Joint Evaluation of TSN Topology and Scheduling Algorithm in Autonomous Driving Scenarios

This study proposes and constructs two typical topology architectures based on operational domains and functional domains for time-sensitive network (TSN) applications in automotive Ethernet, conducting comparative analyses of their latency performance and scalability. Building on this foundation, we formally model and evaluate three scheduling algorithms: queue isolation, window mapping, and injection time planning. Through experiments on the Open-planner platform across multiple scenarios, these results demonstrate that the window mapping algorithm excels in multi-period and high-load scenarios; the queue isolation algorithm is suitable for large-scale deployment; while the injection time planning algorithm exhibits strong robustness in complex and extreme high-traffic environments. These findings provide valuable references for automotive Ethernet topology design and scheduling algorithm selection.