Research On Multi-Processor Real-Time Scheduling For Mixed-Criticality Systems

Multi-core processors architectures are more and more widely used in embedded systems, to satisfy the increasing high-performance and low-power requirements. Multi-core processors provide great computing capacities to integrate multiple functionalities with different criticality levels into a shared platform. Because of this, mixed-criticality systems on multi-processor systems are becoming interesting trends on modern real-time embedded systems design. In the past few decades, the uniprocessor scheduling problem and multiprocessor scheduling problem of non-mixed criticality have been well studied. However, the scheduling problem of mixed criticality is a much harder challenge.The scheduling problem of mixed-criticality systems appears to be challenging, even on single-processor platforms. Most previous research on mixed-criticality has focused on uniprocessor system, yet many real-time applications are expected to, or already run on multi-processor platforms. Consequently, the research of mixed-criticality systems for multi-processor will prove useful.This thesis extends the state-of-the-art single-processor mixed-criticality scheduling algorithm EDF-VD to multi-processor systems. First, this thesis integrates EDF-VD into traditional workload partitioning schemes to get a multiprocessor mixed-criticality scheduling algorithm MC-PEDF. Although MC-PEDF performs better than previous solutions, this thesis find that the traditional workload partitioning schemes are not suitable for mixed-criticality systems as it does not explore the asymmetricity of workload on different criticality levels. To overcome this problem, this thesis proposes a novel workload partitioning policy OCOP (one criticality one partition). OCOP allows tasks to be reassigned to a different processor when criticality mode switch occurs, thus can better balance the resource utilization among processors on different criticality levels. Based on OCOP, this thesis proposes our second partitioned scheduling algorithm MC-MP-EDF. Experiments with randomly generated workload show that MC-MP-EDF can drastically improve the system schedulability comparing with MC-PEDF and other previous algorithms, especially for systems with more processors.