Self Whitening Adaptive Equalization and Deconvolution Algorithms

In equalization and deconvolution tasks the correlated nature of the input signal slows the convergence speeds of the least mean square LMS and other stochastic gradient adaptive lters Prewhitening techniques have been proposed to improve convergence performance but the addi tional coe cient memory and updates for the prewhitening lter can be prohibitive in some applica tions In this report we present two simple algorithms that employ the equalizer as a prewhitening lter within the gradient updates A statistical analysis of these self whitening algorithms indi cates that they provide quasi Newton convergence locally about the optimum coe cient solution for deconvolution and equalization tasks Simulations indicate that the algorithms have excellent adaptation properties both for supervised and unsupervised blind adaptation criteria Extensions of the techniques to multichannel deconvolution and equalization tasks are also described