Least-squares CMA with decorrelation for fast blind multiuser signal separation

Abstract

A blind algorithm with implicit signal selectivity capability is proposed. The algorithm is an evolution of the original multiuser constant modulus algorithm of Papadias and Paulraj (see IEEE Sig. Proc. Letters, vol.4, no.6, p.178-81, 1997). The new algorithm features a least-squares type updating rule for fast convergence rate and an adaptive control of the weight of the decorrelation term which improves the steady-state error variance. The expected improvements of the proposed algorithm are verified through simulations with smart antennas in a spatial-division multiple access system.