Fast adaptive wide-band power amplifier feed-forward linearizer

A new algorithm is proposed for a wide-band power amplifier (PA) feed-forward linearizer applying DSP techniques. Using a PA model with memory and the input signal statistics, we effectively predict the linearizer parameters. Unlike other methods, this algorithm does not suffer from the slow-convergence problem. Two major factors cause the control parameters to be time-varying: the input signal statistics and other time-dependent environment factors. The input signal statistics are the fast time-varying factors, while the time-dependent environment factors or equivalently the PA operating point are the slow time-varying ones. In the proposed algorithm, we try to separate the fast time-varying factors from the slow time-varying ones. The input signal statistics are difficult to track but are assumed to be known to the system. Therefore, we only need to track the slow-varying PA operating point, represented by the coefficients of a proposed PA model, to determine the control parameters of the PA linearizer.