Statistical analysis of RaptorQ failure probability applied to a data recovery software

Abstract

In this work, we have implemented a data recovery software integrating the most recent rateless codes, i.e., RaptorQ codes. Thanks to the above-mentioned software, it is possible to recover data loss occurring on several kinds of network conditions. We have performed a statistical analysis of failure probabilities at several configurations of RaptorQ parameters. We have found a good agreement with the theoretical values of a random linear fountain code over Galois Field GF(256). Moreover, we have shown that the probability of having a certain number of failed decoded source blocks - when sending a fixed-size file - follows a Poisson distribution.