UC Riverside

In this thesis, we investigate robust Gaussian joint source-channel coding with a distortion-noise profile. A distortion-noise profile is a function indicating the maximum allowed distortion value for each noise level. We analyze three different scenarios, and propose novel hybrid digital-analog based joint source-channel coding schemes which generalize or outperform existing schemes.In the first scenario, we look at power-distortion trade-off for the case of bandwidth compression when the bandwidth ratio is near zero. We propose hybrid digital-analog schemes and prove a general lower bound for minimum power which is valid for any distortion-noise profile. We also find upper bounds for minimum power for specific profiles including rational profiles with order one and two. In the second scenario, we consider energy-distortion trade-off for bandwidth expansion when the bandwidth ratio is near infinity. As in the bandwidth compression case, we propose hybrid digital-analog schemes and derive a general lower bound for minimum achievable energy for any profile. We discuss certain profiles including inversely linear, exponential, square-law, and staircase in more detail and establish upper bounds for minimum energy for these profiles. In the third scenario, we add side information available at the receiver onto the second scenario, and similarly calculate a general lower bound for minimum energy. We also propose coding schemes providing upper bounds on the minimum energy for linear and staircase profiles.