A Study on Iterative Decoding With LLR Modulator by Neural Network Using Adjacent Track Information in SMR System
Our previous work focused on a log-likelihood ratio (LLR) computed as the decoding reliability by a posteriori probability (APP) decoder in a two-dimensional magnetic recording (TDMR) system using shingled magnetic recording (SMR). In this article, we propose a neural network LLR modulator to reduce the influence of pattern dependent medium noise and to perform efficiently iterative decoding using a low-density parity-check (LDPC) code. In this study, we clarify that the performance improvement of iterative decoding can be obtained by considering LLRs of the adjacent track bits affecting the decoding bit as the input of the neural network. Furthermore, we clarify that the application of a hybrid genetic algorithm (HGA) selects LLRs by using the adjacent track LLRs affecting the decoding bit and realizes more effective iterative decoding.
Paper
Full text
A Study on Iterative Decoding With LLR Modulator by Neural Network Using Adjacent Track Information in SMR System
Semantic Scholar · Computer Science · 2019
Abstract
Our previous work focused on a log-likelihood ratio (LLR) computed as the decoding reliability by a posteriori probability (APP) decoder in a two-dimensional magnetic recording (TDMR) system using shingled magnetic recording (SMR). In this article, we propose a neural network LLR modulator to reduce the influence of pattern dependent medium noise and to perform efficiently iterative decoding using a low-density parity-check (LDPC) code. In this study, we clarify that the performance improvement of iterative decoding can be obtained by considering LLRs of the adjacent track bits affecting the decoding bit as the input of the neural network. Furthermore, we clarify that the application of a hybrid genetic algorithm (HGA) selects LLRs by using the adjacent track LLRs affecting the decoding bit and realizes more effective iterative decoding.