Abstract:

Since LT codes adopt the random encoding method,high coding overheads were required for a stable recovery probability of information.a random permute edge-growth-based coding algorithm was proposed, which could enhance the decoding probability of LT code through restricting the random connections between random permutation edges in the Tanner graphs.With this proposed coding algorithm, a decoding algorithm based on stopping set Gaussian elimination was proposed, which efficiently used the redundancy information in the stopping set after the belief propagation decoding.As a result, this decoding scheme could approach the performance of maximum-likelihood decoding.The simulation results show that, the proposed joint design on coding/decoding schemes could obviously reduce the overheads of such short length LT codes as 10＜sup＞3＜/sup＞, while achieve the decoding failure rate of 10＜sup＞- 4＜/sup＞.

Key words: LT codes, Tanner graph, random permute edge-growth, BP decoding, stopping set