Parallel frequency domain adaptive line enhancer

Publication Date



Proceedings, 2nd Symposium on the Frontiers of Massively Parallel Computation


A comparative study is presented of parallel implementations of three least-mean-square FDALE (frequency domain adaptive line enhancer) algorithms based on circular and/or linear convolution of the filter delayed input and its impulse response. The first algorithm is based on circular convolution only. The second algorithm is a constrained least-mean-square algorithm that performs strictly linear convolution. The third is an unconstrained least-mean-square algorithm that allows either linear or circular convolution, whichever best minimizes the mean square error. Two models of the Connection Machine (CM-1 and CM-2) and one model of the Balance Machine (Balance 8000) were used in this study. Simulation results, speedup factors, and estimated spectra for algorithms based on circular and linear convolution are presented.


Conference Paper

This document is currently not available here.