A Sequential Coding Approach for Short Length LT Codes over A WGN Channel

Abstract

Luby Transform (LT) codes which can be considered as one of the first type of rate less codes are introduced to be an efficient erasure code. In such codes a fountain of encoded symbols is continuously transmitted until the successfully delivery of all the data packets. Whenever the receiver collects N noisy encoded symbols the decoding operation starts. Usually belief propagation (BP) algorithm is used to decipher the code and extract the k data symbols. In this paper, short length LT codes are generated using efficient sequential encoding approach (SEA) and performance measurements have been done over additive white Gaussian (A WGN) channel. This coding technique generates its degree in a repeated sequential manner which yields a mutual relation between adjacent codes that will be used in the decoding part. The decoding complexity of the proposed structure is similar to that of the Raptor codes. The simulation results show that the proposed approach has better performance in terms of error floor and successful decoding ratio when compared to LT codes using robust soliton distribution (RSD) and memorybased-RSD (MBRSD) even when supported by belief propagation-pattern recognition (BP-PR) technique.