Source-channel rate allocation for progressive transmission of images

Progressive image transmission is difficult in the presence of a noisy channel, mainly due to the propagation of errors during the decoding of a progressive bitstream. Excellent results for this problem are made possible through combined source-channel coding – a method that matches the channel code to the source operational rate-distortion as well as channel conditions. This paper focuses on the key component of combined source-channel coding: rate allocation. We develop a parametric methodology for rate allocation in progressive source-channel coding. The key to this technique is an empirical model of decoded bit error rate (BER) as a function of the channel code rate. We investigate several scenarios. In the case of the memoryless channel, we present closed form expressions. For the fading channel and channels with feedback, where closed form results are elusive, our analysis leads to low-complexity algorithms. The results presented in this paper are applicable to any progressive source code, and any family of channel codes.