Finite‐length extrinsic information transfer analysis and design of protograph low‐density parity‐check codes for ultra‐high‐density magnetic recording channels. Issue 11 (1st July 2016)

• Record Type: Journal Article
• Authors: Fang, Yi; Han, Guojun; Guan, Yong Liang; Bi, Guoan; Lau, Francis C.M.; Kong, Lingjun
• Is Part Of: IET communications. Volume 10:Issue 11(2016)
• Page Start: 1303
• Page End: 1311
• Publication Date: 2016-07-01
• Subjects: magnetic recording -- parity check codes -- error correction codes; finite length extrinsic information transfer analysis -- protograph low density parity check codes -- ultra high density magnetic recording channels -- LDPC codes -- intersymbol interference -- ISI channels -- finite length -- FL extrinsic information transfer -- EXIT algorithm -- protograph codes -- regular column weight‐3 code -- error performance -- error correction codes -- data storage systems; Telecommunication systems -- Periodicals; Speech processing systems -- Periodicals; 621.38205
• DOI: 10.1049/iet-com.2015.1233
• Languages: English
• ISSNs: 1751-8628

Abstract

Abstract : The authors study the performance of protograph low‐density parity‐check (LDPC) codes over two‐dimensional (2D) intersymbol interference (ISI) channels in this study. To begin with, the authors propose a modified version of finite‐length (FL) extrinsic information transfer (EXIT) algorithm so as to facilitate the convergence analysis of protograph codes. Exploiting the FL‐EXIT analyses, the authors observe that the protograph codes optimised for 1D ISI channels, e.g. the 1D‐ISI protograph code, cannot maintain their advantages in the 2D‐ISI scenarios. To address this problem, the authors develop a simple design scheme for constructing a family of rate‐compatible improved protograph (RCIP) codes particularly for 2D‐ISI channels, which not only outperform the 1D‐ISI protograph code, but also are superior to the regular column‐weight‐3 code and optimised irregular LDPC codes in terms of the convergence speed and error performance. More importantly, such RCIP codes benefit from relatively lower error‐floor as well as linear encoding and fast decoding. Thanks to these advantages, the proposed RCIP codes stand out as better alternatives in comparison with other error‐correction codes for ultra‐high‐density data storage systems.