Multi-Step Online Scheduling to Improve Schedulability for In-Vehicle Time-Sensitive Networking

With the rapid development of intelligent vehicles, traditional in-vehicle networks struggle to provide real-time communication capacity. Time-Sensitive Networking (TSN) enables intelligent vehicles with high bandwidth, ultra-low latency and deterministic communications. One of the key mechanisms is the Time-Aware Shaper (TAS), which schedules real-time flows by a system-wide table to guarantee low latency and bounded jitter. Different from factory scenarios, the scheduling process of flows in vehicle networks needs to be online and make low disturbance to the system considering the safety-critical applications. Besides, with the large amount of flows involved in vehicle systems, the high schedulability demand of the scheduling algorithm is significantly arising. In this paper, we propose a multi-step online scheduling method, trying to reschedule flows selected by a metric, measuring the disturbance, when there is no feasible schedule for the new arriving flow. Evaluation shows that our method can improve 10% of schedulability in large-scale dynamic networks such as modern Intelligent vehicles.