Configurable basis-function generation for nonlinear modeling

• US 8,576,941 B2
• Filed: 11/16/2010
• Issued: 11/05/2013
• Est. Priority Date: 11/16/2010
• Status: Active Grant

• Current Assignee: Telefonaktiebolaget LM Ericsson
• Original Assignee: Telefonaktiebolaget LM Ericsson
• Inventors: Bai, Chunlong, Morris, Bradley John, Lehman, Brian
• Primary Examiner(s): KASSA, ZEWDU A
• Application Number:
  US13/119,165
• Publication Number:
  US 20120119811A1
• Time in Patent Office: 1,085 Days
• Field of Search: 375/296, 375/295, 375/297, 455/114.3, 455/114.2, 455/126
• US Class Current: 375/296
• CPC Class Codes: H03F 1/3247 using feedback acting on predistortion circuits H03F1/3264 takes precedence
• CPC Class Codes: H03F 1/3252 using multiple parallel paths between input and output H03F1/3258, H03F1/3282, H03F1/3294 take precedence

Abstract

Digital predistorter circuits with selectable basis function configurations are described. In some embodiments, an input scaling block is introduced prior to a basis function generator structure. The input scaling factor is based on the input signal'"'"'s average power. In other embodiments, configurable connection coefficients are used to construct the orthogonal basis functions. Multiple sets of tap weights for the predistorter are maintained, each set corresponding to a given basis function configuration. In an example method for pre-distorting an input signal to compensate for distortion introduced by an electronic device, a statistic characterizing the input signal is calculated, and one of a pre-determined set of basis function configurations is selected, based on the statistic. A set of pre-distortion model weights corresponding to the selected basis function configuration are determined, after which the selected basis function configuration and the corresponding set of pre-distortion model weights are applied to the input signal.

Claims

1. A method for compensating an input signal for distortion introduced by an electronic device operating on the input signal to produce an output signal, characterized in that the method comprises:

calculating a statistic characterizing the input signal over a selected time interval;

selecting, based on the statistic, one of a predetermined plurality of basis function configurations for a non-linear model of pre-distortion for compensating the distortion introduced by the electronic device, wherein each basis function configuration corresponds to a different basis function set;

determining a set of pre-distortion model weights corresponding to the selected basis function configuration;

applying the selected basis function configuration and the set of pre-distortion model weights to the input signal, to produce a pre-distorted input signal for input to the electronic device.

2. The method of claim 1, wherein the statistic comprises an average signal power over the selected time interval.

3. The method of claim 1, wherein each of the predetermined plurality of basis function configurations comprises an input scaling factor that differs for each of the plurality of basis function configurations and a basis function set that is the same for all of the plurality of basis function configurations, and wherein selecting one of the predetermined plurality of basis function configurations comprises selecting the input scaling factor, based on the statistic.

4. The method of claim 1, wherein each of the predetermined plurality of basis function configurations comprises a basis function set that differs for each of the plurality of basis function configurations, and wherein selecting one of the predetermined plurality of basis function configurations comprises selecting one of the basis function sets, based on the statistic.

5. The method of claim 4, wherein each basis function set comprises one or more polynomials comprising a sum of power functions weighted by connection coefficients, and wherein selecting one of the predetermined plurality of basis function configurations comprises selecting a set of connection coefficients, based on the statistic.

6. The method of claim 1, wherein determining the set of pre-distortion model weights comprises retrieving, from memory, previously calculated pre-distortion model weights corresponding to the selected basis function configuration.

7. The method of claim 6, further comprising dynamically adapting pre-distortion model weights corresponding to at least one of the pre-determined plurality of basis function configurations, wherein the previously calculated pre-distortion model weights comprise previously adapted pre-distortion model weights corresponding to the selected basis function configuration.

8. The method of claim 7, wherein dynamically adapting pre-distortion model weights corresponding to a given basis function configuration comprises:

collecting first signal samples, corresponding to the input signal, over two or more time intervals during which the given basis function configuration is applied to the input signal;

collecting second signal samples, corresponding to the output signal, such that the second signal samples correspond in time to the first signal samples; and

calculating adapted pre-distortion model weights from the first signal samples and the second signal samples.

9. The method of claim 1, wherein determining the set of pre-distortion model weights comprises calculating the set of pre-distortion model weights from first signal samples corresponding to the input signal and second signal samples corresponding to the output signal, based on the selected basis function configuration.

10. The method of claim 9, wherein calculating the set of pre-distortion model weights from the first signal samples and the second signal samples comprises directly estimating the pre-distortion model weights from the first signal samples and the second signal samples.

11. The method of claim 9, wherein calculating the set of pre-distortion model weights from the first signal samples and the second signal samples comprises:

estimating device distortion parameters for a device distortion model from the first and second signal samples, wherein the device distortion model is based on the selected basis function configuration and reflects distortion introduced by the non-linear electronic device; and

calculating the pre-distortion model weights from the device distortion parameters.

12. A circuit for compensating an input signal for distortion introduced by an electronic device operating on the input signal to produce an output signal, characterized in that the circuit comprises:

a processor circuit configured to;

calculate a statistic characterizing the input signal over a selected time interval, from samples of the input signal;

select, based on the statistic, one of a predetermined plurality of basis function configurations for a non-linear model of pre-distortion for compensating the distortion introduced by the electronic device, wherein each basis function configuration corresponds to a different basis function set; and

determine a set of pre-distortion model weights corresponding to the selected basis function configuration; and

a pre-distortion application circuit configured to apply the selected basis function configuration and the set of pre-distortion model weights to the input signal, to produce a pre-distorted input signal for input to the electronic device.

13. The circuit of claim 12, wherein the statistic comprises an average signal power over the selected time interval.

14. The circuit of claim 12, wherein each of the predetermined plurality of basis function configurations comprises an input scaling factor that differs for each of the plurality of basis function configurations and a basis function set that is the same for all of the plurality of basis function configurations, and wherein the processor circuit is configured to select one of the predetermined plurality of basis function configurations by selecting the input scaling factor, based on the statistic.

15. The circuit of claim 12, wherein each of the predetermined plurality of basis function configurations comprises a basis function set that differs for each of the plurality of basis function configurations, and wherein the processor circuit is configured to select one of the predetermined plurality of basis function configurations by selecting one of the basis function sets, based on the statistic.

16. The circuit of claim 15, wherein each basis function set comprises one or more polynomials comprising a sum of power functions weighted by connection coefficients, and wherein the processor circuit is configured to select one of the predetermined plurality of basis function configurations by selecting a set of connection coefficients, based on the statistic.

17. The circuit of claim 12, wherein the processor circuit is configured to determine the set of pre-distortion model weights by retrieving previously calculated pre-distortion model weights corresponding to the selected basis function configuration.

18. The circuit of claim 17, wherein the processor circuit is further configured to dynamically adapt pre-distortion model weights corresponding to at least one of the pre-determined plurality of basis function configurations, wherein the previously calculated pre-distortion model weights comprise previously adapted pre-distortion model weights corresponding to the selected basis function configuration.

19. The circuit of claim 18, wherein the processor circuit is further configured to:

collect samples of the input signal over two or more time intervals during which the given basis function configuration is applied to the input signal, to obtain first signal samples;

collect samples of the output signal that correspond in time to the first signal samples, to obtain second signal samples; and

dynamically adapt pre-distortion model weights corresponding to a given basis function configuration by calculating adapted pre-distortion model weights from the first signal samples and the second signal samples.

20. The circuit of claim 12, wherein the processor circuit is configured to determine the set of pre-distortion model weights by calculating the set of pre-distortion model weights from first signal samples corresponding to the input signal and second signal samples corresponding to the output signal, based on the selected basis function configuration.

21. The circuit of claim 20, wherein the processor circuit is configured to calculate the set of pre-distortion model weights from the first signal samples and the second signal samples by directly estimating the pre-distortion model weights from the first signal samples and the second signal samples.

22. The circuit of claim 20, wherein the processor circuit is configured to calculate the set of pre-distortion model weights from the first signal samples and the second signal samples by:

estimating device distortion parameters for a device distortion model from the first and second signal samples, wherein the device distortion model is based on the selected basis function configuration and reflects distortion introduced by the non-linear electronic device; and

calculating the pre-distortion model weights from the device distortion parameters.

23. A wireless transmitter apparatus comprising a power amplifier configured to amplify an input signal to produce an output signal, wherein the wireless transmitter apparatus is characterized by further comprising:

a processor circuit configured to;

calculate a statistic characterizing the input signal over a selected time interval, from samples of the input signal;

select, based on the statistic, one of a predetermined plurality of basis function configurations for a non-linear model of pre-distortion for compensating distortion introduced by the power amplifier, wherein each basis function configuration corresponds to a different basis function set; and

determine a set of pre-distortion model weights corresponding to the selected basis function configuration; and

a pre-distortion application circuit configured to apply the selected basis function configuration and the set of pre-distortion model weights to the input signal, to produce a pre-distorted input signal for input to the power amplifier.