Adaptive filtering via wind driven optimization technique

Abstract

In this paper, an improved version of a recent structured population based evolutionary optimization algorithm called Wind Driven Optimization (WDO) is employed for the task of adaptive filtering. Adaptive filtering, the essence of all digital signal processing applications, often faces the challenge of selecting a suitable adaptive algorithm which can mitigate the effects of inter-symbol interference (ISI) and random noise. Recent trend has been to adopt soft computing optimization algorithms to optimize the parameters of adaptive filters. WDO is one of them. The objective of developing modifications in the original WDO is to improve the capability of carrying out global search and increase the accessibility of the global optimality. The proposed improved WDO seeks the help of Genetic Algorithm (GA) to bring about diversity in the population of swarm in each iteration, so that probability of the algorithm getting stuck in local optimum is greatly reduced. Improved WDO is implemented to optimize the coefficients of FIR based adaptive filter utilized to design Adaptive Channel Equalizer (ACE) and in System Identification (SI) task — two vital applications of adaptive filtering. The performance of the proposed algorithm is compared with original WDO, GA and the most popular conventional adaptive algorithm, Least Mean Square (LMS). The results confirm the efficacy of the said algorithm over other algorithms considered in this work.