A Semantic-Aware TSN Framework for Minimizing Age of Informative Data in Real-Time Industrial Monitoring Systems

Abstract

Real-time industrial monitoring systems rely on the timely delivery of semantically important data, such as anomaly-indicating sensor readings, to enable accurate and responsive decision-making. Credit-Based Shaping (CBS), a key mechanism in Time-Sensitive Networking (TSN), is well-suited for such systems due to its ability to dynamically manage aperiodic and bursty traffic without rigid transmission schedules. However, CBS was originally designed for Audio Video Bridging (AVB) traffic, which is periodic and less delay-sensitive, and thus lacks mechanisms to prioritize packets based on their semantic importance or freshness. As a result, critical updates may be delayed by the transmission of redundant or outdated packets. Motivated by these limitations, this paper presents an enhanced CBS mechanism, aiming to ensure the timeliness of semantically informative data in real-time industrial monitoring systems. Specifically, we encapsulate this enhancement within a semantic-aware TSN framework, which integrates three tightly coupled techniques: 1) semantic-based packet prioritization, 2) age-aware traffic shaping, and 3) age-aware packet forwarding and filtering. These mechanisms work in synergy to detect and expedite the transmission of high-priority, semantically meaningful packets, while suppressing redundant updates. Simulation results demonstrate that the proposed approach significantly improves anomaly detection responsiveness while maintaining efficient bandwidth utilization.