Modeling digital predistorter
Methods and apparatus are disclosed for predistorting an input signal to compensate for non-linear distortions introduced by an electronic device, for example, a power amplifier. The non-linear effects of a power amplifier can be modeled using different behavior models. Coefficients for one behavior model can be converted into coefficients for a different behavior model using a conversion function. A conversion circuit implementing the conversion function can be used in between a predistorter and an adaptation circuit that use different models.
1. A method of compensating for distortions introduced by an electronic device, comprising:
receiving an input signal to a predistorter and an output signal from the electronic device;
generating a first set of model coefficients based on a first predistorter model used in an adaptation circuit, the input signal to the predistorter, and the output signal;
generating a second set of model coefficients for a second predistorter model by applying a conversion function to the first set of model coefficients, wherein the second predistorter model is used by the predistorter;
adapting the predistorter based on the second set of model coefficients; and
predistorting input signals to the predistorter based on the second predistorter model.
2. The method of claim 1 , wherein the first predistorter model comprises a set of polynomial basis functions.
3. The method of claim 2 , wherein the set of polynomial basis functions includes a set of orthogonal basis functions.
4. The method of claim 2 , wherein the set of polynomial basis functions includes a set of power basis functions.
5. The method of claim 2 , wherein the second predistorter model is based on a look-up table (LUT) and the second set of model coefficients corresponds to entries in the LUT.
6. The method of claim 5 , wherein generating the second set of model coefficients comprises:
determining a desired input range for the LUT and a desired number of entries in the LUT;
dividing the input range into multiple bins, the number of bins being the same as the number of entries;
selecting one or more input samples within each of the multiple bins;
generating output samples corresponding to the one or more input samples based on the one or more input samples, the first set of model coefficients and the first predistorter model; and
evaluate a LUT entry for each of the multiple bins.
7. The method of claim 3 , wherein the second predistorter model is a set of power basis functions, and wherein the conversion function is a matrix.
8. The method of claim 4 , wherein the second predistorter model is a set of orthogonal basis functions, and wherein the conversion function is a matrix.
9. A circuit for compensating for distortions introduced by an electronic device to produce a predistorted input signal, wherein the electronic device operates on an input signal to a predistorter, said circuit comprising:
an adaptation circuit configured to compute a first set of model coefficients for a first predistorter model from the input signal to the predistorter;
a coefficient conversion circuit configured to generate a second set of model coefficients for a second predistorter model from the first set of model coefficients; and
a modeling circuit configured to adapt the predistorter based on the second set of model coefficients.
10. The circuit of claim 9 , wherein the first predistorter model comprises a set of polynomial basis functions.
11. The circuit of claim 10 , wherein the set of polynomial basis functions is a set of orthogonal basis functions.
12. The circuit of claim 10 , wherein the set of polynomial basis functions is set of power basis functions.
13. The circuit of claim 10 , wherein the second predistorter model is based on a look-up table (LUT), and the second set of model coefficients corresponds to entries in the LUT.
14. The circuit of claim 13 , wherein the coefficient conversion circuit is configured to:
determine a desired input range for the LUT and a desired number of entries in the LUT;
divide the input range into multiple bins, the number of bins being the same as the number of entries;
select one or more input samples within each of the multiple bins;
generate output samples corresponding to the one or more input samples based on the one or more input samples, the first set of model coefficients and the first predistorter model, wherein the output samples represent a gain factor over the corresponding input samples; and
evaluate a LUT entry based on the corresponding output samples for each of the multiple bins.
15. The circuit of claim 11 , wherein the second predistorter model is a set of power basis functions, and wherein the conversion function is a matrix.
16. The method of claim 12 , wherein the second predistorter model is a set of orthogonal basis functions, and wherein the conversion function is a matrix.