Decorrelation algorithm for blind decision feedback equalizer with lattice structures

Abstract

A new algorithm was previously introduced for blind, adaptive equalization, known as the decorrelation algorithm. The algorithm is based on decorrelating the input to the decision or threshold device of a decision feedback equalizer to reduce the intersymbol interference at the equalizer's output. To increase the rate of convergence of this blind, adaptive, decision feedback equalizer, a fast Kalman structure was proposed, but not without a dramatic increase in complexity and limited numerical stability. In the present paper, more computationally efficient lattice-based structures are proposed. Using the decorrelation algorithm to control these structures, the authors maintain a high rate of convergence with better numerical stability in finite-precision environments.<>