An Efficient Decoding Algorithm for Concatenated RS-Convolutional Codes

• Main Authors: Yu-shun Hsiao, 蕭宇舜
• Other Authors: Wan-De Weng
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/52255455802878279297

碩士 === 國立雲林科技大學 === 電機工程系碩士班 === 101 === Concatenated schemes using Reed-Solomon (RS) codes as the outer code and convolutional codes as the inner code have been used for digital video broadcasting systems. The convolutional codes can sweep the imperfectness of the channel while RS codes can fight burst errors which are caused by the the decoding process of convolutional codes when a Viterbi decoder is employed. Using this method, the concatenated coding scheme can achieve a relatively high coding gain. The concatenated Reed-Solomon/Convolutional (RS/Conv) coding scheme has been around for a long time, but using iterative decoding schemes based on turbo principle for decoding these codes is relatively new. The main reason is that the soft input soft output (SISO) decoding algorithms for RS code are usually complicated, especially when codeword length is long. In 1994, Pyndiah et al. developed a simple and effective SISO decoder for RS codes based on the Chase algorithm. Pyndiah''s algorithm has been used for turbo decoding of both product and serial concatenated codes . In this paper, we improve the decoding algorithm for the concatenated RS/Convo scheme investigated in by adopting Kaneko''s algorithm for RS decoding. This idea was inspired by "An efficient decoding algorithm for block turbo codes", in which the authors S. Dave, J. Kim, and S. Kwatra, discussed how to reduce decoding complexity of turbo product codes. The proposed algorithm reduces the computation by employing stopping criteria during the Chase decoding process. While conventional Chase algorithm generates constant number of codeword candidates, we expect to examine fewer number of codeword candidates using Kaneko''s stopping criteria. The major advantage of the proposed algorithm is lower computational complexity without suffering performance loss.