Optimal Design of IIR Filter Using Dragonfly Algorithm

Abstract

The field of infinite impulse response (IIR) filter design mainly focused on the proper selection of filter parameters from the numerous possible combination. This filter design problem is based on determining the optimal set of parameters for unknown model such that its closely matches with the parameters of the benchmark filter. Many researchers have designed IIR filters using gradient based techniques like least mean square (LMS) method, etc. But, these gradient-based techniques have drawback of getting trapped into local solutions. To overcome this problem, evolutionary optimization techniques are used, which give global solutions. This paper utilizes a novel optimization technique known as dragonfly algorithm (DA) for the computation of the parameters of unknown IIR filter. Two benchmark functions are considered to prove the efficacy of the DA for IIR filter design problem. The results obtained using DA are compared with three existing algorithms namely, cat swarm optimization (CSO), particle swarm optimization (PSO), and bat algorithm (BA). The obtained results verify that the performance of DA-based IIR filter design is superior than that achieved by PSO, CSO, and BA.