# Digital Predistortion

Laboratory demonstration of nonlinear compensation

Different amplitude distortion curves from memory effects

Cumulative complementary probability density function with and without crest factor reduction

Click image for enlarged view

Power spectra with and without crest factor reduction. Click
Power spectra with and without crest factor reduction.

Digital predistortion is an important enabling technology for software-defined radio because it improves spectrum control and efficiency, and can support more flexible radio frequency (RF) hardware.

Engineers at Southwest Research Institute (SwRI) are developing advanced algorithms and demonstration systems for digital predistortion. Projects range from consulting to assistance in subsystem development and testing to full system design, prototyping, and validation.

## Digital Predistortion

Internal research at SwRI has produced a set of algorithms for digital predistortion. SwRI engineers have also developed proof-of-concept subsystems for digital predistortion in:

- Third-generation (3G) cellular
- Television
- Other broadcast systems

Digital predistortion offers an economical and efficient mitigation for nonlinear operation of high power RF amplifiers and other RF hardware.

The SwRI digital predistortion program has four main components:

- Nonlinear compensation
- Memory effects compensation
- Linear compensation
- Crest factor reduction

## Nonlinear Compensation

Nonlinear compensation is a method of digital predistortion, where each sample of the base-band input signal is adjusted for the magnitude and phase distortion based on the instantaneous power level of that sample. The amount of distortion is estimated with a feedback system, and the compensation is usually implemented in an FPGA (Field Programmable Gate Array) for speed.

Nonlinear compensation is often what is meant when the term digital predistortion is used; however digital predistortion also includes:

- Memory effects compensation
- Linear compensation
- Crest factor reduction

The figure above shows the adjacent channel power ratio improvement of an amplifier with and without nonlinear compensation. The signal is a four-channel WCDMA (Wideband Code Domain Multiple Access) signal at 2.14 GHz.

## Memory Effects Compensation

Memory effects compensation builds on nonlinear compensation by taking into account several input signal samples at a time for improved performance. Although a significant amount of the nonlinear distortion in a typical amplifier is a function of the particular input signal level, in many high-power amplifiers, much of the nonlinearity is based on recent signal levels.

### Linear Compensation

SwRI has developed feedback system algorithms for automatically calibrating the output signals for up-conversion even with broadband signals, where the image and local oscillator leakage are completely covered up in the frequency domain by the main signal. Although these spurious signals may not be large enough in power relative to the main signal to cause problems in receivers, they do become a problem in feedback-based nonlinear and memory effects compensation systems. Therefore, linear compensation is a necessary prerequisite for optimal performance of nonlinear and memory effects compensation systems.

## Crest Factor Reduction

SwRI has developed crest factor reduction algorithms that reduce the peak-to-average power ratio of multi-carrier signals while minimally impacting the integrity of the signals. There are a number of methods that can be used, depending upon system latency and loss requirements.

## Related Terminology

digital predistortion • memory effects compensation • linear compensation • nonlinear compensation • crest factor reduction • RF hardware • software-defined radio • algorithm development