RAPID FREQUENCY-DOMAIN ADAPTATION OF CAUSAL FIR FILTERS

Normalizing the convergence coefficient of the block frequency-domain least mean square (LMS) algorithm in each frequency bin can improve th e convergence rate, but in some applications can lead to a biased stea dy-state solution if the filter is constrained to be strictly causal. An algorithm is presented in which the spectral factors of the bin-nor malized convergence coefficient are used before and after the causalit y constraint is applied in the adaptation algorithm, which converges r apidly to the optimal causal filter.