BER ANALYSIS OF DSSS MDPSK OVER THE SELECTIVE MULTIPATH RAYLEIGH FADING CHANNEL

Abstract

In this project, the DSSS MDPSK communication system with differential detection is analysed. The transmitter consists of two blocks. The signal, which is M­ ary, is first modulated by the MDPSK modulator. It is then spreaded by the DSSS modulator. Finally, it is mixed with a carrier before transmission. The receiver is actually a RAKE receiver with maximal ratio combining (MRC). In the receiver, the received signal is brought back to the baseband first by mixing with the carrier. It is then passed through each branch of the receiver, also known as the finger. Each finger of the receiver consists of a pair of in-phase and quadrature demodulators. The demodulators despread the signal before demodulating the MDPSK modulated signal. The channel is a slow selective multipath Rayleigh fading channel. The fading process is assumed constant over one symbol duration. The fading process is assumed linear with first order Butterworth filter characteristic.

Conventionally, two assumptions are made in the derivation of the bit error rate (BER) of the system. First, the fading in the channel is constant for at least two symbol intervals. By adopting a more effective analytical method given in [10] and [12], the assumption can be removed. From then on, it could be recognised that the error rate floor in the graph of BER vs. channel SNR is due not solely to the interference from the adjacent multipaths which is also known as the inter-symbol interference (ISI). It is also due to the fluctuations in the channel fading gain. Second, the outputs of the matched filters in the MRC receiver are assumed independent. Simulations are done to validate the BER derived from the second assumption. Simulation results indicate that the theoretical BER is sensitive to the path delays. However, the general shapes and behaviours of both the simulation and the theoretical curves are similar.

Other aspects like the effects of number of chips, order of diversity, channel gain, etc. have also been investigated. It is found that the BER decreases with increasing number of chips and order of diversity. The error rate floor of the system with the fading autocorrelation has also been investigated.