Optimal Step-Size LMS Algorithm Using Exponentially Averaged Gradient Vector

This paper proposes a new algorithm, which uses an optimal step size (OSS) weight adjustment scheme. This strategy leads to a better convergence rate and misadjustment in environments with sudden changes of parameters and for colored input data. The computational complexity is comparable with the well known RLS. The performance of the novel approach is verified by simulations under system identification scenario and compared with that of the NLMS and RLS algorithms. The strategy uses averaged values of the correlation matrix and the cross-correlation vector. Experimental results for car interior echo cancelation are presented including analysis of convergence rate and misadjustment