The Research Of Mixed Criticality Scheduling Based On Virtual Deadline

With the rapid development of information technology,embedded systems are used more and more widely.Due to the cost,volume,energy consumption,utilization and other aspects of the constraints,using a shared processor platform to handle different criticality functions has become a new development direction.This kind of system is also called mixed criticality system,the generation of mixed criticality provides a new development space for design of the embedded system.The design of mixed criticality scheduling algorithms and the correctness analysis of scheduling have become the focus of academic research.Mixed criticality real-time scheduling must simultaneously satisfy two objectives: in the worst case execution time,to meet the high criticality functional requirements;at the same time,meet the design requirements : efficient use of system resources.This paper makes a detailed analysis of the existing mixed criticality task scheduling algorithm.The earliest deadline virtual mixed criticality task scheduling algorithm is a great algorithm,and has become a mainstream scheduling algorithm.For high criticality tasks,algorithm set two different deadline in different modes,and the performance of low criticality tasks are improved.Although this algorithm is optimal in a single processor scheduling algorithm of mixed criticality,at the mode conversion,all high criticaity tasks are assumption simultaneous enter into high mode.In practical,the probability of occurrence this phenomenon application is very small.This paper,on the premise of meet high criticality tasks scheduling,at the same time in order to improve the low mixed criticality tasks and system resource utilization,propose an improved algorithm based on virtual deadline scheduling algorithm.First the improved algorithm put the probability to worst-case execution time and the probability of system failure rate threshold to the task model,then based on task clustering algorithm provide the additional resources for high criticality tasks under high criticality mode.Due to only some of the high criticality simultaneous enter into the high key mode,some of the low mixed criticality tasks can still be scheduled,and no need to completely abandon.The simulation results also show that the proposed algorithm is more efficient than the classical earliest virtual deadline algorithm.