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.