A dynamic coefficient polynomial predistorter based on direct learning architecture

A dynamic coefficient polynomial predistorter based on direct learning architecture is proposed. Compared to the existing polynomial predistorter, on the one hand, the proposed predistorter based on the direct learning architecture is more robust to initial conditions of the tap coefficients than that based on indirect learning architecture; on the other hand, by using two polynomial coefficient combinations, different polynomial coefficient combination can be selected when the input signal amplitude changes, which effectively decreases the estimate error. This paper introduces the direct learning architecture and gives the dynamic coefficient polynomial expression. A simplified nonlinear recursive least-squares (RLS) algorithm for polynomial coefficient estimation is also derived in detail. Computer simulations show that the proposed predistorter can attain 31dB, 28dB and 40dB spectrum suppression gain when our method is applied to the traveling wave tube amplifier (TWTA), solid state power amplifier (SSPA) and polynomial power amplifier (PA) model, respectively.