Interference Reduction in Multicarrier CDMA Via Blind Minimum Variance Multiuser Detector

This paper investigates the problem of reducing interference via an adaptive minimum variance (AMV) receiver in the Multi - Carrier Code Division Multiple Access (MC CDMA) forward link. In the MC - CDMA forward link, receivers are decentralized hence a computationally traceable approach to filter design is via adaptive estimation. It is well known that explicit training reduces the spectral efficiency of communication channels subject to time variation, while on the other hand, implicit training (i.e. Blind estimation) is not capacity preserving for noisy channels. It is necessary, therefore, to consider whether any potential advantage can be gained capacity wise in using implicit training for slowly time varying channels where some degree of channel estimation error can be expected. In this paper, the AMV algorithm is examined for MC - CDMA where results are presented indicating that for BPSK, 16QAM, and 64QAM modulation, the probability of error and capacity are nearly identical to the LMS MMSE approach provided that perfect channel side information (CSI) is available at the receiver. In the case where the CSI is subject to uncertainty, i.e. as practically would be the case when measuring a varying channel response in noise, the results show that approximately 2dB loss in performance occurs when the uncertainty is one tenth the symbol energy.