Adaptive coding and transmitter diversity for slow fading channels

In this dissertation we consider several communications techniques to counteract the fading nature of the wireless channel. In particular we focus on the slow fading indoor channels, where the rate of change of the channel is low compared to the frame rate of the transmission system. In the present work we look at two methods to enhance the performance of the communications system: (i) the transmitter diversity system, and (ii) the Automatic Repeat Request (ARQ) protocols. In Chapter 2, we study the transmitter diversity system under the condition of the diversity channels being correlated. In Rayleigh fading, the performance of the diversity system deteriorates relative to the independent fading case. However, for the Rician fading system, when the specular diversity signals are not significantly out-of-phase with each other, we have discovered that the performance of the diversity system under correlated fading conditions can exceed the uncorrelated fading performance. In Chapter 3, we present a simple method to evaluate the throughputs of the basic ARQ protocols in slow fading channels. In Chapter 4, we design and analyse an adaptive ARQ/FEC protocol which changes the transmitter code-rate to match the channels conditions. The channel is estimated very simply from the implicit information contained in the ARQ feedback packets. The protocol is found to give a substantial gain over the basic ARQ protocols at almost all average SNR levels. In Chapter 5 we analyse a periodically switched antenna ARQ system. It is shown that at low fading rates this system is significantly superior to the single antenna system. In Chapter 5, we also analyse the performance of transmitter diversity ARQ systems. We have also proposed and analysed an adaptive ARQ/Diversity system in which the transmitter switches between the single and dual antenna transmission mode depending on the channel conditions. The adaptive ARQ/Diversity system has been found to significantly out-perform realistic block-coded systems