Joint Source-Channel Coding Based on Cosine-Modulated Filter Banks for Erasure-Resilient Signal Transmission

This paper examines erasure resilience of oversampled filter bank (OFB) codes, focusing on two families of codes based on cosine-modulated filter banks (CMFB). We first revisit OFBs in light of filter bank and frame theory. The analogy with channel codes is then shown. In particular, for paraunitary filter banks, we show that the signal reconstruction methods derived from the filter bank theory and from coding theory are equivalent, even in the presence of quantization noise. We further discuss frame properties of the considered OFB structures. Perfect reconstruction (PR) for the CMFB-based OFBs with erasures is proven for the case of erasure patterns for which PR depends only on the general structure of the code and not on the prototype filters. For some of these erasure patterns, the expression of the mean-square reconstruction error is also independent of the filter coefficients. It can be expressed in terms of the number of erasures, and of parameters such as the number of channels and the oversampling ratio. The various structures are compared by simulation for the example of an image transmission system.