Turbo multiuser detection for TD-CDMA

In this paper the novel detection scheme multi-step joint detection for TD-CDMA mobile radio systems is presented. Multi-step joint detection uses the turbo principle for iteratively improving data detection. Extrinsic information obtained by FEC-decoding in a previous step is used for a joint reduction of interference and improvement of asymptotic efficiency of the linear multiuser detector. Multi-step joint detection helps to overcome the problems of small asymptotic efficiencies present in linear multiuser detectors like the zero forcing equalizer in TD-CDMA mobile radio systems in the case of high system loads. Simultaneously complexity is much lower than that of optimum nonlinear multiuser detectors based on the Viterbi algorithm. As an alternative to simulations the performance of multi-step joint detection can be theoretically determined under certain not too restrictive assumptions. Both approximation and simulative performance results are presented in the paper. It is shown that in typical mobile radio systems the required SNR at the receiver input can be reduced by approximately 10 dB compared to linear multiuser detection. Thus, multi-step joint detection helps to increase the permissible number of mobile stations per cell or to decrease the required cluster size in TD-CDMA mobile radio systems and thus improves spectrum efficiency and capacity.