Hybrid-GA based static schedule generation for time-triggered ethernet
Bingqian Li, Wang Yong · 2016
Time-triggered Ethernet (TTE) has its unique TT traffic schedule to guarantee a global communication scheme. Scheduling problem in TTE's static segment is regarded as NP complete problem for its mess constrains, complex topology and large scale of messages. Another problem in TT traffic scheduling is lack of flexibility. Former scheduled communication table is unable to be compatible with extended changes in TTE. To promote the performance of scheduled TT traffic and create extra space in time domain for further use, we introduced genetic algorithm in searching for global solution and proposed a hybrid genetic algorithm (hybrid-GA), which combined bin-packing and genetic algorithm. We transformed the problem into a typical bin-packing model and further on solved it with hybrid-GA, which fused both bin-packing algorithm and genetic algorithm. Our experiments using hybrid-GA show that the algorithm we proposed could effectively satisfy real-time demands and perform well in time slot consuming. Compared to pure genetic algorithm, hybrid-GA converges faster for its local searching ability benefiting by fused bin-packing algorithm.