An Improved Parameter Free Genetic Algorithm for CSD-FIR Filter design

A Canonical Signed Digit FIR filter (CSD-FIR filter) design problem is difficult to solve in polynomial computation time. Therefore, metaheuristics are used to obtain a semi-optimal solution to the problem. Parameter free genetic algorithm (PfGA) does not require to set any parameters that we need to adjust in advance. On the other hand, diversity of the search is relatively inferior for the CSD-FIR filter design problem. In this paper, we propose the CSD-FIR filter design method with PfGA with adaptive size of offspring (PfGA-AO) for better diversity and keeping the parameter free characteristic of the algorithm. PfGA-AO is applied for several design examples to show it∧s effectiveness.