Energy efficient polynomial kernel generation in full-duplex radio communication

1. A system for mitigating nonlinearity from a nonlinear behavior comprising:

a transceiver, comprising a transmitter and a receiver, configured to transmit and receive signals in a signal processing pipeline configured to cancel self-interference originating from a transmitter path of the transmitter;

a memory storing executable components; and

a processor, coupled to the memory, configured to execute or facilitate execution of the executable components, comprising:

a polynomial kernel generator component configured to perform a polynomial computation in a log domain, generate a plurality of kernels that model the nonlinear behavior of the transmitter in the log domain using fixed-point arithmetic with one or more inputs and cancel, via the plurality of kernels, the self-interference to the signal processing pipeline of the receiver, wherein the polynomial kernel generator component comprises a scheduling component that generates selection signals that schedule generation of specific polynomial kernels via the plurality of adders based on a state of a finite state machine.

2. The system of claim 1 , wherein the signal processing pipeline comprises one or more serial pipelines corresponding to input components of an input signal of the one or more inputs and a plurality of parallel kernel outputs corresponding in number to the plurality of kernels, wherein the inputs are fixed-point inputs or floating point inputs.

3. The system of claim 1 , wherein the executable components further comprise:

a distortion component configured to generate a model of the nonlinear behavior with segmented polynomial approximations based on the plurality of kernels to cancel the self-interference in the receiver.

4. The system of claim 1 , wherein the transmitter is configured to generate the self-interference to the receiver by consuming more power than the receiver.

5. The system of claim 1 , wherein the signal processing pipeline is configured as an all-digital, full-duplex or frequency division duplexing pipeline.

6. The system of claim 1 , wherein the polynomial kernel generator component comprises a plurality of adders serially coupled to one another to generate the plurality of kernels.

7. The system of claim 1 , wherein the signal processing pipeline comprises a quadrature component and an in-phase component that are processed independently along separate serial processing pathways of the polynomial kernel generator component into a decommutator that is configured to output the plurality of kernels in parallel.

8. The system of claim 1 , wherein the polynomial kernel generator component is further configured to generate the plurality of kernels with a latency that comprises K+1, wherein K is an integer greater than zero and comprises a number of kernels of the plurality of kernels.

9. The system of claim 1 , wherein the polynomial kernel generator component comprises:

a plurality of adders coupled to one another along the signal processing pipeline configured to process an in-phase component and a quadrature component of an input signal;

a plurality of log computation components configured to generate a log output in the log domain to at least two of the plurality of adders; and

a plurality of anti-log components, respectively coupled to the plurality of log computation components, configured to generate an anti-log output from a logarithmic input provided by the at least two of the plurality of adders.

10. The system of claim 1 , the polynomial kernel generator component is further configured to generate each kernel of the plurality of kernels in one clock cycle and the plurality of kernels in K+1 cycles, wherein K is a positive integer.

11. The system of claim 1 , wherein the executable components further comprise:

a kernel number component configured to determine a number of kernels of the plurality of kernels that linearly filter the self-interference based on a set of criteria related to an input of the receiver.

12. An apparatus configured to generate one or more polynomial kernels to cancel a non-linear behavior in a full-duplex communication mode, comprising:

a polynomial kernel generator, via a controller, configured to perform a polynomial computation in a log domain, generate a plurality of kernels in the log domain that model the nonlinear behavior of a transmitter path and cancel, via the plurality of kernels, a self-interference caused by the nonlinear behavior of a non-linear component of the transmitter path in a receiver path, wherein the polynomial kernel generator comprises a differential input path comprising, a first signal processing pathway and a second signal processing pathway, configured to receive complex input components of input signals respectively and generate the plurality of kernels independently of a multiplier and as a function of three adders.

13. The apparatus of claim 12 , wherein the polynomial kernel generator further comprises:

the differential input path comprising, the first signal processing pathway and the second signal processing pathway, configured to receive a quadrature component and an in phase component of a differential input signal, respectively;

a first select component configured to provide the in phase component to a first adder of the first signal processing pathway based on a finite state of the controller;

a second select component configured to provide the quadrature component to a second adder of the second signal processing pathway based on the finite state of the controller; and

a third select component configured to provide an output of the first signal processing pathway and the second signal processing pathway to a third adder based on the finite state of the controller.

14. The apparatus of claim 13 , wherein the polynomial kernel generator further comprises a feedback path configured to provide an output of third adder to the second select component coupled to the second adder.

15. The apparatus of claim 12 , wherein the polynomial kernel generator further comprises:

the differential input path comprising the first signal processing pathway and the second signal processing pathway respectively comprising:

a log component configured to utilize a piece-wise linear approximation with a base-2 logarithm calculation of an input signal having self-interference from the transmitter path, and generate a logarithmic output; and

an anti-log component configured to generate an anti-log of the logarithmic output from an anti-log computation adder.

16. The apparatus of claim 15 , wherein a first adder of the first signal processing pathway and a second adder of the second signal processing pathway generates a different kernel of the plurality of kernels at each clock cycle.

17. The apparatus of claim 16 , wherein the polynomial kernel generator further comprises a decommutator configured to output the plurality of kernels in parallel within the receive path and cancel the self-interference caused by the non-linear component of the transmitter path in the receiver path.

18. The apparatus of claim 12 , wherein the non-linear component comprises a power amplifier, and the transmit path consumes more power than the receive path in response to operation in the full-duplex communication mode.

19. The apparatus of claim 12 , wherein the polynomial kernel generator is further configured to generate the plurality of kernels that model the nonlinear behavior of the transmitter path, via a piecewise nonlinear polynomial approximation, in a cycle delay that is equal to a number of kernels of the plurality of kernels plus one.

20. A method for canceling self-interference of a full-duplex communication in a transceiver comprising a polynomial kernel generator, comprising:

receiving, via a receiver of the transceiver, a differential input signal comprising self-interference from a transmitter of the transceiver;

generating, via a controller, a polynomial computation in a log domain;

generating one or more polynomial kernels that enable modeling of the nonlinear behavior of the transmitter in the log domain;

eliminating the self-interference from the transmitter within the receiver based on a nonlinear model with one or more piecewise polynomial approximations utilizing the one or more polynomial kernels; and

performing an anti-log computation of the polynomial computation in the log domain and providing a result of the anti-log computation to at least one adder configured to generate the one or more polynomial kernels in a receive path of the receiver independent of a multiplication or a multiplier.

21. The method of claim 20 , further comprising:

scheduling inputs based on the differential input signal to be selectively inputted into three adders of the receiver;

wherein the generating the one or more polynomial kernels comprises generating outputs of the three adders to a decommutator configured to provide a parallel output of the one or more polynomial kernels in K+1 clock cycles, wherein K comprises a number comprising a positive integer of the one or more polynomial kernels.

22. The method of claim 20 , wherein the generating the polynomial computation in the log domain comprises performing a log of base 2 computation with a piece-wise linear approximation.