Channel state information feedback using channel compression and reconstruction
Systems and methods facilitating feedback of robust channel state information (CSI), such as to provide full CSI feedback or otherwise providing CSI feedback, are described. CSI encoders and/or decoders used by network nodes may implement channel compression/reconstruction based upon neural-network (NN) training of collected channels. A structured payload having an interpretable payload portion and an uninterpretable payload portion may utilized with respect to CSI feedback. The channel compression provided according to some aspects of the disclosure supports feedback of robust CSI, in some instances including full CSI, as determined by a particular network node. Other aspects and features are also claimed and described.
1. A method of wireless communication, comprising:
receiving, by a first network node from a second network node, encoded channel state information (CSI) included in an uninterpretable payload portion of a structured payload including an interpretable payload portion that is interpretable without decoding and the uninterpretable payload portion that is uninterpretable without decoding;
generating, by the first network node, reference signal (RS) based side information, based at least partially on a RS, wherein the RS based side information is in addition to the CSI that is configured for use in association with the CSI; and
decoding, by the first network node, the encoded CSI from the uninterpretable payload portion using information from the interpretable payload portion and the RS based side information to provide reconstructed channel information.
2. The method of claim 1 , further comprising:
signaling, by the first network node to the second network node, to indicate that the second network node is to use the structured payload for feedback of the encoded CSI.
3. The method of claim 1 , further comprising:
signaling, by the first network node to the second network node, to indicate that the second network node is to encode CSI of the uninterpretable payload portion based at least partially on the RS based side information.
4. The method of claim 1 , further comprising:
inputting a RS estimated channel to a neural-network to generate dimensional information for the RS based side information provided to a CSI decoder decoding the reconstructed channel information from the encoded CSI of the uninterpretable payload portion.
5. The method of claim 1 , wherein the encoded CSI in the uninterpretable payload portion comprises CSI encoded by a CSI encoder using neural-network based channel compression.
6. The method of claim 5 , wherein the encoded CSI in the uninterpretable payload portion is encoded based on the RS based side information.
7. The method of claim 1 , wherein the encoded CSI in the uninterpretable payload portion comprises CSI information defining an estimated channel associated with the second network node.
8. The method of claim 1 , wherein the interpretable payload portion comprises information configured to facilitate early decisions by the first network node with respect to decoding the encoded CSI or utilization of the reconstructed channel information.
9. The method of claim 8 , wherein the information of the interpretable payload portion includes at least one of burst interference information, recommended rank information, modulation and coding scheme (MCS) information, or information regarding which reference signal encoding of the encoded CSI is based upon.
10. The method of claim 1 , further comprising:
training a CSI decoder configured to perform the decoding of the encoded CSI from the uninterpretable payload portion using an autoencoder framework based on online data collection at the first network node, wherein the online data collection includes CSI information collected from the second network node and reference signal information monitored by the first network node, and wherein the encoded CSI in the uninterpretable payload portion is compressed using encoder parameters derived from the autoencoder framework.
11. The method of claim 10 , wherein both the interpretable payload portion and the uninterpretable payload portion are generated by a CSI encoder, wherein the CSI encoder has been trained using the autoencoder framework used in training the CSI decoder.
12. The method of claim 10 , wherein the uninterpretable payload portion is generated by a CSI encoder and the interpretable payload portion is added to the structured payload after encoding of the uninterpretable payload portion by the CSI encoder.
13. A method of wireless communication, comprising:
transmitting, by a first network node, a reference signal (RS);
encoding, by the first network node, channel state information (CSI) to provide encoded CSI, wherein the encoded CSI is based at least partially on the RS; and
transmitting, by the first network node to a second network node, the encoded CSI in an uninterpretable payload portion of a structured payload including an interpretable payload portion that is interpretable without decoding and the uninterpretable payload portion that is uninterpretable without decoding.
14. The method of claim 13 , further comprising:
receiving, by the first network node from the second network node, an indication that the first network node is to use the structured payload with respect to the encoded CSI.
15. The method of claim 13 , further comprising:
signaling, by the first network node to the second network node, to indicate that the first network node is using the structured payload for feedback of the encoded CSI.
16. The method of claim 13 , further comprising:
receiving, by the first network node from the second network node, an indication that the first network node is to encode the CSI with consideration of the RS.
17. The method of claim 13 , wherein the RS comprises a sounding reference signal (SRS).
18. The method of claim 13 , wherein the encoded CSI in the uninterpretable payload portion comprises information compressed by a CSI encoder using neural-network based channel compression.
19. The method of claim 13 , wherein encoded CSI of the uninterpretable payload portion comprises CSI information defining an estimated channel associated with the first network node.
20. The method of claim 13 , wherein the interpretable payload portion comprises information configured to facilitate early decisions by the second network node with respect to decoding the encoded CSI or utilization of reconstructed channel information obtained by decoding the encoded CSI.
21. The method of claim 20 , wherein the information of the interpretable payload portion includes at least one of burst interference information, recommended rank information, modulation and coding scheme (MCS) information, or information regarding which reference signal encoding of the encoded CSI is based upon.
22. The method of claim 13 , further comprising:
training a CSI encoder configured to perform the encoding of the CSI of the uninterpretable payload portion using an autoencoder framework based on online data collected at the first network node, wherein the online data collection includes reference signal observation information and decoder parameters derived from observation of the reference collected from the second network node and CSI reference signal information monitored by the first network node, and wherein the encoded CSI in the uninterpretable payload portion is compressed using encoder parameters derived from the autoencoder framework.
23. The method of claim 22 , wherein both the interpretable payload portion and the uninterpretable payload portion are generated by the CSI encoder.
24. The method of claim 22 , wherein the uninterpretable payload portion is generated by the CSI encoder and the interpretable payload portion is added to structured CSI feedback channel compression after encoding of the uninterpretable payload portion by the CSI encoder.
25. An apparatus configured for wireless communication, the apparatus comprising:
a memory; and
at least one processor coupled to the memory, wherein the at least one processor is configured:
to receive, by a first network node from a second network node, encoded channel state information (CSI) included in an uninterpretable payload portion of a structured payload including an interpretable payload portion that is interpretable without decoding and the uninterpretable payload portion that is uninterpretable without decoding;
to generate, by the first network node, a reference signal (RS) based side information, based at least partially on a RS, wherein the RS based side information is in addition to the CSI that is configured for use in association with the CSI; and
to decode, by the first network node, the encoded CSI from the uninterpretable payload portion using information from the interpretable payload portion and the RS based side information to provide reconstructed channel information.
26. The apparatus of claim 25 , wherein the at least one processor is further configured:
to input a RS estimated channel to a neural-network to generate dimensional information for the RS based side information provided to a CSI decoder decoding the reconstructed channel information from the encoded CSI of the uninterpretable payload portion.
27. The apparatus of claim 25 , wherein the interpretable payload portion comprises information configured to facilitate early decisions by the first network node with respect to decoding the encoded CSI or utilization of the reconstructed channel information, wherein the information of the interpretable payload portion includes at least one of burst interference information, recommended rank information, modulation and coding scheme (MCS) information, or information regarding which reference signal encoding of the encoded CSI is based upon.
28. An apparatus configured for wireless communication, the apparatus comprising:
a memory; and
at least one processor, wherein the at least one processor is configured:
to transmit, by a first network node, a reference signal (RS);
to encode, by the first network node, channel state information (CSI) to provide encoded CSI, wherein the encoded CSI is based at least partially on the RS; and
to transmit, by the first network node to a second network node, the encoded CSI in an uninterpretable payload portion of a structured payload including an interpretable payload portion that is interpretable without decoding and the uninterpretable payload portion that is uninterpretable without decoding.
29. The apparatus of claim 28 , wherein the interpretable payload portion comprises information configured to facilitate early decisions by the second network node with respect to decoding the encoded CSI or utilization of reconstructed channel information obtained by decoding the encoded CSI, wherein the information of the interpretable payload portion includes at least one of burst interference information, recommended rank information, modulation and coding scheme (MCS) information, or information regarding which reference signal encoding of the encoded CSI is based upon.
30. The apparatus of claim 28 , further comprising:
training a CSI encoder configured to perform the encoding of the CSI of the uninterpretable payload portion using an autoencoder framework based on online data collected at the first network node, wherein the online data collection includes reference signal observation information and decoder parameters derived from observation of the reference collected from the second network node and CSI reference signal information monitored by the first network node, and wherein the encoded CSI in the uninterpretable payload portion is compressed using encoder parameters derived from the autoencoder framework.