Optimization of Degree Distribution for Layer-aligned Multipriority Rateless Codes based on Safety Criteria of Ripple

Data loss or errors happen frequently on real networks. A common way of addressing this issue is packet retransmission, at the cost of potentially excessive transmission delay. For applications such as multimedia streaming, using channel coding to combat packet loss or errors is a more attractive option. Rateless codes have been an interesting approach due to their lower coding complexity and versatility. Layer-aligned multipriority rateless codes were originally designed for streaming with the capability of unequal error protection. In this paper, we are interested in finding a better degree distribution for such codes. We firstly derive the safety criteria of ripple sizes using a proposed leaping random walk model. In addition, we design an estimate function to predict the ripple size variation for the layer-aligned multipriority rateless codes. To achieve a better degree distribution, we use a genetic algorithm to optimize a multi-objective problem that we have formulated. This enabled us to improve the overall performance of the coding methods. Our simulation results demonstrate that the optimized degree distribution lead to significant improvements in error correction and data recovery rates.