Research On Real-time Scheduling Algorithm In Mixed-Criticality System(MCS)Based On Response-time Analysis

With the rapidly growing popularity of safty-critical real-time embedded systems and applications in life,more and more criticality functions and different components are embedded on the same hardware platform,the structure of the system are more complex.The development of integration not only bring high consumption,large space,also make the system increasingly difficult to maintain.And it presents great challenges for the aviation spacecraft,nuclear reactor control systems,automotive electronic control systems and financial online trading systems which require higher safety-criticality.Not only we should ensure the safety-criticality tasks to be certified strictly,but also consider the pessimism estimate during verification,which is a waste of the system's resources.The current researches on Mixed-Criticality Systems are mostly based on the dual-criticality systems,without considering the multi-criticality systems.When the computation of the higher criticality tasks increases,the system will abandon less criticality tasks to ensure the normal execution of higher criticality tasks.It is very passive and poor practicality.To solve these problems,this paper made an in-depth study from some aspects,the main works as follows:First,the current mixed criticality tasks's scheduling model based on the response-time analysis is dual-criticality state,but the criticality of the existing mixed-criticality system is often more than two.And there is not a very accurate method about the response time analysis of multi-criticality system tasks.In this paper,we analyzed the existing response time analysis strategy AMC.based on this,we added a critical level for the system,analyzed tasks' response time of the middle state level.Similarly We could analyze the response time of the tasks between the lowest and highest criticality for multi-criticality system.we improved the AMC by extending coverage to multi-criticality system,and introduced a new response time analysis algorithm which is a good solution to this problem.Then,in traditional mixed-criticality system,when the computation of the higher criticality tasks increases,the system will abandon less criticality tasks for ensuring the normal execution of higher criticality tasks.Those less criticality tasks will not be executed until the system is initialized.Such a negative approach will not be able to ensure the Quality of Service(QoS)for some of the tasks which are less criticality but very important to the system.In this paper,we introduced a new response time analysis policy with weakly-hard constraints for multi-criticality systems.After the event of system's criticality mode increased,it rules that the losing of less criticality tasks cannot exceed a certain percentage,so as to achieve a positive handling of less criticality tasks.This algorithm is a compromise between algorithm FPPS and algorithms AMC by ensuring the Quality of Service(QoS)for some of the tasks with less criticality but high importance to the system in the event of a criticality mode change,as well as providing scalable performance trade-offs between the QoS of less criticality tasks in higher criticality mode and the schedulability of the system.