Multiuser detection using iterative (turbo) decoding techniques

Abstract

Multiuser detection has become an important technique for the third-generation wireless communications, in particular, code division multiple access (CDMA). A soft-in soft-out (SISO) multiuser detection architecture based on a cross-entropy minimization technique has been developed. Minimum cross entropy is an optimal lossless decoding algorithm but its complexity limits its practical implementation. Hence, the algorithm is implemented analogous to turbo decoding. The proposed architecture uses a conventional CDMA system followed by two cascaded logarithm maximum a posterior (log-MAP) SISO decoders. Channel information is fed into the SISO decoder, which generate the L-values and processing proceeds on an iterative manner similar to the decoding of the turbo codes. The cross-entropy minimization provides a stop criterion for the multiuser detection to prevent any unnecessary iteration when no further improvement is possible.