An efficient algorithm for encoding and decoding of raptor codes over the binary erasure channel

Abstract

As the most advanced rateless fountain codes, Systematic Raptor codes has been adopted by the 3GPP standard as a forward error correction scheme in Multimedia Broadcast/Multicast Services (MBMS). It has been shown to be an efficient channel coding technique which guarantees high symbol diversity in overlay networks. The 3GPP standard outlined a time-efficient maximum-likelihood (ML) decoding scheme that can be implemented using Gaussian elimination. But when the number of encoding symbols grows large, Gaussian elimination need to deal with a large matrix with O (K 3 ) binary arithmetic operations, so the larger K becomes, the worse ML decoding scheme performs. This paper presents a better time-efficient encoding and decoding scheme while maintaining the same symbol recoverable performance, this encoding and decoding scheme is named Rapid Raptor Code. It will be shown that the proposed Rapid Raptor code Scheme significantly improves traditional Raptor codes' efficiency while maintaining the same performance.