Multiuser demodulation for DS-CDMA systems in fading channels.

Abstract

The problems of optimal as well as suboptimal detection for CDMA transmissions over an additive white Gaussian noise (AWGN) channel, have been the focus of study in the recent past. However, CDMA transmissions are frequently made over channels which exhibit fading and/or dispersion; hence receivers need to be designed which take into account this behaviour. In spite of the major research effort invested in multiuser demodulation techniques, several practical as well as theoretical open problems still exist. Some of them are considered in more detail in this thesis. The aim of the thesis is to develop multiuser demodulation algorithms for mobile communication systems in frequency-selective fading channels, and to analyze their implementation complexity. The emphasis is restricted to the uplink of an asynchronous DS-CDMA system where the users transmit in an uncoordinated manner and are received by one centralized receiver. The original work that is undertaken for the MScEng study is the evaluation of a multiuser receiver structure for a frequency-selective fading channel, where there exists a steady specular path and two fading paths. Furthermore, the effect of using selection diversity is investigated by examining the bit error rate, asymptotic multi user efficiency and near-far resistance of the proposed detector structure. These results are confirmed both analytically and by simulation in the thesis. An investigation is also conducted into the application of neural networks to the problem of multiuser detection in code division multiple access systems. The neural network will be used as a classifier in an adaptive receiver which incorporates an extended Kalman filter for joint amplitude and delay estimation. Finally, some open problems for future research will be pointed out in the thesis. Keywords: AWGN channel , DS-CDMA system, frequency-selective, multi user demodulation, asymptotic multiuser efficiency, near-far resistance, neural network, Kalman filter.