| While standard variable length codes, such as Huffman codes, typically exhibit high coding efficiency, they are also extremely sensitive to transmission errors. Due to their high efficiency, these codes are often used in data compression standards, but their low error-resilience represents a considerable drawback, particularly for the applications characterized by extremely noisy channels (e.g. wireless communications), or the applications that require superior transmission quality (e.g. HDTV).; This dissertation considers the design of variable length codes with improved error-resilience, and joint source-channel coding systems that utilize these codes.; We first consider the algorithms for construction of efficient reversible variable length codes (RVLCs). RVLCs have recently gained considerable attention in video-compression standards (e.g. H263, MPEG-4), because of improved error-resilience relative to Huffman codes. However, the feasible design of highly efficient RVLCs is still an open problem. This dissertation proposes a novel RVLC construction algorithm that produces the codes of higher efficiency relative to the other algorithms in the literature.; Furthermore, we study the design of VLCs with respect to the trellis-based distance metrics, such as free distance. These metrics have shown to be an important indicator of the VLC error-resilience, particularly when advanced decoding strategies are utilized (e.g. soft-input VLC decoding, iterative source-channel decoding). This dissertation proposes an algorithm for construction of RVLCs with free distance greater than one. Since variable length codes typically have free distance equal to one, the proposed RVLCs exhibit significant improvement in symbol error rate relative to the VLCs constructed using standard methods.; Finally, we consider joint source-channel coding with error-resilient VLCs. We specifically focus on the design of iterative joint decoders, which utilize error-resilient RVLCs in combination with turbo codes. We propose several decoding systems that exhibit capacity-approaching performance, and significantly outperform standard separate source-channels decoders, as well as the previously introduced joint source-channel iterative decoders that utilize VLCs in combination with convolutional codes. |
