High-performance scheduling of DAG tasks on heterogeneous distributed mixed-criticality system

Real-time embedded systems filed tend to deploy functions of different critical levels on a unified platform for reasons related to SWaP (size, weight and power) and cost considerations. Highly critical tasks represent actual tasks with high urgency or high importance, which means that missing the deadline could have severe consequences, so priority guarantees are often required. Usually, the function of real-time systems is abstracted as a directed acyclic graph (DAG). In order to meet different computing requirements, heterogeneous computing units connected by buses are deployed in the system, which is generally called a heterogeneous distributed system. Aiming at the real-time scheduling problem of periodic functions of heterogeneous distributed mixed-criticality systems in the vehicle filed, a mixed integer liner programming method (MILP) is proposed to reduce the high- criticality functions and the system's overall deadline miss rate. Finally, the feasibility of the proposed method is verified by simulation experiments.