Designing Linear Phase FIR Filters with Particle Swarm Optimization and Harmony Search

Abstract

In recent years, evolutionary methods have shown great success in solving many combinatorial optimization problems such as FIR (Finite Impulse Response) filter design. An ordinary method in FIR filter design problem is Parks-McClellan, which is both difficult to implement and computationally expensive. The goal of this paper is to design a near optimal linear phase FIR filter using two recent evolutionary approaches; Particle Swarm Optimization (PSO) and Harmony Search (HS). These methods are robust, easy to implement, and they would not trap in local optima due to their stochastic behavior. In addition, they have distinguishing features such as less variance error and smaller overshoots in both stop and pass bands. To prove these benefits, two case studies are presented and obtained results are compared with previous implementations. In both cases, better and reliable results are achieved.