A Genetic Algorithm based Scheduling Method for Automotive Ethernet

Abstract

Time-sensitive networking (TSN), which is well known as one of deterministic Ethernet, can ensure a real-time property of time-critical traffic by providing various standards. In particular, IEEE 802.1Qbv, which is one of the standards about TSN, can minimize a problem that occurred due to delay by ensuring the real-time transmission of a message that has high priority through scheduling. However, because scheduling messages in TSN is an NP-hard problem as same as the job schedule in a manufacturing process, an optimal schedule is needed to solve the problem. This study proposed a method to optimize a time-critical traffic schedule in TSN using a genetic algorithm. A chromosome in the genetic algorithm consists of messages to be scheduled, and schedules are created by the order of messages arranged in a chromosome. The performance of the chromosome is evaluated through the fitness function, which uses three performance indicators (end-to-end delay, jitter, and bandwidth utilization for guard band) as parameters. Finally, the proposed schedule optimization algorithm was performed over a simulation environment that simulated a real autonomous driving vehicle network to verify the applicability.