Source-Channel Coding Techniques in the Presence of Interference and Noise
Abou Saleh, Ahmad
Joint source-channel coding , Shannon-Kotel'nikov mapping , Broadcast channels , Correlated interference
As wireless systems proliferate worldwide, interference is becoming one of the main problems for system designers. Interference, which occurs when multiple transmissions take place over a common communication medium, limits system performance. Wireless devices can coordinate the use of scarce radio resources in order to manage the interference and establish successful communication. To effectively deal with the interference problem, some wireless devices must have a certain level of knowledge about the interference. In practice, this knowledge comes at the expense of using more resources (such as employing a proper channel training mechanism). With the remaining available resources, the question is how to achieve reliable communication? This thesis investigates an information theoretic approach and employs several coding techniques to improve system performance by either cancelling the interference or extracting knowledge from it about the information signal. The first part of this thesis considers the transmission of information signals over a fading channel that is disturbed with additional interference. The system's information theoretic limit in terms of mean square error distortion is assessed. Moreover, hybrid coding schemes are proposed and analyzed to obtain an achievable performance. As an extension to this problem, source channel-state estimation is investigated; in this case, the receiver is interested in estimating the information signal and the channel-state (interference). It is shown that the achievable performance is close to the derived limit. Motivated by multi-terminal systems, the transmission of a pair of sources over the two-user Gaussian broadcast channel in the presence of interference is then studied. Inner and outer bounds on the system's mean square error distortion are obtained. Unlike the first part of the thesis, in the second part, two additional constraints are added to the picture: 1) low coding delay and 2) low complexity. Similar communication scenarios as the ones in the first part are investigated under low delay and complexity requirements. These two constraints are motivated by the increased popularity of wireless sensor networks; a sensor node, often conceived as having limited lifetime and processing power, communicates its sensed field information to a fusion centre over a noisy wireless channel. To meet these challenges, we propose to use analog source-channel mappings. Parametric and nonparametric mappings are investigated in this last part of the thesis; it is noted that parametric mappings based on spiral and sawtooth curves are able to outperform linear scheme. Moreover, nonparametric mappings based on joint optimization of the encoder and the decoder are shown to outperform other low delay schemes.