Dynamic Reservation-Based Mixed-Criticality Task Set Scheduling

Mixed-criticality systems allow tasks of LO-criticality and HI-criticality to share the same platform. Such a design approach provides high resource usage for modern embedded systems. However, there are many challenges need to be addressed before mixed-criticality systems can be widely deployed. One key problem is how to schedule the task set to guarantee that all HI-criticality tasks meet their deadlines, and at the same time, minimize LO-criticality tasks´ deadline miss ratio. To address the problem, we present a dynamic reservation-based mixed-criticality task set scheduling algorithm, i.e., The DRBS algorithm, to minimize the LO-criticality tasks´ deadline miss ratio while guarantee HI-criticality tasks always meet their deadlines. Our experimental results show that the proposed scheduling algorithm can improve the schedulability of the existing approaches such as the EDF-VD and the Greedy algorithm by 70% and 10%, respectively. In addition, the effectiveness of the dynamic reservation-based (DRBS) algorithm with respective to the deadline miss ratio reduction is also experimentally validated.