Nonlinear MMSE interference cancellation for asynchronous CDMA channels with convolutional coding

A trellis-based algorithm is derived in this paper for nonlinear MMSE estimation of convolutionally encoded signals in asynchronous CDMA channels with random spreading. Iterative nonlinear MMSE multiuser detection, proposed by the authors earlier for uncoded systems, is extended to forward error-correcting (FEC) coded CDMA systems by employing this algorithm. The resulting receiver is a Turbo-like combined FEC decoder and multiuser detector whose complexity is linear in the number of users. The proposed trellis-based estimation algorithm has the same order of complexity in the number of states of the FEC code as the Viterbi algorithm for optimal decoding of convolutional codes. Simulation results indicate that the proposed cancellation scheme can offer substantial improvement in performance over three other schemes in terms of cancellation error, capacity increase and tolerance to deep fades.