A Distortion Optimal Rate Allocation Algorithm for Transmission of Embedded Bitstreams over Noisy Channels

In this paper, a globally distortion optimal solution is proposed for FPP. The method, which is applicable to any source with arbitrary distortion-rate characteristics, has quadratic complexity in N, where N is the number of transmitted packets. It is based on constructing a search trellis in which trellis levels represent the number of packets, trellis states at a given level i embody the possible source rates corresponding to i packets, and edges represent different code rates. Such a trellis, starts from a single root state and spreads out in N levels. We prove that the backward application of a Viterbi-like algorithm to this trellis starting from the final states and working towards the root results in a survivor path that provides us with the distortion optimal rate allocation solution. The proposed method can also be applied to VPP, providing an alternative to the algorithm of V. Chande and N. Farvardin (2000) with comparable complexity