Multi-Task Semantic Communication System for DNA Transmission

  • Do, Tung Son; 
  • Truong, Thanh Phung; 
  • Tran, Hao Hoang; 
  • Tran, Kiet Nguyen Tuan; 
  • Ho, Manh Cuong; 
  • ... Cho, Sungrae; 
  • 외 2명
Citations

WEB OF SCIENCE

0
Citations

SCOPUS

1

초록

This paper presents DNA-Mamba-DeepSC, a novel multi-task semantic communication system specifically designed for efficient DNA sequence transmission. The proposed system addresses the critical challenges of preserving biological semantics while achieving high compression ratios and transmission reliability by leveraging Mamba-2 architecture with linear computational complexity. Our system simultaneously performs DNA sequence reconstruction and classification tasks through joint optimization, enabling superior performance compared to traditional communication methods. Experimental results demonstrate significant improvements across multiple metrics: up to 98.3% edit distance reduction compared to conventional UTF-8+Turbo methods, consistently high classification accuracy of 95−96% across various SNR conditions (from -6 dB to 18 dB), 62.1% latency reduction for processing long sequences, and 85.3% parameter reduction compared to baseline DeepSC while maintaining superior performance. The multi-task learning framework successfully preserves both sequence-level fidelity and semantic information during transmission, making the system particularly suitable for genomics research, personalized medicine, and biotechnology applications requiring reliable DNA data exchange.

키워드

DNA transmission; multi task learning; semantic communication
제목
Multi-Task Semantic Communication System for DNA Transmission
저자
Do, Tung Son; Truong, Thanh Phung; Tran, Hao Hoang; Tran, Kiet Nguyen Tuan; Ho, Manh Cuong; Tran, Anh Tien; Vu, Vinh Ngoc; Cho, Sungrae
DOI
10.1109/ICOIN68469.2026.11480607
발행일
2026
유형
Proceedings Paper
저널명
International Conference on Information Networking
페이지
359 ~ 364