UE capability reporting and handling of artificial intelligence/machine learning models
A user device, UE, for a wireless communication network, like a 3 rd Generation Partnership Project, 3GPP, network, is described. The UE has a plurality of UE capabilities, the UE capabilities indicating functions the UE is capable to perform. The UE is to inform on one or more parts of the plurality of UE capabilities using one or more identification values, e.g., hash values.
1 . A user device, UE, for a wireless communication network, like a 3 rd Generation Partnership Project, 3GPP, network,
wherein the UE comprises a plurality of UE capabilities, the UE capabilities indicating functions the UE is capable to perform, and
wherein the UE is to inform on one or more parts of the plurality of UE capabilities using one or more identification values, e.g., hash values,
wherein the UE is capable to use one or more artificial intelligence/machine learning, AI/ML, models, and the plurality of UE capabilities comprises information on the one or more AI/ML models,
wherein the UE is to store additional information for identifying and/or executing a certain AI/ML model, the additional information including a validity of the certain AI/ML model and a cell ID linked to the certain AI/ML model, and
wherein the validity of an AI/ML model is associated with a signaling of AI/ML model update or capability information, and the validity of the certain AI/ML model is determined when the UE is associated with the cell ID, and/or the certain AI/ML model is valid when the UE is associated with the cell ID.
2 . The user device, UE, of claim 1 , wherein:
each of the one or more AI/ML models is identified by an individual identification value, or
two or more of the AI/ML models are identified by a common identification value.
3 . The user device, UE, of claim 1 , wherein:
the UE is to receive a request, like a UE capability enquiry message, from the wireless communication network or from a further UE to report some or all of the AI/ML models supported by the UE, and/or
responsive to the request, the UE is to send the one or more identification values representing the supported AI/ML models to the wireless communication network or to the further UE.
4 . The user device, UE, of claim 3 , wherein:
the UE is to receive from the wireless communication network or from a further UE a request indicating one or more identification values or a common identification value representing AI/ML models, and/or
responsive to the request, the UE is to:
deactivate the one or more AI/ML models indicated by the one or more identification values or the common identification value, or
reset the one or more AI/ML models indicated by the one or more identification values or the common identification value into a predefined state.
5 . The user device, UE, of claim 1 , wherein:
the UE is to receive from the wireless communication network or from a further UE one or more additional AI/ML models to be supported by the UE, and/or
store the one or more additional AI/ML models.
6 . The user device, UE, of claim 1 , wherein:
the UE is to receive from the wireless communication network or from a further UE a message, like a Master Information Block, MIB, or a System Information Block, SIB, indicating one or more identification values or a common identification value representing AI/ML models, and/or
responsive to the message, the UE is to:
use the one or more AI/ML models indicated by the one or more identification values or the common identification value as default AI/ML model, or
use among the one or more AI/ML models indicated by the one or more identification values or the common identification value a first AI/ML model supported by the UE as default AI/ML model, or
bar the one or more AI/ML models indicated by the one or more identification values or the common identification value from being used by the UE.
7 . The user device, UE, of claim 1 , wherein the UE is to include within barring information one or more:
the one or more identification values of the one or more AI/ML models it has barred,
a barring time, e.g., how long the model is barred,
a trigger condition that has to be fulfilled to unbar the AI/ML model.
8 . The user device, UE, of claim 7 , wherein the UE is to signal the barring information to the wireless communication network or to a further UE.
9 . The user device, UE, of claim 1 , wherein an AI/ML model is identified by a non-unique identification value used at different origins of the AI/ML model, like different AI/ML servers or different wireless communication networks, and wherein the UE is to link the AI/ML model with the origin of the AI/ML model.
10 . The user device, UE, of claim 1 , wherein the UE is to associate each of the stored AI/ML models with an identification value, like a hash value, and wherein the UE is to re-activate an AI/ML model after the last use using the identification value.
11 . The user device, UE, of claim 1 , wherein
an AI/ML model is updated partly or in full, and/or
the UE is to pull an AI/ML model update, and/or
an AI/ML model update is pushed to the UE, and/or
an AI/ML model is updated in case a difference is detected between the AI/ML model stored at the UE and the AI/ML model stored at an origin of the AI/ML model, and/or
an AI/ML model update comprises an update of one or more validity conditions of the AI/ML model, e.g., a change in the validity time, or a change of a geographical area in which the UE is located, or a use of a different set of resources, like a different resource pool.
12 . The user device, UE, of claim 1 , wherein:
each AI/ML model is identified by an individual identification value, or
two or more of the AI/ML models are identified by a common identification value.
13 . The user device, UE, of claim 1 , wherein each of the AI/ML models is identified by the individual identification value, e.g., an individual hash value, and the common identification value identifying two or more of the AI/ML models is acquired using the individual identification values, e.g., by hashing the individual hash values.
14 . The user device, UE, of claim 1 , wherein:
the UE is to store the AI/ML models using a hierarchical structure defining a plurality of hierarchy levels, e.g., for defining for each feature, a certain scenario and a certain PLMN, and/or
an AI/ML model is identified using the identification value of the AI/ML model and the hierarchy level at which the AI/ML model is stored.
15 . A network entity for a wireless communication network, like a 3 rd Generation Partnership Project, 3GPP, network,
wherein the network entity is to receive from a user device, UE, information on the UE capabilities, the UE capabilities indicating functions the UE is capable to perform,
wherein the information comprises one or more identification values, like hash values, identifying one or more parts of the plurality of UE capabilities, and
wherein the network entity is to determine UE capabilities using the received one or more identification values,
wherein the UE is capable to use one or more artificial intelligence/machine learning, AI/ML, models, and the plurality of UE capabilities comprises information on the one or more AI/ML models,
wherein the UE is to store additional information for identifying and/or executing a certain AI/ML model, the additional information including a validity of the certain AI/ML model and a cell ID linked to the certain AI/ML model, and
wherein the validity of an AI/ML model is associated with a signaling of AI/ML model update or capability information, and the validity of the certain AI/ML model is determined when the UE is associated with the cell ID, and/or the certain AI/ML model is valid when the UE is associated with the cell ID.
16 . A method for operating a network entity for a wireless communication network, like a 3 rd Generation Partnership Project, 3GPP, network, the method comprising:
receiving from a user device, UE, information on the UE capabilities, the UE capabilities indicating functions the UE is capable to perform, wherein the information comprises one or more identification values, like hash values, identifying one or more parts of the plurality of UE capabilities, and
determining UE capabilities using the received one or more identification values,
wherein the UE is capable to use one or more artificial intelligence/machine learning, AI/ML, models, and the plurality of UE capabilities comprises information on the one or more AI/ML models,
wherein the UE stores additional information for identifying and/or executing a certain AI/ML model, the additional information including a validity of the certain AI/ML model and a cell ID linked to the certain AI/ML model, and
wherein the validity of an AI/ML model is associated with a signaling of AI/ML model update or capability information, and the validity of the certain AI/ML model is determined when the UE is associated with the cell ID, and/or the certain AI/ML model is valid when the UE is associated with the cell ID.