Semantic-Aware Multipath TCP: Enhancing Industrial IoT Flexible Manufacturing Networks With GAN-Based Packet Recovery

Citations

WEB OF SCIENCE

0
Citations

SCOPUS

0

초록

Flexible manufacturing in Industrial Internet of Things (IIoT) environments leverages these capabilities to adapt dynamically to changing demands. However, IIoT applications face significant challenges in ensuring reliable data transmission due to high packet loss, resource constraints, and strict latency requirements. Multipath transmission control protocol (MPTCP) has emerged as a critical protocol for IIoT flexible manufacturing, leveraging multiple network paths to enhance throughput and reliability. Despite its advantages, MPTCP suffers from Head-of-Line (HoL) blocking. This article addresses the challenge of reliable data transmission in IIoT flexible manufacturing networks, focusing on the latency and throughput degradation caused by HoL blocking in MPTCP. We propose semantic-aware MPTCP (SA-MPTCP), integrating semantic communication with a generative adversarial network-based packet recovery mechanism that mitigates reordering delays and minimize retransmission overhead. SA-MPTCP prioritizes semantically significant data recovery, minimizes retransmission delays, and stabilizes congestion control. Evaluations show SA-MPTCP achieves 96% recovery accuracy in low-loss and over 85% in high-loss conditions, reducing delays by up to 90% compared to retransmission. SA-MPTCP enhances throughput and reliability, particularly for IIoT streaming and real-time control applications.

키워드

MPTCPGenerative adversarial networksInternet of ThingsIndustrial Internet of ThingsTimingModelingDelaysGeneratorsJoining processesPacket lossGenerative adversarial network (GAN)industrial internet of things (IoT)multipath transmission control protocol (MPTCP)packet recoverysemantic communication
제목
Semantic-Aware Multipath TCP: Enhancing Industrial IoT Flexible Manufacturing Networks With GAN-Based Packet Recovery
저자
Han, DeajinDao, Nhu-NgocCho, SungraeNa, Woongsoo
DOI
10.1109/TII.2026.3693342
발행일
2026-06
유형
Article; Early Access
저널명
IEEE Transactions on Industrial Informatics