Delay Optimization for Switchless ARINC 664 Mesh Networks with Cyclic Dependencies

Abstract

Switchless mesh topologies for modern avionics networks have been gaining popularity in recent years due to its inherent advantages in terms of the size, weight, and power (SWaP). However, it is a challenging task to calculate Network Calculus (NC) delay bounds for switchless mesh topologies due to the cyclic dependency problem, where the paths of interfering traffic flows form cycles. This paper, first, presents a method to automatically calculate the NC worst-case delay bounds for any switchless mesh topology with arbitrary traffic flows, where the cyclic dependency problem is solved using Time Stopping Method (TSM). Furthermore, we propose a Genetic Algorithm (GA) based delay optimization mechanism for switchless ARINC 664 mesh networks, where GA is used for exploring alternative paths to obtain tighter NC worst-case end-to-end delay bounds. The performance evaluation shows that the proposed GA based delay optimization provides consistently and significantly tighter delay bounds.