Granted Patent B2

US 12,651,142 · App. 17/484,580 · Granted Jun 9, 2026

Inventors: Xipeng Zhu (San Diego, CA); Gavin Bernard Horn (La Jolla, CA); Rajat Prakash (San Diego, CA); Rajeev Kumar (San Diego, CA); Shankar Krishnan (San Diego, CA); Taesang Yoo (San Diego, CA); Eren Balevi (San Diego, CA); Aziz Gholmieh (Del Mar, CA)

Assignee: QUALCOMM Incorporated

G06N3/04H04W24/02

Abstract

Claims (50)

1 . A method of wireless communication by a base station, comprising:

receiving a user equipment (UE) radio capability and a UE machine learning capability;

determining a neural network function (NNF) identifier (ID) based on the UE radio capability;

selecting a neural network model from a plurality of neural network models based on the NNF ID, the UE machine learning capability, and a capability of a network entity, the neural network model comprising a model structure and a parameter set;

configuring the network entity with the neural network model;

transmitting, from a centralized unit control plane (CU-CP) to a centralized unit machine learning plane (CU-XP), an NNF request including the NNF identifier (ID);

transmitting a model setup request from the CU-XP to the network entity, the model setup request including a model ID corresponding to the model structure and a parameter set ID corresponding to the parameter set;

transmitting the neural network model to the network entity based on the parameter set ID and the model ID; and

receiving a model setup response from the network entity confirming the neural network model is configured in the network entity.

2 . The method of claim 1 , in which the network entity comprises one or more units of the base station, including a distributed unit (DU), the centralized unit control plane (CU-CP), a centralized unit user plane (CU-UP), or the centralized unit machine learning plane (CU-XP).

3 . The method of claim 1 , in which the network entity comprises another network device including a radio access network intelligent controller (RIC).

4 . The method of claim 1 , further comprising transmitting the neural network model to the network entity via a user plane protocol.

5 . The method of claim 1 , further comprising activating the neural network model.

6 . The method of claim 5 , in which the activating comprises transmitting a network model activation message from the centralized unit control plane (CU-CP) to the centralized unit machine learning plane (CU-XP), and transmitting the network model activation message from the CU-XP to the network entity.

7 . The method of claim 5 , further comprising activating a UE model for operation at a UE by transmitting a UE model activation message to a distributed unit (DU) for forwarding to the UE.

8 . The method of claim 1 , further comprising receiving a UE message from a UE triggering the configuring of the network entity.

9 . The method of claim 8 , in which the UE message comprises a UE assistance information message comprising the NNF ID or a model request.

10 . The method of claim 1 , further comprising:

transmitting, to a UE, a prohibit timer for a UE machine learning request, in response to receiving the UE radio capability and the UE machine learning capability;

transmitting, to the UE, a white list of allowed NNF IDs and neural network models based on the UE radio capability and the UE machine learning capability;

transmitting, to the UE, a black list of prohibited NNF IDs and neural network models based on the UE radio capability and the UE machine learning capability; and

receiving, from the UE, a request for configuring the network entity with the neural network model.

11 . The method of claim 10 , further comprising transmitting, to the UE, a list of events triggering the UE to report a change in conditions for initiating an update of the neural network model.

12 . The method of claim 10 , in which the request for configuring the network entity comprises a UE assistance information message.

13 . An apparatus for wireless communication by a base station, comprising:

a memory; and

at least one processor coupled to the memory, the at least one processor configured:

to receive a user equipment (UE) radio capability and a UE machine learning capability;

to determine a neural network function (NNF) identifier (ID) based on the UE radio capability;

to select a neural network model from a plurality of neural network models, based on the NNF ID, the UE machine learning capability, and a capability of a network entity, the neural network model comprising a model structure and a parameter set;

to configure the network entity with the neural network model;

to transmit, to the UE, a white list of allowed NNF IDs and neural network models based on the UE radio capability and the UE machine learning capability;

to transmit, to the UE, a black list of prohibited NNF IDs and neural network models based on the UE radio capability and the UE machine learning capability; and

to receive, from the UE, a request for configuring the network entity with the neural network model.

14 . The apparatus of claim 13 , in which the network entity comprises one or more units of the base station, including a distributed unit (DU), a centralized unit control plane (CU-CP), a centralized unit user plane (CU-UP), or a centralized unit machine learning plane (CU-XP).

15 . The apparatus of claim 13 , in which the network entity comprises another network device including a radio access network intelligent controller (RIC).

16 . The apparatus of claim 13 , in which the at least one processor is further configured:

to transmit, from a centralized unit control plane (CU-CP) to a centralized unit machine learning plane (CU-XP), an NNF request including the NNF identifier (ID);

to transmit a model setup request from the CU-XP to the network entity, the model setup request including a model ID corresponding to the model structure and a parameter set ID corresponding to the parameter set;

to transmit the neural network model to the network entity based on the parameter set ID and the model ID; and

to receive a model setup response from the network entity confirming the neural network model is configured in the network entity.

17 . The apparatus of claim 16 , in which the at least one processor is further configured to transmit the neural network model to the network entity via a user plane protocol.

18 . The apparatus of claim 13 , in which the at least one processor is further configured to activate the neural network model.

19 . The apparatus of claim 18 , in which the at least one processor is configured to activate by transmitting a network model activation message from a centralized unit control plane (CU-CP) to a centralized unit machine learning plane (CU-XP), and transmitting the network model activation message from the CU-XP to the network entity.

20 . The apparatus of claim 18 , in which the at least one processor is further configured to activate a UE model for operation at a UE by transmitting a UE model activation message to a distributed unit (DU) for forwarding to the UE.

21 . The apparatus of claim 13 , in which the at least one processor is further configured to receive a UE message from a UE triggering the configuring of the network entity.

22 . The apparatus of claim 21 , in which the UE message comprises a UE assistance information message comprising the NNF ID or a model request.

23 . The apparatus of claim 13 , in which the at least one processor is further configured to transmit, to the UE, a list of events triggering the UE to report a change in conditions for initiating an update of the neural network model.

24 . The apparatus of claim 13 , in which the request for configuring the network entity comprises a UE assistance information message.

25 . The apparatus of claim 13 , further comprising receiving a request for configuration based on a prohibit timer for a UE machine learning request.

Assignments (1)

Continuity (1)

References Cited (11)