Performance comparisons of two adaptive subband structures

Abstract

This paper compares the performances of two subband adaptive FIR filter structures. In the first structure, the errors used in the adaptation of the subband filters are computed by comparing the decomposed and filtered input signal to the decomposed desired response. In the second structure, an overall error is computed by comparing the overall output and desired signals, and is then decomposed in order to be used in the adaptation algorithm. Subband structures with down-sampling factors both equal and smaller than the critical subsampling factor are considered. The convergence behaviors of the two structures are investigated and compared, taking into account the non-ideal filter bank characteristics. Computer simulations are presented verifying the convergence analysis.