Abstract
A general optimum block adaptive (GOBA) algorithm for adaptive FIR filtering is presented. In this algorithm, the correction terms of the filter coefficients in each block, instead of the convergence factors, are optimized in a least squares sense. Unlike the optimum block adaptive algorithm with individual adaptation of parameters (OBAI), which can be used only for the case where the block length L is greater than or equal to the filter-tap number N, the GOBA algorithm has no constraints on L and N. It is shown that the GOBA algorithm is reduced to the normalized least mean squares (LMS) algorithm when L=1, and is equivalent to the OBAI algorithm when L>or=N. The convergence of the GOBA algorithm can be assured if the correlation matrix of the input signal is positive definite. Computer simulation based on an efficient computing procedure shows that the GOBA algorithm has better tracking capability than existing block adaptive algorithms.