Fixed point implementation for parameters extraction in a digital predistorter using adaptive algorithms

In this paper, the parameters extraction from the Volterra series to analyze the performance of a Digital Predistorter (DPD), for the Power Amplifier (PA) with memory, is introduced in two different ways: (1) different numerical methods for the parameters extraction and (2) the fixed point numerical format implementation for this numerical method. The parameters in the Volterra Model are typically calculated based on the mean square error criteria. In this paper, we present some alternatives to reduce the complexity, number of operations, and a PA linearization time ,with DPD dealing with OFDM signals. The simulation results show that with the Volterra model, both the LMS and the VSS algorithms are faster and more effective to calculate the parameters and mantain their convergence properties for a 32-bits implementation.