A fast converging adaptive multiuser detection scheme with reduced complexity for CDMA systems

A low-complexity adaptive linear MMSE multiuser detection architecture for estimation of all users in a CDMA system is presented. The order of linear filter is reduced substantially by taking advantage of the knowledge of the signal space for demodulation of all users' signals on the up-link of a multi-access network. This results in fast convergence of the adaptive algorithms and also the feasibility of employing least-squares type of algorithms which are impractical otherwise for systems with large spreading gains. Further reduction in complexity without compromising on performance is shown to be possible by using a set-membership adaptive filtering algorithm that features optimal and sparse updating of the filter coefficients.