Generating an estimated non-linear echo signal

1. A method of generating an estimated non-linear echo signal of a transceiver, comprising:

the transceiver simultaneously transmitting a transmit signal, and receiving a receive signal;

receiving a plurality of data inputs of the transmit signal;

partitioning the plurality of data inputs into subsets;

computing a weight vector for each of the subsets;

computing a vector of addresses to memory locations for each of the subsets;

accessing values of interpolants at the memory locations based on the vector of addresses for each of the subsets;

calculating the estimated non-linear echo signal based on the values of the interpolants and the weight vector corresponding to each subset;

combining the estimated non-linear echo signal with the received signal.

2. The method of claim 1 , wherein each interpolant is a piece-wise linear function.

3. The method of claim 1 , wherein the plurality of data inputs is a continuous steam of data, and the partitioning of the data inputs into subset occurs continuously at a rate that each data input is received.

4. The method of claim 1 , wherein each subset comprises N i data inputs.

5. The method of claim 1 , wherein each subset comprises M i interpolant sites, and M i is dependent upon a degree of dependency of a non-linear echo signal being estimated on a number of data inputs.

6. The method of claim 4 , wherein the vector of addresses for each subset comprises N i +1 components.

7. The method of claim 6 , wherein the memory is partitioned into N i +1 blocks, a block corresponding with each component of the vector of addresses of a subset.

8. The method of claim 4 , wherein the weight vectors comprises N i components.

9. The method of claim 3 , wherein the value for the estimated non-linear echo signal is calculated once per received data input.

10. The method of claim 4 , further comprising determining an interpolation site for each subset, and further comprising partitioning the memory into M 1 interpolation sites for each subset, and each of N+1 blocks of the memory comprises (M 1 )/(N1+1) words.

11. The method of claim 10 , wherein each of the words are partitioned into M 2 subwords.

12. The method of claim 1 , wherein a first subset comprises N1 data inputs and a second subset comprises N2 data inputs, and wherein calculating a value for the estimated non-linear echo signal based on the values of the interpolants and the weight vector corresponding to each subset comprises:

generating N2+1 piece-wise linear approximations from (N1+1)*(N2+1) values of interpolants from memory using N1 components of a weight vector corresponding to the first subset;

generating the value for the estimated non-linear each signal by piece-wise approximating the N2+1 piece-wise linear approximations using N2 components of a weight vector corresponding to a second subset.

13. The method of claim 12 , wherein N1=N2.

14. The method of claim 1 , wherein computing a vector address for each subset comprises:

determining N i +1 closest interpolation sites for each subset;

for a first subset, computing a unique linear address that identifies a memory block and address of words comprising values of interpolants at the memory locations;

for a second subset, computing a unique address of values of interpolants within each word in the memory block.

15. The method of claim 1 , wherein computing a weight vector for each subset comprises:

for each subset, computing N i +1 interpolation sites;

selecting one of the interpolation sites as an origin, and designating other of the interpolation sites as terminal sites;

computing N i displacement vectors, wherein the displacement vectors are defined by the origin and the terminal sites;

computing a subset representing vector, wherein the subset representing vector is identified by the origin and a point that represents the subset, computing coefficients of a linear expansion of the subset representing vector based on the N i displacement vectors;

computing the weight vector based on the coefficients of the linear expansion.

16. A method of generating a non-linear replica signal of a transmit DAC of a transceiver, comprising:

the transceiver simultaneously transmitting a transmit signal, and receiving a receive signal;

receiving a plurality of data inputs to the transmit DAC, wherein the transmit DAC generates the transmit signal;

partitioning the plurality of data inputs into subsets;

computing a weight vector for each of the subsets;

computing a vector of addresses to memory locations for each of the subsets;

accessing values of interpolants at the memory locations based on the vector of addresses for each of the subsets;

calculating the estimated non-linear echo signal based on the values of the interpolants and the weight vector corresponding to each subset;

combining the estimated non-linear echo signal with the received signal.

17. The method of claim 16 , wherein the data inputs further comprise echo DAC data inputs of an echo DAC of the transceiver.

18. The method of claim 17 , wherein a first subset comprises transmit DAC data inputs and a second subset comprises echo DAC data inputs.

19. The method of claim 17 , wherein a first subset comprises transmit DAC data inputs and a second subset comprises ADC data outputs of a receive ADC of the transceiver.

20. The method of claim 16 , further comprising summing the value of the estimated non-linear echo signal with an output of a receive ADC of the transceiver.