Pre-distortion with enhanced convergence for linearization

1. A pre-distorter for distorting an RF input signal to provide a pre-distorted radio frequency (RF) input signal to an amplifier that provides an amplified RF output signal, wherein the RF input signal has an envelope, comprising:

a radio-frequency signal processing circuit that distorts the RF input signal according to a polynomial of powers of the envelope, each power of the envelope being weighted by a corresponding pre-distortion weight; and

a performance monitor operable to compare a version of the amplified RF output signal to a delayed version of the RF input signal to provide an error signal, wherein the performance monitor is configured to iteratively adapt the pre-distortion weights based upon a gradient of a cost function, the cost function being a function of the error signal.

2. The pre-distorter of claim 1 , wherein the performance monitor is operable to generate the cost function based upon a power spectral density of the error signal.

3. The pre-distorter of claim 2 , wherein the performance monitor is adapted to divide the power spectral density into an in-band portion and an out-of-band portion, the cost function being based upon a weighted combination of the in-band and out-of-band portions.

4. The pre-distorter of claim 1 , wherein the performance monitor is operable to generate the cost function based upon an importance sampling of the error signal.

5. The pre-distorter of claim 1 , wherein the cost function is the mean square of the error signal.

6. The pre-distorter of claim 1 , wherein the performance monitor iteratively adapts the pre-distortion weights by iteratively adapting algorithmic coefficients to provide a set of first coefficients and linearly transforming the first coefficients into the pre-distortion weights.

7. The pre-distorter of claim 6 , wherein the performance monitor linearly transforms the first coefficients into the pre-distortion weights to increase a dimensional independence of an error space defined by the error signal.

8. The pre-distorter of claim 1 , wherein the cost function is based on an importance sampling of the error signal.

9. The pre-distorter of claim 1 , wherein the performance monitor comprises a digital signal processor.

10. A pre-distorter for distorting an RF input signal to provide a pre-distorted radio frequency (RF) input signal to an amplifier that provides an amplified RF output signal, wherein the RF input signal has an envelope, comprising:

a first means for distorting the RF input signal according to a polynomial of powers of the envelope, each power of the envelope being weighted by a corresponding pre-distortion weight; and

a second means for comparing a version of the amplified RF output signal to a delayed version of the RF input signal to provide an error signal, wherein the second means is configured to iteratively adapt the pre-distortion weights based upon a gradient of a cost function, the cost function being a function of the error signal.

11. The pre-distorter of claim 10 , wherein the second means is operable to generate the cost function based upon a power spectral density of the error signal.

12. The pre-distorter of claim 11 , wherein the second means is adapted to divide the power spectral density into an in-band portion and an out-of-band portion, the cost function being based upon a weighted combination of the in-band and out-of-band portions.

13. The pre-distorter of claim 10 , wherein the second means is operable to generate the cost function based upon an importance sampling of the error signal.

14. The pre-distorter of claim 10 , wherein the cost function is the mean square of the error signal.

15. A method for distorting an RF input signal to provide a pre-distorted radio frequency (RF) input signal to an amplifier that provides an amplified RF output signal, wherein the RF input signal has an envelope, comprising:

comparing a first version of the RF output signal to a delayed version of the RF input signal to calculate an error signal;

generating a polynomial of powers of the envelope, wherein each power is weighted by a corresponding pre-distortion weight; and

iteratively adapting the pre-distortion weights according to a gradient of a cost function of the error signal.

16. The method of claim 15 , wherein the cost function is a power spectral density of the error signal.

17. The method of claim 15 , wherein iteratively adapting the pre-distortion weights comprises:

iteratively adapting a first set of coefficients according to the gradient of the cost function to form an adapted first set of coefficients; and

linearly transforming the adapted first set of coefficients such that dimensions for an error space defined by the error signal have increased independence.

18. The method of claim 15 , wherein the cost function is an importance sampling of the error signal.

19. The method of claim 15 , further comprising:

forming a polynomial signal of powers of the envelope weighed by the pre-distortion weights;

multiplying a second version of the input signal with the polynomial signal to produce a multiplied signal, the polynomial signal having no constant component such that the multiplication produces no linear versions of the input signal;

adding the multiplied signal to the RF input signal to provide the pre-distorted input signal; and

processing the pre-distorted input signal through the amplifier to provide the amplified output signal, wherein the amplified output signal is a substantially linear function of the RF input signal.