Serial Concatenated Codes Constructed Based on 2-state Convolutional Codes

• Main Authors: Chien-i Lu, 呂建逸
• Other Authors: Mao-ching Chiu
• Format: Others
• Language: zh-TW
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/73923169141327347988

碩士 === 國立中正大學 === 通訊工程研究所 === 96 === Shannon’s noisy channel coding theorem has been published since 1948. Shannon showed that randomly chosen codes, along with maximum-likelihood decoding (MLD), can provide capacity-achieving performance. Afterward, a large amount of researches was proposed, which provides algebraic structures for developing easily implementable decoding algorithms. This is an important topic about channel coding. Turbo code, proposed in 1993, was the first one which can approach Shannon limit. Low-density parity check (LDPC) codes are a class of linear block codes and were published by Gallager in 1963. Tanner considered LDPC codes and showed how they may be represented effectively by a so-called bipartite graph, now called a Tanner graph. LDPC codes were confirmed to provide near capacity performance with iterative decoding over AWGN channels when the code length is large. Therefore, LDPC codes have become strong competitors to Turbo codes for error control in many communication and digital storage systems where high reliability is required. Currently, LDPC codes are widely used in many wireless communication standard, for example, IEEE 802.11n, IEEE 802.16e, and so on. In the thesis, we propose a code construction based on serial concatenation of accumulate codes. The constructed codes have the properties of capacity-approaching and easily encodable and easily decodable. In the proposed code construction, every component code constitutes a simple 2-state trellis structure. Due to the simple trellis structure of the proposed codes, an low-complexity iterative soft-in soft-out decoding algorithm was proposed. According to simulations, the proposed code with the iterative 2-state trellis decoding algorithm provides a better performance and a fast convergence rate as compared to conventional LDPC code with BP algorithm.