Information shortening for joint source-channel coding schemes based on low-density parity-check codes

We propose a simple shortening algorithm for joint source-channel coding schemes based on low-density parity-check codes. Due to finite-length effects, such systems present high error floors when optimised for low-entropy binary symmetric sources and transmission over an AWGN channel. To mitigate such effects, we propose the reduction of the compression rate by inserting infinite reliability variable nodes which form cycles with length equal to the girth of the code used as source compressor. Simulation results show that such a strategy is able to efficiently trade a slight reduction of the compression rate for a significant performance enhancement.