Artificial intelligence (AI) based channel state information (CSI) coding

1 . A method performed by a user equipment (UE) for artificial intelligence (AI) based channel state information (CSI) encoding, comprising:

receiving, a signaling message including at least one encoding parameter from a base station (BS), wherein the at least one encoding parameter comprises an AI model indicator (AMI);

segmenting a precoder data based on the signaling message into a plurality of data segments;

selecting at least one AI model from a plurality of AI models based on the at least one encoding parameter received from the BS;

encoding each data segment of the plurality of data segments of the precoder data using the at least one selected AI model; and

sending at least one encoded data segment of the precoder data to the BS for decoding.

2 . The method of claim 1 , wherein the at least one AI model is selected among the plurality of AI models based on the AMI in response to the at least one encoding parameter including the AMI.

3 . The method of claim 1 , wherein the at least one encoding parameter further includes at least one of time measurements or frequency sub-band measurements.

4 . The method of claim 1 , wherein the precoder data is encoded based on at least one of time measurement occasions, at least one frequency sub-band measurement, at least one time measurement in which each occasion comprises of CSI measured for at least one frequency sub-band, or the AMI for selecting at least one AI model.

5 . The method of claim 1 , wherein the at least one encoded data segment of the precoder data is decoded by the BS based on at least one of at least one frequency sub-band measurement, at least one time measurement in which each occasion comprises of CSI measured for at least one frequency sub-band, or the AMI for selecting at least one AI model.

6 . A user equipment (UE) for artificial intelligent (AI) based channel state information (CSI) encoding, the UE comprising:

memory comprising a plurality of AI models and corresponding weight of each of the plurality of AI models;

a processor; and

wherein the processor is configured to:

receive a signaling message including at least one encoding parameter from a base station (BS), wherein the at least one encoding parameter comprises an AI model indicator (AMI);

segment a precoder data into a plurality of data segments;

select at least one AI model from a plurality of AI models based on the at least one encoding parameter received from the BS;

encode each data segment of the plurality of data segments of the precoder data using the at least one selected AI model; and

send the at least one encoded data segments of the precoder data to the BS for decoding.

7 . The UE of claim 6 , wherein the at least one AI model is selected among the plurality of AI models based on the AMI in response to the at least one encoding parameter including the AMI.

8 . The UE of claim 6 , wherein the at least one encoding parameter further includes at least one of time measurements or frequency sub-band measurements.

9 . The UE of claim 6 , wherein the precoder data is encoded based on at least one of time measurement occasions, at least one frequency sub-band measurement, at least one time measurement in which each occasion comprises of CSI measured for at least one frequency sub-band, or the AMI for selecting at least one AI model.

10 . The UE of claim 6 , wherein the at least one encoded data segment of the precoder data is decoded by the BS based on at least one of at least one frequency sub-band measurement, at least one time measurement in which each occasion comprises of CSI measured for at least one frequency sub-band, or the AMI for selecting at least one AI model.

11 . A base station (BS) for artificial intelligent (AI) based channel state information (CSI), the BS comprising:

a memory;

a processor; and

wherein the processor is configured to:

send a signaling message including at least one encoding parameter to a user equipment (UE), wherein the at least one encoding parameter comprises an AI model indicator (AMI);

receive at least one data segment of a precoder data from the UE;

select at least one AI model from a plurality of AI models based on the at least one encoding parameter; and

decode the at least one data segments of the precoder data using the at least one selected AI model.

12 . The BS of claim 11 , wherein the at least one AI model is selected among the plurality of AI models based on the AMI.

13 . The BS of claim 11 , wherein the at least one encoding parameter further includes at least one of time measurements or frequency sub-band measurements.

14 . The BS of claim 11 , wherein the precoder data is encoded by the UE based on at least one of time measurement occasions, at least one frequency sub-band measurement, at least one time measurement in which each occasion comprises of CSI measured for at least one frequency sub-band, or the AMI for selecting at least one AI model.

15 . The BS of claim 11 , wherein the at least one encoded data segment of the precoder data is decoded based on at least one of at least one frequency sub-band measurement, at least one time measurement in which each occasion comprises of CSI measured for at least one frequency sub-band, or the AMI for selecting at least one AI model.