A subband adaptive filter with the statistically optimum analysis filter bank

Conventional subband adaptive digital filters using filter banks have shown degradation in their performance because of the nonideal nature of analysis filters. For this problem, we propose two kinds of subband adaptive digital filters. The first one has the statistically optimum analysis filter bank which minimizes error signals. This first method, however, needs some statistics of the signals; therefore, we propose another type of subband adaptive digital filter which requires no a priori knowledge of the statistics regarding the signals. Computer simulations are also given in order to show the efficiency of the proposed subband adaptive digital filters.