Study of a new fast adaptive filtering algorithm

Abstract

A new fast adaptive filtering algorithm was presented by using the correlations between the signal s former and latter sampling times. The proof of the new algorithm was also presented, which showed that its optimal weight vector was the solution of generalized Wiener equation. The new algorithm was of simple structure, fast convergence, and less stable maladjustment. It can handle many signals including both uncorrelated signal and strong correlation signal. However, its computational complexity was comparable to that of the normalized least-mean-square (NLMS) algorithm. Simulation results show that for uncorrelated signals, the stable maladjustment of the proposed algorithm is less than that of the VS-NLMS algorithm, and its convergence is comparable to that of the algorithm proposed in references but faster than that of L.E-LMS algorithm. For strong correlation signal, its performance is superior to those of the NLMS algorithm and DCR-LMS algorithm.