FAST EUCLIDEAN DIRECTION SEARCH ALGORITHM IN ADAPTIVE NOISE CANCELLATION AND SYSTEM IDENTIFICATION

摘要

Over the years, many algorithms have been proposed that can be used for adaptive filtering; least mean square (LMS) and normalized least mean square (NLMS) algorithms have been used in a wide range of signal processing applications because of their simplicity in computation and implementation. On the other hand, the best adaptive filter algorithm, i.e., recursive least square (RLS), converges significantly faster than the LMS and NLMS algorithms but its weakest point is the high computational complexity. The numerical instability problem of RLS algorithm motivates the use of simplified or partial RLS algorithms as a viable alternative to full RLS. In particular, we point out that the fast Euclidean direction search (FEDS) which is a recently introduced algorithm for adaptive filtering can indeed be interpreted as such partial RLS algorithms exhibiting a nice tradeoff between complexity and performance, and numerical robustness. This paper describes a new FEDS algorithm in noise cancellation for speech enhancement, interference cancellation and system identification applications. Furthermore, the performance of the introduced algorithm is investigated by using the energy conservation relation. The simulation results demonstrate the good performance of the FEDS algorithm in attenuating the noise and system identification. We also show that the tracking property of the proposed scheme is better than the RLS algorithm. Likewise, simulations are conduced to corroborate the presented studies and show that the theoretical results agree well with the simulation results.