Zero-delay source-channel coding

In this thesis, we investigate the zero-delay transmission of source samples over three di↵erent types of communication channel models. First, we consider the zero-delay transmission of a Gaussian source sample over an additive white Gaussian noise (AWGN) channel in the presence of an additive white Gaussian (AWG) interference, which is fully known by the transmitter. We propose three parameterized linear and non-linear transmission schemes for this scenario, and compare the corresponding mean square error (MSE) performances with that of a numerically optimized encoder, obtained using the necessary optimality conditions. Next, we consider the zero-delay transmission of a Gaussian source sample over an AWGN channel with a one-bit analog-to-digital (ADC) front end. We study this problem under two di↵erent performance criteria, namely the MSE distortion and the distortion outage probability (DOP), and obtain the optimal encoder and the decoder for both criteria. As generalizations of this scenario, we consider the performance with a K-level ADC front end as well as with multiple one-bit ADC front ends. We derive necessary conditions for the optimal encoder and decoder, which are then used to obtain numerically optimized encoder and decoder mappings. Finally, we consider the transmission of a Gaussian source sample over an AWGN channel with a one-bit ADC front end in the presence of correlated side information at the receiver. Again, we derive the necessary optimality conditions, and using these conditions obtain numerically optimized encoder and decoder mappings. We also consider the scenario in which the side information is available also at the encoder, and obtain the optimal encoder and decoder mappings. The performance of the latter scenario serves as a lower bound on the performance of the case in which the side information is available only at the decoder.