| Adaptive digital signal processing has matured to the point where it now constitutes an important part of statistical signal processing. However, high speed adaptive digital filtering and nonlinear adaptive digital filtering are yet to be developed.;In this dissertation, a family of high speed linear adaptive digital filters, which are suitable for parallel realization, are first presented. They are (1) Delayed N-path adaptive finite impulse response (DNA-FIR) digital filters; (2) Delayed N-path adaptive linear phase finite impulse response (DNALP-FIR) digital filters; (3) Delayed N-path equation-error based adaptive infinite impulse response (DNEEBA-IIR) digital filters. By using multiple digital signal processors in parallel, the processing speed of this type of filters can be increased compared with the conventional realizations. They are contrived to be useful for general applications in adaptive digital signal processing. Comparison studies have been conducted among the proposed DNA-FIR digital filter, the block implementation of adaptive finite impulse response (BIA-FIR) digital filter, and the conventional adaptive finite impulse response (CA-FIR) digital filter. It has been shown that the processing speed of the proposed DNA-FIR digital filter is N times that of the BIA-FIR digital filter (given block length N and time domain realization).;Then a nonlinear delayed N-path adaptive finite impulse response (NDNA-FIR) digital filter is developed, the parallel structure of which lends itself to high speed implementation. This NDNA-FIR digital filter compares favorably with the CA-FIR digital filter and the nonlinear median filter when applied to broadband noise cancellation. The last part of this dissertation describes the structure and its adaptive algorithm of a nonlinear adaptive infinite impulse response (IIR) digital filter. This IIR digital filter is bounded input bounded output (BIBO) stable. Based on this structure, an individual adaptation scheme is incorporated into the adaptive algorithm to improve the convergence speed. |
