Machine-learning architectures for simultaneous connection to multiple carriers

1. A method performed by a network entity associated with a wireless communication system, the method comprising:

determining at least one deep neural network (DNN) configuration for processing information exchanged with a user equipment (UE) over the wireless communication system using a simultaneous connection to a first component carrier and a second component carrier,

the at least one DNN configuration including:

a first portion for forming a first DNN that generates a first output associated with the first component carrier at the network entity and a second output associated with the second component carrier;

a second portion forming a second DNN at the UE;

communicating, to the UE, an indication of the second portion; and

using the first DNN to exchange, over the wireless communication system, the information with the UE using the simultaneous connection to the first component carrier and the second component carrier.

2. The method as recited in claim 1 , wherein the first component carrier is in a licensed band, and

wherein the second component carrier is in an unlicensed band.

3. The method as recited in claim 1 , wherein the simultaneous connection to the first component carrier and the second component carrier is downlink carrier aggregation, and

wherein determining the at least one DNN configuration comprises:

determining, for the second portion, a second DNN configuration that forms the second DNN to process the information by aggregating a first input associated with the first component carrier and a second input associated with the second component carrier.

4. The method as recited in claim 3 , further comprising:

determining a first sub-DNN configuration for forming a first sub-DNN that generates a split output including the first output and the second output;

determining a second sub-DNN configuration for forming a second sub-DNN that processes, based on the first output associated with the first component carrier, the information exchanged using the first component carrier; and

determining a third sub-DNN configuration for forming a third sub-DNN that processes, based on the second output associated with the second component carrier, the information exchanged using the second component carrier.

5. The method as recited in claim 1 , wherein the simultaneous connection to the first component carrier and the second carrier is uplink carrier aggregation, and

wherein the first DNN processes the information by aggregating a first input associated with the first component carrier and a second input associated with the second component carrier.

6. The method as recited in claim 1 , wherein the determining the at least one DNN configuration is based, at least in part, on a count of component carriers included in the simultaneous connection.

7. The method as recited in claim 1 , wherein the determining the at least one DNN configuration further comprises:

determining an end-to-end machine-learning configuration (E2E ML configuration) as the at least one DNN configuration.

8. The method as recited in claim 1 , wherein the second portion includes a sub-DNN configuration that forms a sub-DNN to process the information exchanged using the second component carrier, and

wherein the communicating the indication of the second portion to the UE comprises:

communicating, using the first component carrier, the sub-DNN configuration that forms the sub-DNN to process the information exchanged using the second component carrier.

9. The method as recited in claim 1 , further comprising:

receiving feedback from the UE;

determining a modification to the at least one DNN configuration by analyzing the feedback; and

communicating the modification to the UE for updating the second DNN based on the modification.

10. A method performed by a user equipment (UE) associated with a wireless communication system, the method comprising:

receiving an indication of at least one deep neural network (DNN) configuration for processing information exchanged over the wireless communication system using a simultaneous connection to multiple component carriers that include a first component carrier and a second component carrier;

forming, at the UE, at least one DNN based on the indication; and

using the at least one DNN to process the information exchanged over the wireless communication system using the simultaneous connection to the multiple component carriers by aggregating multiple inputs of the at least one DNN.

11. The method as recited in claim 10 , wherein the simultaneous connection to the multiple component carriers comprises downlink carrier aggregation, and

wherein using the at least one DNN to process the information comprises:

processing, using the at least one DNN, a first portion of the information exchanged using the first component carrier to generate a first input of the multiple inputs;

processing, using the at least one DNN, a second portion of the information exchanged using the second component carrier to generate a second input of the multiple inputs; and

aggregating, using the at least one DNN, the first input and the second input to recover the information.

12. The method as recited in claim 11 , wherein processing the first portion of the information comprises using a first sub-DNN to generate the first input;

wherein processing the second portion comprises using a second sub-DNN to generate the second input; and

wherein aggregating the first input and the second input comprises using a third sub-DNN to receive the first input and the second input and aggregate the first input and the second input to recover the information.

13. The method as recited in claim 12 , wherein the indication comprises a first indication, and the method further comprises:

receiving a second indication that indicates a modification to the first sub-DNN, the second sub-DNN, or the third sub-DNN; and

updating the first sub-DNN, the second sub-DNN, or the third sub-DNN based on the second indication.

14. The method as recited in claim 10 , wherein the simultaneous connection to the multiple component carriers comprises uplink carrier aggregation, and

wherein using the at least one DNN to process the information comprises:

generating, using the at least one DNN, a first output to exchange a first portion of the information over the first component carrier of the wireless communication system; and

generating, using the at least one DNN, a second output to exchange a second portion of the information over the second component carrier of the wireless communication system.

15. The method as recited in claim 14 , wherein forming the at least one DNN comprises:

forming the at least one DNN based on a portion of a distributed end-to-end machine-learning configuration.

16. The method as recited in claim 10 , wherein the first component carrier is in a licensed band, and

wherein the second component carrier is in an unlicensed band.

17. A network entity apparatus comprising:

a processor; and

computer-readable storage media comprising instructions that direct the network entity apparatus to perform operations comprising:

determining at least one deep neural network (DNN) configuration for processing information exchanged with a user equipment (UE) over a wireless communication system using a simultaneous connection to a first component carrier and a second component carrier,

the at least one DNN configuration including:

a first portion for forming a first DNN that generates a first output associated with the first component carrier at the network entity apparatus and a second output associated with the second component carrier;

a second portion for forming a second DNN at the UE;

communicating, to the UE, an indication of the second portion; and

using the first DNN to exchange the information, over the wireless communication system and with the UE, using the simultaneous connection to the first component carrier and the second component carrier.

18. The network entity apparatus as recited in claim 17 , the operations further comprising:

receiving feedback from the UE;

determining a modification to the at least one DNN configuration by analyzing the feedback; and

communicating the modification to the UE for updating the second DNN.

19. The network entity apparatus as recited in claim 18 , wherein the determining the modification comprises:

determining one or more architectural changes to the at least one DNN configuration; or

determining one or more parameter changes to the at least one DNN configuration.

20. The network entity apparatus as recited in claim 17 , wherein determining the at least one DNN configuration comprises:

determining, for the second portion, a second DNN configuration that forms the second DNN to process the information by aggregating a first input associated with the first component carrier and a second input associated with the second component carrier.

21. The method as recited in claim 1 , wherein the second DNN generates, at a second network entity, the second output associated with the second component carrier.