This invention relates to the linearization of power amplifiers (or other nonlinear devices) having nonlinear conversion distortion when operated near saturation. Such operation introduces a nonlinear distortion to the transmitted signal evidenced by a spectrum re-growth that diverts some of the energy to adjacent communications channels, thereby decreasing the performance of the desired channel and creating interference in adjacent channels.

To solve this problem, predistortion techniques have been put forward.  However the challenge remains to estimate the predistortion function in order to fully linearize the amplifier. Most of the traditional estimation solutions contain drawbacks that are not conducive to the optimum efficiency and cost of power amplifier subsystems. Some of these drawbacks include the need for down-conversion and demodulation of the power amplifier output, the need for high speed signal processing, difficulty to ensure convergence to the solution, susceptibility to modeling errors, and the use of non-real time applications or lack of adaptivity.

This invention is based on two observations: (a) the envelope of a signal with a given modulation scheme and pulse shaping function has a unique statistical distribution or “type” characteristic, and (b) the response of the power amplifier is continuous (and monotonous for AM/AM distortions) up to its saturation point. Thus the key to this invention is the use of the distribution function of the modulated signal's envelope at the output of the power amplifier (PA) to derive the predistortion function instead of the conventional time domain approach.

The estimation technique consists in gathering a number of samples of the signal envelope at the output of the amplifier.  Once sufficient samples are gathered, the statistics of the output signal envelope are calculated and compared to the expected envelope based on either a priori knowledge or on the statistics of the measured amplifier input signal envelope. The difference between the statistics at the input and output of the amplifier contains the information necessary to derive the predistortion function. The process can be iterative to track changes over time and temperature.  

FEATURES

The key feature of this invention is the "type-based" technique that is used to adaptively estimate the predistortion function. This unique statistical distribution method precludes the need for parametric modeling of the PA and therefore avoids any model mismatches or fitting errors. As well, it allows operation over a wider range of nonlinearities.

In contrast to the prior art, this invention is also simpler and easier to implement:

• It does not require reconstruction of the modulating signal and can operate on a very low sampling rate independently of the modulation rate;
• It is functionally adaptive to cope with time-induced variations in PA response;
• It requires no synchronism of PA output measurement with other signals;
• It does not need a calibration signal;
• It does not interrupt the transmission.

STAGE OF DEVELOPMENT

MATLAB simulations have been used to validate the approach and the design.  A 30 GHz, QPSK transmitter prototype with a 2W SSPA with linearizer is under development.  Results for AM/AM compensation can be found in a conference paper (see below link to publication database).  AM/PM linearization prototype is under development.

APPLICATIONS

This invention is applicable to various digital communications systems where the efficiency and cost considerations of the power amplifier are important factors, such as in satellite, microwave and millimeter wave communications systems.

DELIVERABLES

This technology and other proprietary implementation know-how are available from CRC through a license agreement. Engineering support is also available on a cost-recovery basis for customization and/or implementation of these algorithms for specific client applications.

PROTECTION

U.S. Patent #6,885,241 B2
European Patent Application #PCT/CA2003/000458
Canadian Patent Application #2,457,404

MORE DETAILS

About CRC's Satellite Systems Research (RSS) Division: http://www.crc.gc.ca/rss

RSS Publications Database: http://www.crc.gc.ca/en/html/rss/home/publications/publications

CONTACTS

Please send an email to rss_dsp@crc.gc.ca