Erasure resilient channel coding strategy for distributed sensor networks

Abstract

Rateless erasure codes have recently attracted much attention in the communications research community. Luby transform codes being the most well known, were the first practical realisation of record-breaking sparse-graph codes, particularly for binary erasure channels. These codes have the advantage of not requiring a priori knowledge of specific channel conditions and lend itself to application in nondeterministic distributed wireless sensor networks. This paper revisits the Luby transform code, and proposes a novel multi-path routing scheme where each part of the code follows any available transmission path. The packet delivery is not guaranteed. However, if ample packets are received from the multifarious paths, decoding the information can be accomplished at high probability. Our results indicate that just a few conditions are sufficient to define an optimal routing path with regard to the proposed encoding and decoding strategy. This work is of much interest to Wireless Sensor Networks (WSN's), due to typical topological characteristics.