A Study on Strategies and Mutant Factor in Differential Evolution Algorithm for FIR Filter Design

Differential evolution (DE) is a population-based optimizer that obtains the solution by iteratively improving the given measure of quality. In this work, DE has been applied in FIR Filter design to optimize the frequency response and the signed-power-of-two (SPT) terms of the filter coefficients. DE is influenced mainly by three design parameters, Strategy S, Mutant factor F and Cross-over probability Cr. The focus of this paper is the mutation step of DE which involves F and S. Simulations with all possible combinations of F and S show that strategy 2 and 3 with F=0.7 give the desired frequency response till filter order N=50. However, with consideration of SPT terms, S=2 and F=0.7 gives the optimized solution.