + ‘No Level used’ indication), and one field indicating the associated list of cell IDs (list of PCIs), directly or indirectly as index in a list of ‘list of Cell IDs’ Additional fields can be included in the Model ID which contain other ML Model related information (e.g. dataset ID, context ID). Level_3 The Model ID indicates applicability conditions valid for a specific cell and its relevant characteristics (NW-side additional condition) The Model IDs at this level (0_1_1_1_1, 0_1_1_2_1, ...) are identified via the online model identification procedure and are configured such that ACTIVE mode UEs can use the Model ID specified at this level when connected to the respective cell. Each Model ID at this level (0_1_1_1_1, 0_1_1_2_1, ...) contains at least two fields (mandatory): a) one field indicating the Level to which this Model ID belongs to (e.g. 3 bits for a total of 7 levels + ‘No Level used’ indication), and b) one field indicating the associated cell ID (PCI), directly or indirectly as index in a list Cell IDs Additional fields can be included in the Model ID which contain other ML Model related information (e.g. dataset ID, context ID, etc.). In the embodiment above, only one RA and one RNA are configured for a UE 110 for 3GPP access, therefore there is only one Model ID at each Level 0 and Level 1 which needs to be identified. For Level 2 and Level 3, there can be multiple Model IDs at each level. We show four main examples of how the Model ID Levels are being used, in different UE mobility (Step 8) and / or configuration scenarios (Step 5): FIG. 13- UE 110 changes serving cell under the same gNB, and uses one of the a priori identified Model ID levels in the new serving cell, as signaled by the NW. FIG. 14 - UE 110 changes serving cell and gNB, under the same RNA, and uses one of the a priori identified Model ID levels in the new serving cell, as signaled by the NW. FIG. 15 - UE 110 changes serving cell under the same gNB, and uses the associated conditionalReconfiguration signaled from NW prior the mobility event, to determine the Model ID level to use in the new serving cell. Exactly the same mechanism can also be used when the UE 110 changes serving cell and gNB. FIG. 16 - UE 110 changes serving cell under the same gNB, and uses the a priori associated conditionalReconfiguration to determine the Model ID level to use in the new serving cell. Exactly the same mechanism can also be used when the UE 110 changes serving cell and gNB. The method of FIG 13B continues directly from FIG. 13A; the method of FIG. 14B continues directly from FIG. 14A; the method of FIG. 15B continues directly from FIG. 15A; and the method of FIG. 16B continues directly from FIG. 16A. In both these scenarios, the initial configuration and model identifications steps 1-4 are the same. The differences are highlighted in the description below. Step 1-2: UE ML capability exchange procedure is performed resulting in UE ML Functionalities / Features / Feature Groups exposure to the NW. Step 3: The NW determines howto configure the Model ID Levels for the UE 110 based on the acquired UE ML capabilities in Step 1-2. In this example the NW configures N=4 levels of Model IDs as described above. Step 4 : The RRC connection establishment in the serving cell (and initial gNB). Step 5: Based on the outcome of Step 3, the NW initiates model identification procedure with the UE, including identification of at least one Model ID at each of the configured N=4 levels. Alternatively, the Model ID for Level 0 can also be identified via offline identification procedure. In FIGs 13 and 14, the Step 5.* can be understood as initial conditional configurations for the usage of the different Model IDs. In Step 5.*, the model identification procedure results in N=4 Model IDs identified, one for each configured level: 0_1, 0_1_1, 0_1_1_1 and 0_1_1_1_1. This can be achieved with N=4 identification procedures or one combined procedure. Alternatively, the Level 0 Model ID can also be identified via offline identification procedure when the network operator allows this and has the necessary agreements with the UE vendor. In these initial identification Steps 5.* can incorporate a conditionalReconfiguration information to be signaled from the NW to the UE 110 in one or more of the Step 5* (see Error! Reference source not found. 16). The new conditionalReconfiguration indication can indicate e.g., if UE 110 leaves the current beam / cell / group of cells / RNA and / or is handed over to another beam / cell / group of cells / RNA, UE 110 should start using the model ID with „X“ level higher in the hierarchy for the ML functionality. When this conditionalReconfiguration indication is provided. Step 6-7 : The NW (serving cell) configures and activates the use of Model ID Level 3 for the UE. The UE 110 applies the configurations received from the NW, the ML Functionality is activated, and the corresponding UE reports are generated (performance, monitoring, etc.). The new conditionalReconfiguration, can be alternatively, or additionally, be signaled by the NW (serving cell) to the UE, after Step 7, or as part of Step 7 (prior to any mobility event like in Step 8). This is similar to how Conditional Handover (CHO) is configured / managed in the prior-art, and the new conditionalReconfiguration can be either included in the CHO configuration message or provided as stand-alone configuration. FIG. 15 shows this alternative. In the example of FIG. 15: Step 7: conditionalReconfiguration, is additionally signaled by the NW (serving cell) to the UE. Step 8: The mobility event which lead to required change in the Model ID to be used, Scenario A in FIG. 13 or Scenario B in FIG. 14. This event does not necessarily have to be in RRC COONECTED mode, and any cell / gNB / RNA change can be handled by the subsequent steps i.e., when UE 110 moves RRC IDEL / INACTIVE in the source cell and then back to RRC CONNECTED in a new cell / gNB / RNA. In the examples of FIGs 13 and 14: Step 9: the NW (serving cell) can assume, configure and request the activation of the corresponding model ID in the UE 110 without any additional model identification in the new serving cell, because of the Step 5 and the hierarchical structure of the Model IDs. In the example of FIG. 13: Step 9: The NW (serving cell) assumes the UE 110 can use Model ID Level 2 as identified in Step 5.3. Step 10: The NW (serving cell) configures and activates the use of Model ID Level 2, based on information from Step 9, by providing an indicator to the UE The indicator can be the explicit Model ID (from the ones identified in Step 5) from higher level to be used by the UE The UE 110 must have a Model ID Level 2 available, because the identification process in Step 5.3 has yielded one. The current applicability of this model is assumed to be checked, if needed, according to prior-art solutions. However, if Model ID Level 2 is not determined to not be applicable in the current UE conditions, then the UE 110 needs to fall-back to the Model ID Level 1 (from Step 5.2). This fallback is then signaled to the NW (case not shown in FIG. 13). Alternatively, in this step a simple indicator can be provided instead of explicit model ID, indicating that UE 110 shall use / assume model ID with a higher hierarchy. E.g. the indicator can be a 2-bit word where the 4 Model ID levels from Step 5 are mapped to the available combinations, 00, 01,10 and 11. Similar approach ca be used if more than one Model ID is identified for each level. Step 11: UE 110 configures and activates Functionality using Model ID based on Step 10 configuration received from NW Step 12: the UE 110 can use the Level 2 Model ID (0_1_1_1) until, and only if, the NW decides to trigger the configuration and activation of a new Level 3 Model ID (see step 13); in this case the online model identification is required for the Level 3 Model ID). The use of the Level 2 Model ID (0_1_1_1) can be interpreted as the ‘fallback’ Mode ID configured for the UE 110 in Step 5.3. Step 13: Optional. NW decides to trigger the configuration and activation of a new Level 3 Model ID; in this case the online model identification is required for the Level 3 Model ID. The use of the Level 2 Model ID (0_1_1_1) can be interpreted as the ‘fallback’ Mode ID configured for the UE 110 in Step 5.3. The switching / selection between local models in Step 13.1 and model identification in Step 13.2 can be based on UE-side / NW-side additional conditions. In the example of FIG. 14: Step 9: The NW (serving cell) assumes the UE 110 can use Model ID Level 1 as identified in Step 5.2. Step 10 : The NW (serving cell) configures and activates the use of Model ID Level 1, based on information from Step 9. The indicator can be the explicit Model ID (from the ones identified in Step 5) from higher level to be used by the UE. The UE 110 must have a Model ID Level 1 available, because the identification process in Step 5.2 has yielded one. The current applicability of this model is assumed to be checked, if needed. However, if Model ID Level 1 is not determined to not be applicable in the current UE conditions, then the UE 110 needs to fall-back to the Model ID Level 0 (from Step 5.1). This fallback is then signaled to the NW (case not shown in FIG. 14). Alternatively, in this step a simple indicator can be provided instead of explicit model ID, indicating that UE 110 shall use / assume model ID with a higher hierarchy. E.g. the indicator can be a 2-bit word where the 4 Model ID levels from Step 5 are mapped to the available combinations, 00, 01,10 and 11. Similar approach ca be used if more than one Model ID is identified for each level. Step 11: UE 110 configures and activates Functionality using Model ID based on Step 10 configuration received from NW Step 12-13: the UE 110 can use the Level 1 Model ID (0_1_1) until the Step 13 when NW decides to trigger the configuration and activation of a new Level 2 Model ID (0_1_1_2); in this case the online model identification is required for the Level 1 Model ID. The use of the Level 1 Model ID (0_1_1) can be interpreted as the ‘fallback’ Mode ID configured for the UE 110 in Step 5.2. The switching / selection between local models in Step 13.1 and model identification in Step 13.2 can be based on UE-side / NW-side additional conditions. Step 14: Optional. NW decides to trigger the configuration and activation of a new Level 3 Model ID (0_1_1_2_1); in this case the online model identification is required for the Level 3 Model ID. The use of the Level 2 Model ID from Step 13 (0_1_1_2) can be interpreted as the ‘fallback’ Mode ID configured for the UE 110 in Step 13.2. In the example of FIG. 16: Step 9: UE 110 evaluates the conditionalReconfiguration received in any of the Step 5.* to select the Model ID level to use after the mobility event in Step 8. Step 10: Optionally, the new serving cell checks the conditionalReconfiguration from Step 5.*, to be aligned with the UE 110 selected Model ID level to use after the mobility event in Step 8. We give a more concrete example for configuring the Model ID Levels, using UE feature groups (FG), available after the UE capability exchange in steps 1-2. The FG is the one related to Beam Management, as example (FG_BM). Each feature group gets an ID. Considering only FGa, the Model IDs at different Levels could it be indicated as follows: 0_1: FG_BM1 0_1_*: FG_BM1_1, FG_BM1_2, ... 0_1_1_*: FG_BM1_1_1, FG_BM1_1_2, ... 0_1_1_1_*: FG_BM1_1_1_1, FG_BM1_1_1_2, ... Another way to assign these Model ID levels is based on their temporal validity, e.g. Level 0 with largest temporal validity, and Level 3 with lowest / shortest temporal validity. Time of the day can be used as a trigger for model ID switching. These triggers can vary between different parts of the network (think of commuting to / from work place), and they are likely to be based on other KPIs the network can estimate such as traffic load, number of mobility events, etc. Temporal validity refers to a time period for which the Model ID can be assumed to be valid, e.g. time of the day-based periods: 00:00-06:00, 00:06-12:00, 12:00-18:00, 18:00-24:00. In another embodiment, Level 0 Model IDs are identified via offline identification, i.e. are based on agreements between UE vendors and NW vendors. Level 1 is CSP / MNO specific and provide Model IDs which are linked to either UE capability categories and / or spatial and / or temporal characteristics of the network deployment. Level 2 is RA specific (see description above) Level 3 is RNA specific (see description above) Level 4 is gNB / cell specific (see description above) In an alternative to embodiment 2) or 3) the RA and RNA specific levels are combined, see example in FIG. 19. The Model IDs at the different levels (configured and enabled) do not have to be used simultaneously by the UE, as they are meant to be used when the UE 110 is in different RRC states and / or CM states. In another embodiment the levels could be configured such that the model IDs from all levels are valid simultaneously for the same UE e.g., the levels are related always to UE CONNECTED mode, at different geographical granularity. In yet another embodiment, the Model ID can be linked to Standalone non-public network (SNPN), operated by NPN operators, which, in general, combination of PLMN ID and NID and may not be globally unique across PLMN list. In another variant, within a PLMN, Model ID to be linked with restrictive cells using Closed Access Group (CAG) ID. In another variant, Model ID can be linked to network slicing or a slice group using NSAG ID within a tracking area. While using a unique global Model ID might be difficult to manage due to large scale of Model IDs as well (bit) size of the ID, the use of hierarchical IDs provides a more flexible Model ID management with variable ID size for different levels. The proposed idea of hierarchical Model ID is not limited to one particular definition of ‘level’ and applicable to various domains, such as area, temporal validity as well as any common functionality applicable to a group of UEs. The proposed approach provides an easy to configure and manage “fall-back” operating modes without additional model identification procedure required, i.e. when the a identified Model ID from Level X is not applicable (conditions are not meant for expected performance), then the identified Model ID Level X-1 is used. The above FIGs illustrate methods performed by a system comprising interaction between different system entities. The FIGs also illustrate a collection of separate methods performed separately by the different system entities. FIG. 17 illustrates an example of a controller 1800 suitable for use in an apparatus 1810. Implementation of a controller 1800 may be as controller circuitry. The controller 1800 may be implemented in hardware alone, have certain aspects in software including firmware alone or can be a combination of hardware and software (including firmware). As illustrated in FIG. 17 the controller 1800 may be implemented using instructions that enable hardware functionality, for example, by using executable instructions 1806 in a general-purpose or special-purpose processor 1802 that may be stored on a machine-readable storage medium (disk, memory etc.) to be executed by such a processor 1802. The processor 1802 is configured to read from and write to the memory 1804. The processor 1802 may also comprise an output interface via which data and / or commands are output by the processor 1802 and an input interface via which data and / or commands are input to the processor 1802. The memory 1804 stores instructions, program, or code 1806 that controls the operation of the apparatus 1810 when loaded into the processor 1802. The computer program instructions, program or code am 1806, provide the logic and routines that enables the apparatus 1810 to perform the methods illustrated in the accompanying FIGs. The processor 1802 by reading the memory 1804 is configured to load and execute the instructions, program, or code 1806. The apparatus 1810 comprises: at least one processor 1802; and at least one memory 1804 storing instructions that, when executed by the at least one processor 1802, cause the apparatus at least to perform the methods as described above. The apparatus 1810 comprises: at least one processor 1802; and at least one memory 1804 storing instructions that, when executed by the at least one processor 1802, cause the apparatus at least to: transmit, to the network node 120, a multi-level identifier capability; receive, from the network node 120, a model identification configuration; and use one of the one or more first machine learning supported functionalities of the UE 110 based on the received model identification configuration. The apparatus 1810 comprises: at least one processor 1802; and at least one memory 1804 storing instructions that, when executed by the at least one processor 1802, cause the apparatus at least to: receive, from the UE, a multi-level identifier capability; and transmit, to the UE, a model identification configuration. In some examples, there is a (computer implemented) system comprising: one or more UEs and one or more NW nodes. As illustrated in FIG. 18, the instructions, program, or code 1806 may arrive at the apparatus 1810 via any suitable delivery mechanism 1808. The delivery mechanism 1808 may be, for example, a machine readable medium, a computer-readable medium, a non-transitory computer-readable storage medium, a computer program product, a memory device, a record medium such as a Compact Disc Read-Only Memory (CD-ROM) or a Digital Versatile Disc (DVD) or a solid-state memory, an article of manufacture that comprises or tangibly embodies the computer program 1806. The delivery mechanism may be a signal configured to reliably transfer the computer program 1806. The apparatus 1810 may propagate or transmit the computer program 1806 as a computer data signal. The term “non-transitory” as used herein, is a limitation of the medium itself (i.e., tangible, not a signal) as opposed to a limitation on data storage persistency (e.g., RAM vs. ROM). Computer program instructions for causing an apparatus to perform at least the methods as described above or for performing at least the methods as described above. Computer program instructions for causing an apparatus to perform at least the following or for performing at least the following: causing transmission, to the network node 120, a multi-level identifier capability; causing reception, from the network node 120, a model identification configuration; and causing use of one of the one or more first machine learning supported functionalities of the UE 110 based on the received model identification configuration. Computer program instructions for causing an apparatus to perform at least the following or for performing at least the following: causing reception, from the UE, of a multi-level identifier capability; and causing transmission, to the UE, of a model identification configuration. The computer program instructions may be comprised in a computer program, a non-transitory computer readable medium, a computer program product, a machine readable medium. In some but not necessarily all examples, the computer program instructions may be distributed over more than one computer program. Although the memory 1804 is illustrated as a single component / circuitry it may be implemented as one or more separate components / circuitry some or all of which may be integrated / removable and / or may provide permanent / semi-permanent / dynamic / cached storage. Although the processor 1802 is illustrated as a single component / circuitry it may be implemented as one or more separate components / circuitry some or all of which may be integrated / removable. The processor 1802 may be a single core or multi-core processor. References to ‘computer-readable storage medium’, ‘computer program product’, ‘tangibly embodied computer program’ etc. or a ‘controller’, ‘computer’, ‘processor’ etc. should be understood to encompass not only computers having different architectures such as single / multi- processor architectures and sequential (Von Neumann) / parallel architectures but also specialized circuits such as field-programmable gate arrays (FPGA), application specific circuits (ASIC), signal processing devices and other processing circuitry. References to computer program, instructions, code etc. should be understood to encompass software for a programmable processor or firmware such as, for example, the programmable content of a hardware device whether instructions for a processor, or configuration settings for a fixed-function device, gate array or programmable logic device etc. As used in this application, the term ‘circuitry’ may refer to one or more or all the following: (a) hardware-only circuitry implementations (such as implementations in only analog and / or digital circuitry) and (b) combinations of hardware circuits and software, such as (as applicable): i. a combination of analog and / or digital hardware circuit(s) with software / firmware and ii. any portions of hardware processor(s) with software (including digital signal processor(s)), software, and memory or memories that work together to cause an apparatus, such as a mobile phone or server, to perform various functions and (c) hardware circuit(s) and or processor(s), such as a microprocessor(s) or a portion of a microprocessor(s), that requires software (for example, firmware) for operation, but the software may not be present when it is not needed for operation. This definition of circuitry applies to all uses of this term in this application, including in any claims. As a further example, as used in this application, the term circuitry also covers an implementation of merely a hardware circuit or processor and its (or their) accompanying software and / or firmware. The term circuitry also covers, for example and if applicable to the claim element, a baseband integrated circuit for a mobile device or a similar integrated circuit in a server, a cellular network device, or other computing or network device. The blocks illustrated in the accompanying Figs may represent steps in a method and / or sections of code in the computer program 1806. The illustration of a particular order to the blocks does not necessarily imply that there is a required or preferred order for the blocks and the order and arrangement of the block may be varied. Furthermore, it may be possible for some blocks to be omitted. As used here ‘module’ refers to a unit or apparatus that excludes certain parts / components that would be added by an end manufacturer or a user. The apparatus 1810 can, for example be a module. A controller 1800 of the apparatus 1810 can, for example be a module. Where a structural feature has been described, it may be replaced by means for performing one or more of the functions of the structural feature whether that function or those functions are explicitly or implicitly described. The systems, apparatus, methods, and computer programs may use machine learning which can include statistical learning. Machine learning is a field of computer science that gives computers the ability to learn without being explicitly programmed. The computer learns from experience E with respect to some class of tasks T and performance measure P if its performance at tasks in T, as measured by P, improves with experience E. The computer can often learn from prior training data to make predictions on future data. Machine learning includes wholly or partially supervised learning and wholly or partially unsupervised learning. It may enable discrete outputs (for example classification, clustering) and continuous outputs (for example regression). Machine learning may for example be implemented using different approaches such as cost function minimization, artificial neural networks, support vector machines and Bayesian networks for example. Cost function minimization may, for example, be used in linear and polynomial regression and K-means clustering. Artificial neural networks, for example with one or more hidden layers, model complex relationship between input vectors and output vectors. Support vector machines may be used for supervised learning. A Bayesian network is a directed acyclic graph that represents the conditional independence of a number of random variables. The above-described example methods are particularly adapted for the implementation in that the design is motivated by technical considerations of the internal functioning of the system or network. The examples are designed to exploit particular technical properties of the technical system on which they are implemented to bring about a technical effect such as efficient use of computer storage capacity, network bandwidth, power consumption. The methods also assign the execution of data-intensive training of a machinelearning algorithm to clients and preparatory steps to a server to take advantage of a server-client architecture. The training data and the training of the reduced machine learning model is technical in that there is distributed training across multiple clients and the training data at each client is secured and remains private. Technology involving the machine learning models can find application in many fields of technology. For example: - classification of digital images, videos, audio, or speech signals based on low-level features (e.g. edges or pixel attributes for images). - controlling a technical system or process, e.g. a computer-controlled classification system or industrial process - determining from measurements an adaptation to an industrial process; - digital audio, image or video enhancement or analysis, - separation of sources in speech signals; speech recognition, - encoding data for reliable and / or efficient transmission or storage (and corresponding decoding); compression of audio, image, video, or sensor data; - encrypting / decrypting or signing electronic communications; - determining a technical parameter (e.g. energy expenditure, core temperature) by processing data obtained from sensors; - providing a reliability estimate for technical information e.g. a genotype - providing a medical diagnosis by an automated system processing physiological measurements. -deriving or predicting a physical state of an existing real object from measurements of physical properties - causally linking sensor data provided as inputs to the ML model to control command outputs for controlling apparatus provided as outputs of the ML model. The above-described examples find application as enabling components of: automotive systems; telecommunication systems; electronic systems including consumer electronic products; distributed computing systems; media systems for generating or rendering media content including audio, visual and audio visual content and mixed, mediated, virtual and / or augmented reality; personal systems including personal health systems or personal fitness systems; navigation systems; user interfaces also known as human machine interfaces; networks including cellular, non-cellular, and optical networks; ad-hoc networks; the internet; the internet of things; virtualized networks; and related software and services. The apparatus can be provided in an electronic device, for example, a mobile terminal, according to an example of the present disclosure. It should be understood, however, that a mobile terminal is merely illustrative of an electronic device that would benefit from examples of implementations of the present disclosure and, therefore, should not be taken to limit the scope of the present disclosure to the same. While in certain implementation examples, the apparatus can be provided in a mobile terminal, other types of electronic devices, such as, but not limited to: mobile communication devices, hand portable electronic devices, wearable computing devices, portable digital assistants (PDAs), pagers, mobile computers, desktop computers, televisions, gaming devices, laptop computers, cameras, video recorders, GPS devices and other types of electronic systems, can readily employ examples of the present disclosure. Furthermore, devices can readily employ examples of the present disclosure regardless of their intent to provide mobility. The term ‘comprise’ is used in this document with an inclusive not an exclusive meaning. That is any reference to X comprising Y indicates that X may comprise only one Y or may comprise more than one Y. If it is intended to use ‘comprise’ with an exclusive meaning then it will be made clear in the context by referring to ‘comprising only one...’ or by using ‘consisting.’ In this description, the wording ‘connect’, ‘couple’ and ‘communication’ and their derivatives mean operationally connected / coupled / in communication. It should be appreciated that any number or combination of intervening components can exist (including no intervening components), i.e., to provide direct or indirect connection / coupling / communication. Any such intervening components can include hardware and / or software components. As used herein, the term "determine / determining" (and grammatical variants thereof) can include, not least: calculating, computing, processing, deriving, measuring, investigating, identifying, looking up (for example, looking up in a table, a database, or another data structure), ascertaining and the like. Also, "determining" can include receiving (for example, receiving information), accessing (for example, accessing data in a memory), obtaining and the like. Also," determine / determining" can include resolving, selecting, choosing, establishing, and the like. In this description, reference has been made to various examples. The description of features or functions in relation to an example indicates that those features or functions are present in that example. The use of the term ‘example’ or ‘for example’ or ‘can’ or ‘may’ in the text denotes, whether explicitly stated or not, that such features or functions are present in at least the described example, whether described as an example or not, and that they can be, but are not necessarily, present in some of or all other examples. Thus ‘example’, ‘for example’, ‘can’, or ‘may’ refers to a particular instance in a class of examples. A property of the instance can be a property of only that instance or a property of the class or a property of a sub-class of the class that includes some but not all the instances in the class. It is therefore implicitly disclosed that a feature described with reference to one example but not with reference to another example, can where possible be used in that other example as part of a working combination but does not necessarily have to be used in that other example. As used herein, “at least one of the following: ” and “at least one of ” and similar wording, where the list of two or more elements are joined by “and” or “or” mean at least any one of the elements, or at least any two or more of the elements, or at least all the elements. Although examples have been described in the preceding paragraphs with reference to various examples, it should be appreciated that modifications to the examples given can be made without departing from the scope of the claims. Features described in the preceding description may be used in combinations other than the combinations explicitly described above. Although functions have been described with reference to certain features, those functions may be performable by other features whether described or not. The description of a feature, such as an apparatus or a component of an apparatus, configured to perform a function, or for performing a function, should additionally be considered to also disclose a method of performing that function. For example, description of an apparatus configured to perform one or more actions, or for performing one or more actions, should additionally be considered to disclose a method of performing those one or more actions with or without the apparatus. Although features have been described with reference to certain examples, those features may also be present in other examples whether described or not. The term ‘a’, ‘an’ or ‘the’ is used in this document with an inclusive not an exclusive meaning. That is any reference to X comprising a / an / the Y indicates that X may comprise only one Y or may comprise more than one Y unless the context clearly indicates the contrary. If it is intended to use ‘a’, ‘an’ or ‘the’ with an exclusive meaning then it will be made clear in the context. In some circumstances the use of ‘at least one’ or ‘one or more’ may be used to emphasis an inclusive meaning but the absence of these terms should not be taken to infer any exclusive meaning. The presence of a feature (or combination of features) in a claim is a reference to that feature or (combination of features) itself and to features that achieve substantially the same technical effect (equivalent features). The equivalent features include, for example, features that are variants and achieve substantially the same result in substantially the same way. The equivalent features include, for example, features that perform substantially the same function, in substantially the same way to achieve substantially the same result. In this description, reference has been made to various examples using adjectives or adjectival phrases to describe characteristics of the examples. Such a description of a characteristic in relation to an example indicates that the characteristic is present in some examples exactly as described and is present in other examples substantially as described. The above description describes some examples of the present disclosure however those of ordinary skill in the art will be aware of possible alternative structures and method features which offer equivalent functionality to the specific examples of such structures and features described herein above and which for the sake of brevity and clarity have been omitted from the above description. Nonetheless, the above description should be read as implicitly including reference to such alternative structures and method features which provide equivalent functionality unless such alternative structures or method features are explicitly excluded in the above description of the examples of the present disclosure. Whilst endeavoring in the foregoing specification to draw attention to those features believed to be of importance the Applicant may seek protection via the claims in respect of any patentable feature or combination of features hereinbefore referred to and / or shown in the drawings whether or not emphasis has been placed thereon. l / we claim:

Claims

1. An apparatus comprising means for:transmitting, to a network node, a multi-level identifier capability,wherein the multi-level identifier capability indicates a capability of a user equipment, UE, to use a model identifier to identify one or more machine learning supported functionalities of the UE associated with one or more model identifier levels,wherein model identifier levels are associated with corresponding different validity domains for machine learning supported functionalities of the UE;receiving, from the network node, a model identification configuration comprising at least a first model identifier identifying one or more first machine learning supported functionalities of the UE and a second model identifier identifying one or more second machine learning supported functionalities of the UE,wherein the one or more first machine learning supported functionalities of the UE are associated with one or more first model identifier levels and the one or more second machine learning supported functionalities of the UE are associated with one or more second model identifier levels; andusing one of the one or more first machine learning supported functionalities of the UE based on the received model identification configuration.

2. An apparatus as claimed in claim 1, comprising means for:receiving, from the network node, a switching configuration configuring the UE with one or more first reconfiguration conditions for switching from using the first identifier identifying the one or more first machine learning supported functionalities of the UE to using the second identifier identifying the one or more second machine learning supported functionalities of the UE, wherein the second model identifier has a broader validity domain than the first model identifier; andbased on at least one of the one or more reconfiguration conditions being met, using one of the one or more second machine learning supported functionalities of the UE based on the second model identifier.

3. An apparatus as claimed in claim 2, comprising means for determining that one or more of the one or more first reconfiguration conditions has been met.

4. An apparatus as claimed in claim 2 or claim 3, wherein at least one of the reconfiguration conditions for switching from using the first identifier identifying the one or more first machine learning supported functionalities of the UE to using the second identifier identifying the one or more second machine learning functionalities of the UE is an invalidity event invalidating use of the first identifier of the one or more first machine learning supported functionalities of the UE.

5. An apparatus as claimed in any preceding claim, wherein the one or more first machine learning supported functionalities of the UE and the one or more second machine learning supported functionalities of the UE perform the same machine learning supported functionality6. An apparatus as claimed in any preceding claim, wherein a validity domain is related to an operating context of the UE.

7. An apparatus as claimed in any preceding claim, wherein the invalidity event is a change in operating context of the UE8. An apparatus as claimed in any preceding claim, wherein the validity domain is related to spatial and / or temporal characteristics of the UE and / or the network.

9. An apparatus as claimed in claim 8, wherein a spatial validity domain corresponds to one of: a beam; a cell; a network node; a RAN notification area (RNA); a Registration Area (RA); or a Public Land Mobile Network (PLMN) serving the UE.

10. An apparatus as claimed in claim 8 or 9, wherein the invalidity event is a mobility event of the UE.

11. An apparatus as claimed in any of claims 8-10, wherein a temporal validity domain corresponds to one of: a time of day; a day; a week; or a month.

12. An apparatus as claimed in any of claims 8-11, wherein the invalidity event is a temporal event related to a temporal change of the corresponding KPIs13. An apparatus as claimed in any of claims 8-12, wherein the validity domain is related to a compatibility of its associated one or more model identifiers with at least one of: the network; a UE vendor; a mobile network operator.

14. An apparatus as claimed in any of claims 8-12, wherein the invalidity event causes the UE to be served by a new cell under the same network node or a new cell under a new network node.

15. An apparatus as claimed in any preceding claim, wherein the model identification configuration indicates a hierarchy of identifier levels associated with corresponding different validity domains within a hierarchy of validity domains16. An apparatus as claimed in any preceding claim, comprising means for managing implementation of the first ML model and managing validity of the first ML model using associated validity domain.

17. An apparatus as claimed in any preceding claim, wherein identifiers having the same validity domain have the same format18. An apparatus as claimed in any preceding claim, wherein identifiers having a narrower validity domain have a shorter format than identifiers having a broader validity domain.

19. An apparatus as claimed in any preceding claim, wherein the one or more machine learning supported functionalities of the UE are associated with one or more validity domains, wherein the validity domains of the one or more machine learning supported functionalities of the UE are related to the validity domains of the one or more model identifier levels.

20. An apparatus as claimed in any preceding claim, wherein a model identifier comprises at least:77 an indication of a model identifier level with which one or more machine learning supported functionalities of the UE identified by the model identifier is associated; andan indication of an association of the model identifier with one or more operating contexts of the UE.

21. An apparatus as claimed in any preceding claim, the model identification configuration comprises a third model identifier identifying one or more third machine learning supported functionalities of the UE, andthe switching configuration configures the UE with one or more second reconfiguration conditions for switching from using the first identifier identifying the one or more first machine learning supported functionalities of the UE or the second identifier identifying the one or more second machine learning functionalities of the UE to using the third identifier identifying the one or more third machine learning functionalities of the UE,wherein the third model identifier has a broader validity domain than the first model identifier and the second model identifier, and wherein the UE comprises means for:based on one or more of the one or more second reconfiguration conditions being met, using one of the one or more third machine learning supported functionalities of the UE based on the third model identifier.

22. An apparatus as claimed in any preceding claim, comprising means for receiving, from the network node, an indication to use a different model identifier of one or more different machine learning supported functionalities of the UE, wherein the different identifier has a narrower validity domain than an identifier of a current machine learning supported functionality of the UE; and using one of the one or more different machine learning supported functionalities of the UE based on the different model identifier.

23. An apparatus as claimed in claim 22, comprising means for, in dependence upon receiving the indication to use the different model identifier of the one or more different machine learning supported functionalities of the UE, determining if the UEhas a model identifier with a narrower validity domain than the model identifier of a current machine learning supported functionality of the UE; andin dependence upon a determination that the UE does not have a model ID with a narrower validity domain than the identifier of the current machine learning supported functionality of the UE, continuing to use the current machine learning supported functionality of the UE.

24. An apparatus comprising means for:receiving, from a user equipment, UE, a multi-level identifier capability,wherein the multi-level identifier capability indicates a capability of the UE to use a model identifier to identify one or more machine learning supported functionalities of the UE associated with one or more model identifier levels,wherein model identifier levels are associated with corresponding different validity domains for machine learning supported functionalities of the UE; determining, based on the received multi-level identifier capability, a model identification configuration comprising at least a first model identifier identifying one or more first machine learning supported functionalities of the UE and a second model identifier identifying one or more second machine learning supported functionalities of the UE,wherein the one or more first machine learning supported functionalities of the UE are associated with one or more first model identifier levels and the one or more second machine learning supported functionalities of the UE are associated with one or more second model identifier levels; and transmitting the model identification configuration to the UE.

25. An apparatus as claimed in claim 24 comprising means for:transmitting, to the UE, a switching configuration configuring the UE with one or more first reconfiguration conditions for switching from using the first identifier identifying the one or more first machine learning supported functionalities of the UE to using the second identifier identifying the one or more second machine learning supported functionality of the UE, wherein the second model identifier has a broader validity domain than the first model identifier.A

Images