Research And Implement Of An Adaptive Dynamic Real-time Scheduling

The key problem of real-time system is to design an efficient task schedule service which guarantee accomplishment of real-time tasks before the deadline.In hard real-time system whose real-time task character is fixed, classic schedule arithmetic is widely used for it ensure completion of real-time tasks in certain predictable time through schedulable analysis. In an overload soft real-time system, a certain range of job lost is tolerated by tasks. Soft real-time system should guarantee the timeliness and the accuracy of real-time tasks, at the same time it should enhance the system throughout. So a well designed suitable schedule model is becomes more and more important.Elastic schedule model with dynamic reflection which is a good research progress on soft real-time schedule divide schedule service into two parts: Upper-level Schedule for soft real-time application with an adjustable load change QoS(Quality of Service) of real-time tasks according to reflective running information; Underlay-level Schedule which responsible for real-time tasks sorting according to their priorities that reflect the relative importance.It is very important to design a schedule service model which has an ability to select appropriate schedule arithmetic according to running environment for soft real-time system with variable running environment such as streaming media system. Adaptive dynamic schedule model that meet the requirement above has two advantages as following. Firstly, schedule service selector, schedule arithmetic and tasks management are designed independently by applying IoC/DI(Invert of Control/Dependence Inject) pattern. It is configuration files that used to organize different parts of the system make schedule arithmetic and tasks management becomes configurable. Secondly, The Separation of schedule service and specific schedule arithmetic make schedule arithmetic can be changed in runtime.