With conventional compression, the encoder exploits the statistics of the source signal. However, efficient compression can also be achieved by exploiting source statistics - partially or wholly - at the decoder only. This surprising insight is the consequence of information theoretic bounds established in the 1970s, particularly those by Wyner and Ziv for lossy coding with side information. We refer to compression that builds on these results and exploits receiver side information as Wyner-Ziv coding. While still in its infancy, practical Wyner-Ziv coding has the potential to widely impact distributed communication systems. For video signals, it might find applications ranging from enhanced digital television broadcasting, low-complexity networked cameras, and improved visual surveillance systems for military and civilian applications. In this project, we develop and explore new Wyner-Ziv coding techniques and apply these to two different video communication problems. Wyner-Ziv coding is a radical departure from conventional, non-distributed coding, and, therefore, a significant part of our effort is devoted to fundamental problems of code design, quantizer design, and the role of transforms. In the first application, we consider a novel video compression system with asymmetric complexity, useful for wireless mobile cameras. Since Wyner-Ziv coding exploits the source statistics in the decoder, the encoder can be very simple, at the expense of a more complex decoder. In the second application, we devise a novel unequal error protection scheme for video broadcasting that protects the video waveform with a lossy source-channel code. Such a scheme does not require a layered, embedded representation of the video signal to achieve graceful degradation with worsening channel conditions and is hence compatible with deployed single-layer video compression standards.
