GA-based on-line optimization of CAN message scheduling

In this paper we describe a design and implementation of an on-line scheduling algorithm for controller area network (CAN) messages that exploits properties of genetic algorithms (GA) to achieve global message throughput optimization. The use of GA-based scheduler has been found effective even in time-critical distributed control systems, such as the considered control system that encompasses several control subsystems at the fire fighting vehicle. The experimental results show that the GA-optimized CAN message scheduler considerably decreased the average message queuing time.