In this thesis, we investigate the performance of minimum mean square error (MMSE) detector for asynchronous multipath fading DS/CDMA systems. BER evaluation of the MMSE detector suffers from the difficulty with calculating an exponential number of users since there exists residual multiple-access-interference (MAI) at the output of the MMSE detector. Especially in case of multipath combining, the evaluation is almost impossible because the residual MAI is correlated with each other in each path. Hence in the derivation of the BERs of MMSE detector, we make use of Gaussian approximation of residual MAI plus background noise. Next, we evaluate the BER of the MMSE detector with correlated Gaussian noise and also consider the effect of imperfect channel estimation. Channel estimation effect is investigated in two ways. The first is that Gaussian error exists around the true channel parameters, and the second is that system employs a concrete channel estimator such as finite impulse response (FIR) moving average (MA) filter or MMSE filter. Some quantitative results on the number of users, that of multipath, the ratio of Gaussian error variance, and doppler frequencies are obtained. Since the performance analysis is considerably accurate, these results can be applied for the all kinds of detectors using MMSE criterion.
The analytical results obtained in this thesis will be good performance measure for MMSE detector applications and help estimate the degradation of channel prediction error in real system.
