AI-based channel estimation method and apparatus

1 . An artificial intelligence (AI)-based channel estimation method, comprising:

obtaining a first training data set, wherein the first training data set is a plurality of initial channel-estimation results obtained by performing channel estimation multiple times on a first channel;

inputting the first training data set into a first initial estimation model for processing, to obtain predicted channel-estimation results;

inputting the predicted channel-estimation results and nominal parameters of the first channel into a first target function, and calculating a first loss value according to the first target function;

adjusting a parameter of the first initial estimation model to obtain a trained first estimation model, when the first loss value is greater than a first preset loss value;

determining the trained first estimation model as a first AI estimation model, when the first loss value is less than or equal to the first preset loss value;

obtaining a first reference signal, the first reference signal being used for channel estimation of the first channel; and

performing channel estimation on the first channel based on the first reference signal, to obtain an initial channel-estimation result; and

inputting the initial channel-estimation result into the first AI estimation model for optimization, to obtain a channel-estimation result of the first channel.

2 . The method of claim 1 , further comprising:

processing the channel-estimation result of the first channel with an AI-based recovery model, to obtain a channel-recovery result, wherein the channel-recovery result indicates a channel-estimation result of a full channel, and the first channel is a subset of the full channel.

3 . The method of claim 2 , wherein:

before processing the channel-estimation result of the first channel with the AI-based recovery model, the method further comprises:

performing channel recovery based on the channel-estimation result of the first channel to obtain an initial channel-recovery result of the first channel; and

the AI-based recovery model is a first AI recovery model, and processing the channel-estimation result of the first channel with the AI-based recovery model comprises:

inputting the initial channel-recovery result into the first AI recovery model for optimization, to obtain the channel-recovery result.

4 . The method of claim 3 , further comprising:

obtaining a third training data set, wherein the third training data set is a plurality of initial channel-recovery results obtained by performing channel recovery according to the channel-estimation result of the first channel;

inputting the third training data set into a first initial recovery model for processing, to obtain predicted channel-recovery results;

inputting the predicted channel-recovery results and nominal parameters of the full channel into a third target function, and calculating a third loss value according to the third target function;

adjusting a parameter of the first initial recovery model to obtain a trained first recovery model, when the third loss value is greater than a third preset loss value; and

determining the trained first recovery model as the first AI recovery model, when the third loss value is less than or equal to the third preset loss value.

5 . The method of claim 2 , wherein the AI-based recovery model is a second AI recovery model, and processing the channel-estimation result of the first channel with the AI-based recovery model comprises:

inputting the channel-estimation result of the first channel into the second AI recovery model for AI-based channel recovery, to obtain the channel-recovery result.

6 . The method of claim 5 , further comprising:

obtaining a fourth training data set, wherein the fourth training data set is a plurality of channel-estimation results of the first channel;

inputting the fourth training data set into a second initial recovery model for processing, to obtain predicted channel-recovery results;

inputting the predicted channel-recovery results and nominal parameters of the full channel into a fourth target function, and calculating a fourth loss value according to the fourth target function;

adjusting a parameter of the second initial recovery model to obtain a trained second recovery model, when the fourth loss value is greater than a fourth preset loss value; and

determining the trained second recovery model as the second AI recovery model, when the fourth loss value is less than or equal to the fourth preset loss value.

7 . An electronic device, comprising:

a transceiver;

a memory storing one or more programs;

a processor configured to executed the one or more programs to:

obtain a first training data set, wherein the first training data set is a plurality of initial channel-estimation results obtained by performing channel estimation multiple times on a first channel;

input the first training data set into a first initial estimation model for processing, to obtain predicted channel-estimation results;

input the predicted channel-estimation results and nominal parameters of the first channel into a first target function, and calculate a first loss value according to the first target function;

adjust a parameter of the first initial estimation model to obtain a trained first estimation model, when the first loss value is greater than a first preset loss value; and

determine the trained first estimation model as a first AI estimation model, when the first loss value is less than or equal to the first preset loss value;

cause the transceiver to obtain a first reference signal, the first reference signal being used for channel estimation of the first channel; and

perform channel estimation on the first channel based on the first reference signal, to obtain an initial channel-estimation result;

input the initial channel-estimation result into the first AI estimation model for optimization, to obtain a channel-estimation result of the first channel.

8 . The electronic device of claim 7 , wherein the processor is further configured to:

process the channel-estimation result of the first channel with an AI-based recovery model, to obtain a channel-recovery result, wherein the channel-recovery result indicates a channel-estimation result of a full channel, and the first channel is a subset of the full channel.

9 . The electronic device of claim 8 , wherein the processor is further configured to:

perform channel recovery based on the channel-estimation result of the first channel to obtain an initial channel-recovery result of the first channel; and

input the initial channel-recovery result into a first AI recovery model for optimization, to obtain the channel-recovery result.

10 . The electronic device of claim 9 , wherein the processor is further configured to:

obtain a third training data set, wherein the third training data set is a plurality of initial channel-recovery results obtained by performing channel recovery according to the channel-estimation result of the first channel;

input the third training data set into a first initial recovery model for processing, to obtain predicted channel-recovery results;

input the predicted channel-recovery results and nominal parameters of the full channel into a third target function, and calculate a third loss value according to the third target function;

adjust a parameter of the first initial recovery model to obtain a trained first recovery model, when the third loss value is greater than a third preset loss value; and

determine the trained first recovery model as the first AI recovery model, when the third loss value is less than or equal to the third preset loss value.

11 . The electronic device of claim 8 , wherein the processor is specifically configured to:

input the channel-estimation result of the first channel into a second AI recovery model for AI-based channel recovery, to obtain the channel-recovery result.

12 . The electronic device of claim 11 , wherein processing unit is further configure to:

obtain a fourth training data set, wherein the fourth training data set is a plurality of channel-estimation results of the first channel;

input the fourth training data set into a second initial recovery model for processing, to obtain predicted channel-recovery results;

input the predicted channel-recovery results and nominal parameters of the full channel into a fourth target function, and calculate a fourth loss value according to the fourth target function;

adjust a parameter of the second initial recovery model to obtain a trained second recovery model, when the fourth loss value is greater than a fourth preset loss value; and

determine the trained second recovery model as the second AI recovery model, when the fourth loss value is less than or equal to the fourth preset loss value.

13 . A non-transitory computer-readable storage medium having stored thereon instructions, which when executed by at least one processor, cause the at least one processor to perform operations comprising:

obtaining a first training data set, wherein the first training data set is a plurality of initial channel-estimation results obtained by performing channel estimation multiple times on a first channel;

inputting the first training data set into a first initial estimation model for processing, to obtain predicted channel-estimation results;

inputting the predicted channel-estimation results and nominal parameters of the first channel into a first target function, and calculating a first loss value according to the first target function;

adjusting a parameter of the first initial estimation model to obtain a trained first estimation model, when the first loss value is greater than a first preset loss value; and

determining the trained first estimation model as a first AI estimation model, when the first loss value is less than or equal to the first preset loss value;

obtaining a first reference signal, the first reference signal being used for channel estimation of the first channel; and

performing channel estimation on the first channel based on the first reference signal, to obtain an initial channel-estimation result; and

inputting the initial channel-estimation result into the first AI estimation model for optimization, to obtain a channel-estimation result of the first channel.

14 . The non-transitory computer-readable storage medium of claim 13 , wherein the at least one processor is further configured to perform:

processing the channel-estimation result of the first channel with an AI-based recovery model, to obtain a channel-recovery result, wherein the channel-recovery result indicates a channel-estimation result of a full channel, and the first channel is a subset of the full channel.

15 . The non-transitory computer-readable storage medium of claim 14 , wherein:

before processing the channel-estimation result of the first channel with the AI-based recovery model, the at least one processor is further configured to perform:

performing channel recovery based on the channel-estimation result of the first channel to obtain an initial channel-recovery result of the first channel; and

the AI-based recovery model is a first AI recovery model, and processing the channel-estimation result of the first channel with the AI-based recovery model comprises:

inputting the initial channel-recovery result into the first AI recovery model for optimization, to obtain the channel-recovery result.

16 . The non-transitory computer-readable storage medium of claim 15 , wherein the at least one processor is further configured to perform:

obtaining a third training data set, wherein the third training data set is a plurality of initial channel-recovery results obtained by performing channel recovery according to the channel-estimation result of the first channel;

inputting the third training data set into a first initial recovery model for processing, to obtain predicted channel-recovery results;

inputting the predicted channel-recovery results and nominal parameters of the full channel into a third target function, and calculating a third loss value according to the third target function;

adjusting a parameter of the first initial recovery model to obtain a trained first recovery model, when the third loss value is greater than a third preset loss value; and

determining the trained first recovery model as the first AI recovery model, when the third loss value is less than or equal to the third preset loss value.

17 . The non-transitory computer-readable storage medium of claim 14 , wherein the AI-based recovery model is a second AI recovery model, and in terms of processing the channel-estimation result of the first channel with the AI-based recovery model, the at least one processor is configured to perform:

inputting the channel-estimation result of the first channel into the second AI recovery model for AI-based channel recovery, to obtain the channel-recovery result.