A Method for Multimodal Optimization with Application to Signal Processing

The paper presents a novel algorithm for the solution of Multimodal Optimization problems. The proposed method is hybrid of Nelder Mead algorithm and gradient method. The algorithm has direct application to problems encountered in the area of signal processing. The proposed method is applied for the estimation of taps in the design of adaptive filters. The convergence of the proposed algorithm is compared with the convergence trajectories of Steepest Descent method and found to be superior. The algorithm is presented for second order optimization problem and was applied for the design of three tap Wiener filter.