Unifying Source, Channel and Network Coding Technology

Lehigh University, Bethlehem PA

Investigators

Abstract

The emerging areas of Network information theory, distributed signal processing and cooperative communication promise unprecedented capabilities for improving the ways people represent, collect, disseminate and interpret information. They are also revolutionizing the communication and information world by introducing new models, new concepts, new methodologies, and new results. Despite the vast excellent results on the information-theoretic aspects, practical approaches that allow for software or hardware implementation with manageable complexity have been far lagging behind. To bridge the theory-practice gap, this proposal focuses on the algorithmic, schematic and strategic design of efficient and practical coding solutions in a cross-disciplinary context. To utilize channel coding technology in source coding is an interesting, albeit not particularly new, idea, whose efficiency has been exemplified by many recent findings in distributed source coding. Complimentary to the research in that line, here the PI proposes to bring source coding ideas to channel coding by investigating a basic but generalizable cooperative communication model where two nodes communicating wirelessly with the assist from a third node. To promote the recently developed notion that "source coding and channel coding are specialization of network coding" and that "network coding is generalization of source coding and channel coding", the PI further expands the system model to include many communicating and/or collaborating terminals and proposes to exploit source and channel coding ideas in network coding. The proposal starts with motivating the proposed methodologies, followed by a brief discussion of the technical background. After demonstrating the efficiency of her approaches with encouraging preliminary results, the PI proposes and seeks support for Leveraging useful source coding ideas to develop efficient channel coding solutions: Developing practical and efficient compress-forward user cooperation schemes by means of practical distributed source coding technology at large, and Slepian-Wolf coding and Wyner-Ziv coding in particular; developing guidelines for quantizer design and error control in the new scenario; and establishing significant analytical and algorithmic results. Unifying source and channel coding technology in network coding: Designing practical graph-based network codes that can effectively adapt to the changing network state and topology; unifying source and channel coding and other signal processing technology in network coding design; exploiting joint source, channel and network coding technology in such problems as data gathering, data dissemination and distributed detection; and examining bounded and asymptotic behavior of such communication/signal-processing systems. Intellectual Merit: The proposed work capitalizes on the PI's extensive experience in coding and cooperative systems, and helps cross-fertilize several important areas in the emerging field of network information theory. The proposal constitutes several refreshing and promising ideas to break the boundary between source coding and channel coding, to exploit the best of each and to incorporate them in network coding. Additionally, through constructive code design in a unified framework, the proposed research allows for the development of meaningful analytical results, one that sets more realistic limits and guidelines for practice than pure information-theoretic analysis. Broader impacts: The ideas and results developed in the proposed research have tight relationship with and bear useful implication in a wealth of problems and applications in the emerging area of network information theory, including, for example, multicast and broadcast problems, multiple access channels, coding for memories with defect, sensor networks, multimedia streaming, multi-source data collection and fusion systems and multi-destination data distribution systems. This is particularly important and timely considering that the current knowledge of network information processing is still very incomplete. The proposal also includes a teaching plan that serves as a useful complement to the proposed research activities.