DSI-ResCNN: A Framework Enhancing the Error-Tolerance Capacity of DNA Storage for Images

Cihan Ruan, Liang Yang, Rongduo Han, Shan Gao, Haoyu Wu, Qiming Yuan, Yanting Guo, Nam Ling

Abstract: Traditional silicon-based storage technologies have reached a bottleneck with the dramatic increase in global data storage demands. High energy consumption, limited natural resources, and environmental concerns are some of the key reasons. DNA storage emerges as a promising alternative method, offering relatively high density and long-term stability. However, this storage method still faces considerable challenges. One such challenge pertains to errors introduced during the DNA storage process, which require the development of information subsystems with enhanced error-tolerance capabilities. In the present study, we propose the integration of two modules into information subsystems to enhance their error-tolerance capacities for DNA storage of images. These two modules, namely the residual convolutional neural network (ResCNN) module and the DNA sequence dropout control (SDC) module, collectively constitute a framework, named DNA storage of images with residual CNN (DSI-ResCNN). DSI-ResCNN enhances the fidelity of images recovered from DNA sequences generated under error-prone conditions, thereby substantiating its potential as an effective solution for mitigating the adverse impacts of errors.

External IDs:doi:10.1109/access.2025.3549673