Power amplifier digital predistortion - fixed or adaptive?

RF linearization and digital predistortion have been shown to be effective at counteracting the distortion due to the non-linear behavior of high-power amplifiers operating near the power-efficient saturation point. The low cost and simple implementation of digital predistortion has made it a highly desirable alternative to expensive and power hungry RF linearizers. A digital predistorter scales and rotates the input samples based on the samplespsila magnitude. The adjustment coefficient depends on the amplifier nonlinearity. These coefficients can be obtained directly by measuring the amplifier characteristics with a network analyzer. The coefficients can also be arrived at by adaptively adjusting an initial estimate such that an error metric is minimized. The error metric is typically the difference between the transmitted and the received samples. The adaptation is accomplished using a standard gradient descent algorithm such as least mean squares (LMS). In this paper we present measurement results comparing the two approaches. The measurements were performed for a space-qualified traveling wave tube amplifier (TWTA). We show that the adaptive approach provides better frequency domain performance (shoulder reduction). However, the improvement in signal to noise ratio and bit error rate is almost identical for the two approaches.