Network Performance Evaluation Based on Interference Analysis in Avionics Switched Networks

The switched networking technology has been widely used in avionics systems due to its high bandwidth, scalability and low cost. In such networks, ensuring real-time guarantees often relies on evaluating worst-case delay. However, for applications prioritizing regular frame delivery, understanding delay variation is crucial. Since flow interference is the underlying cause of delay variation, analysis should be extended from a delay perspective to an interference perspective. Therefore, grasping how flow interference affects the delay distribution becomes an important issue. In this paper, we introduce a framework for analyzing flow interference, which includes interference value calculation, period determination, and two models in both the time and frequency domains. Time-domain interference analysis yields a metric for worst-case delay, while frequency-domain analysis introduces two metrics for assessing the degree and sharpness of delay variation. These metrics not only comprehensively evaluate network performance but also guide optimization efforts. Experimental results confirm that our proposed flow interference framework can effectively evaluate network performance and optimize worst-case delay by 36%, delay variation degree by 21%, and delay variation sharpness by 13.2%.