Expansion Coding for Channel and Source Coding

Abstract: A general method of coding over expansion is proposed,which allows one to reduce the highly non-trivial problems of coding over analog channels and compressing analog sources to a set of much simpler subproblems, coding over discrete channels and compressing discrete sources. More specifically, the focus of this paper is on the additive exponential noise (AEN) channel, and lossy compression of exponential sources. Taking advantage of the essential decomposable property of these channels (sources), the proposed expansion method allows for mapping of these problems to coding over parallel channels (respectively, sources), where each level is modeled as an independent coding problem over discrete alphabets. Any feasible solution to the resulting optimization problem after expansion corresponds to an achievable scheme of the original problem. Utilizing this mapping, even for the cases where the optimal solutions are difficult to characterize, it is shown that the expansion coding scheme still performs well with appropriate choices of parameters. More specifically, theoretical analysis and numerical results reveal that expansion coding achieves the capacity of AEN channel in the high SNR regime. It is also shown that for lossy compression, the achievable rate distortion pair by expansion coding approaches to the Shannon limit in the low distortion region. Remarkably, by using capacity-achieving codes with low encoding and decoding complexity that are originally designed for discrete alphabets, for instance polar codes, the proposed expansion coding scheme allows for designing low-complexity analog channel and source codes.