Schedule construction under precedence constraints in flexray in-vehicle networks

As embedded time-triggered applications have widely replaced mechanical systems in modern automobiles, holistic scheduling of tasks and messages of such applications on in-vehicle networks has become a critical issue. For offering QoS (Quality of Service) guarantees, the holistic schedule must satisfy numerous constraints such as protocol specifications, delay constraints and precedence constraints between tasks schedules and messages transmissions. Existing approaches to this problem search through a vast design space of all possible joint task and message schedules. This leads to a high complexity and limits the scalability of such approaches for scheduling the large scale systems. To cope with this problem, we propose an approach that divides the holistic scheduling problem to two sub-problems: the sub-problem of message scheduling and the sub-problem of task scheduling, while precedence relations and end-to-end information passing between task instances and messages are preserved and the end-to-end deadlines are guaranteed. This helps to reduce the workload on the problem solvers and improves efficiency and scalability. In the first sub-problem, our approach optimizes scheduling the set of messages and allocates time windows for scheduling each task with respect to precedence constraints, end-to-end deadlines and FlexRay protocol specifications. The length of each time window helps to preserve the respective tasks schedulability and to provide flexibility for both task and message scheduling. The objective is defined with respect to extensibility issues. In the second sub-problem, our approach optimizes schedule of the set of tasks with respect to their allocated time windows and timing constraints. The objective is defined with respect to latency issues. We optimize the solution to each sub-problem using Mixed Integer Linear Programming optimization framework. Performance evaluations show that, compared with existing holistic scheduling approaches, our approach is more scalable and obtains better solutions in a reasonable amount of time.