Bootstrap: a fast blind adaptive signal separator

Abstract

A fast multidimensional adaptive algorithm, Bootstrap, is proposed for multiple signal separation. It separates multiple uncorrelated signals imposed on each other. The bootstrap adaptive algorithm, which does not require training sequences, uses an optimization criteria that is based on minimization of output signal correlations. The learning process of this algorithm is compared with that of the least mean square (LMS) algorithm for different eigenvalue spreads. It has been found from computer simulations that the Bootstrap algorithm converges much faster than the LMS algorithm. The learning process of the Bootstrap algorithm is almost independent of eigenvalue spread.