CSI compression and decompression

1 . An apparatus comprising:

at least one processor; and

at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to:

generate channel state information for a channel between the apparatus and a network device;

generate compressed channel state information based on the channel state information by using a compression model; and

transmit the compressed channel state information over a radio link to the network device by:

quantizing the compressed channel state information into a lower dimensional vector of compressed binary elements comprising bit values;

quantizing an identification of the compression model into bit values;

appending the quantized identification to the quantized compressed channel state information; and

transmitting, to the network device, the quantized compressed channel state information and the quantized identification of the compression model.

2 . The apparatus of claim 1 , wherein a decompression model at the network device, for reconstructing the channel state information is common to a plurality of compression models comprising the compression model.

3 . The apparatus of claim 2 , wherein an end-to-end model comprising the compression model and a further decompression model is trained with a training dataset, and the decompression model is trained with a combination of:

an indicator corresponding to the compression model, and

at least a portion of the training dataset.

4 . The apparatus of claim 3 , wherein the identification of the compression model is based on a type of the compression model and a type of the apparatus.

5 . The apparatus of claim 4 , wherein the compression model is a machine learning model.

6 . The apparatus of claim 5 , wherein the apparatus is a user equipment.

7 . A system comprising:

a first apparatus comprising:

a first processor; and

a first memory storing first instructions that, when executed by the first processor, cause the first apparatus at least to:

generate channel state information for a channel between the first apparatus and a network device;

generate compressed channel state information based on the channel state information by using a compression model; and

transmit the compressed channel state information over a radio link to the network device by:

quantizing the compressed channel state information into a lower dimensional vector of compressed binary elements comprising bit values;

quantizing an identification of the compression model into bit values;

appending the quantized identification to the quantized compressed channel state information; and

transmitting, to the network device, the quantized compressed channel state information and the quantized identification of the compression model; and

a second apparatus comprising:

a second processor; and

a second memory storing second instructions that, when executed by the second processor, cause the second apparatus at least to:

receive, from the first apparatus, the quantized compressed channel state information and the quantized identification of the compression model for generating the compressed channel state information;

determine an indicator corresponding to the compression model based on the identification; and

reconstruct, by using a decompression model, channel state information for the channel based on the compressed channel state information and the indicator.

8 . The system of claim 7 , wherein the decompression model, for reconstructing the channel state information, is common to a plurality of compression models comprising the compression model.

9 . The system of claim 8 , wherein an end-to-end model comprising the compression model and a further decompression model is trained with a training dataset, and the decompression model is trained with a combination of:

the indicator corresponding to the compression model, and

at least a portion of the training dataset.

10 . The system of claim 8 , wherein the identification of the compression model is based on a type of the compression model and a type of the first apparatus.

11 . The system of claim 8 , wherein the decompression model is trained with a combination of:

an indicator corresponding to each of the plurality of compression models, and

at least a portion of a training dataset used to train the corresponding compression model.

12 . The system of claim 11 , wherein the second instructions, when executed by the second processor, further cause the second apparatus to:

de-quantize the bit values into a set of values; and

reconstruct, by using the decompression model, the channel state information based on the set of values and the indicator.

13 . A method comprising:

generating, at a user equipment, channel state information for a channel between the user equipment and a network device;

generating, at the user equipment, compressed channel state information based on the channel state information by using a compression model; and

transmitting, at the user equipment, the compressed channel state information over a radio link to the network device by:

quantizing the compressed channel state information into a lower dimensional vector of compressed binary elements comprising bit values;

quantizing an identification of the compression model into bit values;

appending the quantized identification to the quantized compressed channel state information; and

transmitting, to the network device, the quantized compressed channel state information and the quantized identification of the compression model;

receiving, by the network device from the user equipment, the quantized compressed channel state information and the quantized identification of the compression model for generating the compressed channel state information;

determining, by the network device, an indicator corresponding to the compression model based on the identification; and

reconstructing, by the network device by using a decompression model, channel state information for the channel based on the compressed channel state information and the indicator.

14 . The method of claim 13 , wherein the decompression model at the network device, for reconstructing the channel state information, is common to a plurality of compression models comprising the compression model.

15 . The method of claim 14 , wherein an end-to-end model comprising the compression model and a further decompression model is trained with a training dataset, and the decompression model is trained with a combination of:

the indicator corresponding to the compression model, and

at least a portion of the training dataset.

16 . The method of claim 14 , wherein the identification of the compression model is based on a type of the compression model and a type of the user equipment.

17 . The method of claim 14 , wherein the decompression model is trained with an indicator corresponding to each of the plurality of compression models.

18 . The method of claim 17 , wherein the decompression model is further trained with at least a portion of a training dataset used to train the corresponding compression model.

19 . The method of claim 18 , further comprising:

de-quantizing, by the network device, the bit values into a set of values; and

reconstructing, by the network device using the decompression model, the channel state information based on the set of values and the indicator.

20 . The method of claim 19 , wherein an identification of the user equipment is binarized into bits, and wherein the method further comprises appending the binarized identification of the user equipment to the quantized compressed channel state information.