Compressed measurement feedback using an encoder neural network

1. A method for wireless communications at a user equipment (UE), comprising:

compressing a plurality of measurements to obtain an encoder output indicating the plurality of measurements, the encoder output associated with a set of neural network decoder coefficients;

transmitting the set of neural network decoder coefficients for decoding the encoder output; and

transmitting the encoder output indicating the plurality of measurements.

2. The method of claim 1 , further comprising:

performing measurement operations to obtain the plurality of measurements.

3. The method of claim 1 , wherein the plurality of measurements correspond to a first number of bits for reporting the plurality of measurements and the encoder output comprises a second number of bits that is less than the first number of bits for reporting the plurality of measurements.

4. The method of claim 1 , further comprising:

receiving a configuration message indicating a periodicity for neural network decoder coefficient feedback, wherein the set of neural network decoder coefficients is periodically transmitted based at least in part on the periodicity for the neural network decoder coefficient feedback.

5. The method of claim 1 , further comprising:

receiving a configuration message indicating a priority rule for transmitting the encoder output; and

determining to transmit the encoder output on a resource based at least in part on the priority rule.

6. The method of claim 5 , wherein the resource comprises an updated resource, the method further comprising:

determining to transmit the encoder output on an initial resource;

determining that a message scheduled for communication at least partially overlaps with the initial resource; and

preempting transmission of the encoder output based at least in part on the message having a higher priority than the encoder output according to the priority rule, wherein transmitting the encoder output on the updated resource is based at least in part on preempting the transmission of the encoder output.

7. The method of claim 5 , further comprising:

determining that a message scheduled for communication at least partially overlaps with the resource; and

preempting the message scheduled for communication based at least in part on the message having a lower priority than the encoder output according to the priority rule, wherein transmitting the encoder output on the resource is based at least in part on preempting the message scheduled for communication.

8. The method of claim 1 , wherein the encoder output comprises a feature indicating at least two measurements of the plurality of measurements or a statistic associated with the plurality of measurements or both.

9. The method of claim 1 , wherein the plurality of measurements corresponds to a plurality of network devices, a plurality of sensors at the UE, a plurality of radio access technologies, or a combination thereof.

10. The method of claim 1 , wherein the plurality of measurements comprises a reference signal received power measurement, a reference signal received quality measurement, a received signal strength indicator, a reference signal time difference, a global navigation satellite system measurement, a round-trip time measurement, a sensor output, or a combination thereof.

11. The method of claim 1 , further comprising:

transmitting a message indicating that the UE has a capability to compress the plurality of measurements using an encoder neural network.

12. The method of claim 11 , further comprising:

performing a radio resource control setup procedure comprising transmission of the message.

13. The method of claim 1 , further comprising:

receiving a configuration message configuring the UE to operate in a compression enabled mode, wherein compressing the plurality of measurements is based at least in part on the UE operating in the compression enabled mode.

14. The method of claim 1 , further comprising:

receiving an indication of the plurality of measurements, wherein compressing the plurality of measurements further comprises:

inputting the plurality of measurements into an encoder neural network based at least in part on the indication of the plurality of measurements.

15. The method of claim 1 , further comprising:

receiving an uplink grant for the encoder output, wherein transmitting the encoder output is based at least in part on the uplink grant.

16. The method of claim 15 , wherein the uplink grant semi-statically grants a periodic resource for a periodic transmission of the encoder output.

17. The method of claim 15 , wherein the uplink grant dynamically grants a resource for a transmission of the encoder output.

18. The method of claim 1 , further comprising:

performing a handover procedure from a first network device to a second network device different from the first network device based at least in part on transmitting the encoder output.

19. The method of claim 1 , further comprising:

training an encoder neural network at the UE based at least in part on historical measurements at the UE or a UE performance metric or both.

20. The method of claim 19 , further comprising:

training a decoder neural network based at least in part on training the encoder neural network; and

transmitting an indication of the trained decoder neural network.

21. The method of claim 20 , wherein the UE comprises an auto-encoder comprising the encoder neural network and the decoder neural network.

22. A method for wireless communications at a network device, comprising:

receiving a set of neural network decoder coefficients corresponding to a decoder neural network;

receiving an encoder output;

decompressing the encoder output using the decoder neural network to obtain a plurality of measurements; and

communicating with a user equipment (UE) based at least in part on the plurality of measurements.

23. The method of claim 22 , wherein the encoder output comprises a first number of bits and the plurality of measurements corresponds to a second number of bits for reporting the plurality of measurements that is greater than the first number of bits.

24. The method of claim 22 , further comprising:

transmitting a configuration message indicating a periodicity for neural network decoder coefficient feedback, wherein the set of neural network decoder coefficients is periodically received from the UE based at least in part on the periodicity for the neural network decoder coefficient feedback.

25. The method of claim 22 , further comprising:

transmitting a configuration message indicating a priority rule for transmission of the encoder output, wherein the encoder output is received on a resource based at least in part on the priority rule.

26. The method of claim 22 , further comprising:

receiving a message indicating that the UE has a capability to compress the plurality of measurements, wherein decompressing the encoder output using the decoder neural network is based at least in part on the message.

27. The method of claim 22 , further comprising:

transmitting a configuration message configuring the UE to operate in a compression enabled mode, wherein receiving the encoder output is based at least in part on the UE operating in the compression enabled mode.

28. The method of claim 22 , further comprising:

transmitting an indication of the plurality of measurements for compression, wherein the encoder output indicates the plurality of measurements based at least in part on the indication of the plurality of measurements for compression.

29. An apparatus for wireless communications at a user equipment (UE), comprising:

at least one processor; and

at least one memory coupled to the at least one processor; and

instructions stored in the at least one memory and executable by the at least one processor to cause the apparatus to:

compress a plurality of measurements to obtain an encoder output indicating the plurality of measurements, the encoder output associated with a set of neural network decoder coefficients;

transmit the set of neural network decoder coefficients for decoding the encoder output; and

transmit the encoder output indicating the plurality of measurements.

30. An apparatus for wireless communications at a network device, comprising:

at least one processor; and

at least one memory coupled to the at least one processor; and

instructions stored in the at least one memory and executable by the at least one processor to cause the apparatus to:

receive a set of neural network decoder coefficients corresponding to a decoder neural network;

receive an encoder output;

decompress the encoder output using the decoder neural network to obtain a plurality of measurements; and

communicate with a user equipment (UE) based at least in part on the plurality of measurements.