| Performance analysis plays a central role in the design of reliable communication systems. It helps to reduce the probability of error in the transmission of digital data or to improve the quality of the recovered signals when analog signals are communicated. In this dissertation, we first address the problem of performance analysis by presenting analytical methods to derive upper and lower bounds for the error rates of various systems. We then design a joint source-channel coding system for the transmission of analog sources over a multiple antenna communication link.; We begin by considering a single-antenna system with additive white Gaussian noise as well as block Rayleigh fading channels. It is shown that one can reduce the complexity of some of the existing methods and, at the same time, obtain tighter results, which is very favorable particularly when the complexity of the problem (i.e., the codebook size) is prohibitive. Closed-form formulas for the block pairwise error probabilities are derived everywhere.; We then consider a multi-antenna communication link with space-time orthogonal block coding and establish tight upper and lower bounds on the symbol and bit error rates of the system. The signaling scheme and the mapping between the signals and bits can be arbitrary.; We next consider the case where the input bit-stream has a non-uniform distribution, which allows the use of maximum a posteriori (MAP) decoding. We derive the MAP decoding rule, calculate a closed-form formula for the symbol pairwise error probability, and establish tight upper and lower bounds for the symbol and bit error rates. Another contribution is to demonstrate that systems with MAP decoding, which is a form of joint source-channel coding, can outperform those with separate source and channel coding (i.e., tandem systems) at all error rates of interest.; Finally, we propose a joint source-channel coding system for the transmission of analog sources over multiple-antenna links. The proposed setup uses soft-decision decoding channel-optimized vector quantization (COVQ) to achieve a large portion of the soft-decoding gain with reduced complexity. Two COVQs with fixed encoders and adaptive decoders are also designed. The superiority of the proposed COVQ over tandem systems is also demonstrated. |
