Optimized code design for constrained DNA data storage with asymmetric errors

Abstract

With ultra-high density and preservation longevity, deoxyribonucleic acid (DNA)-based data storage is becoming an emerging storage technology. Limited by the current biochemical techniques, data might be corrupted during the processes of DNA data storage. A hybrid coding architecture consisting of modified variable-length run-length limited (VL-RLL) codes and optimized protograph low-density parity-check (LDPC) codes is proposed in order to suppress error occurrence and correct asymmetric substitution errors. Based on the analyses of the different asymmetric DNA sequencer channel models, a series of the protograph LDPC codes are optimized using a modified extrinsic information transfer algorithm (EXIT). The simulation results show the better error performance of the proposed protograph LDPC codes over the conventional good codes and the codes used in the existing DNA data storage system. In addition, the theoretical analysis shows that the proposed hybrid coding scheme stores ~1.98 bits per nucleotide (bits/nt) with only 1% gap from the upper boundary (2 bits/nt).