Research Of Scheduling Algorithms Of Hybrid Tasks Upon Uniprocessor Environment

Due to the development of computer and atuocontrol, real-time system has been applied to avitation, astronautics and nvaigation widely.Real- time seheduling is the core question of realtime system and so on becomes a hot researeh in modern industry and military.Scheduling agriothm is the core of real-time tasks scheduling.At first,we analyze the research status of real-time scheduling algorithm of hybrid tasks, in this paper real-time tasks are categorized 2-dimentionally by arrival pattern and deadline type.Then aim at the shortage of the prevenient scheduling algorithm, we have done a certain amount of study about the two types of hybrid tasks scheduling and proposed the improved algorithm:1) When the task in accordance with the deadline for classifying types can be divided into hard deadline tasks, firm deadline tasks and soft deadline tasks. Aim at the shortage of traditional method, such as the HVF and EDV will result in some hard deadline tasks miss their deadline, a new scheduling algorithm based on the dynamic priority of the hybrid tasks is proposed. This algorithm considered the hard deadline's execution opportunity particularly, and integrate the preemption threshold of the scheduling algorithm and the execution value of the task, and other elements.Given the hard deadline, firm deadline and soft deadline tasks different priorities calculation because of their different characteristics .The simulation test shows that the algorithm can guarantee hard deadline tasks to meet the deadline, and can increase the accomplishment ratio of firm deadline tasks of the total volume.2) When the task in accordance with the arrival mode for classifying types can be divided into periodic tasks and aperiodic tasks. This paper propose an improved scheduling algorithm for hybrid tasks based on TBS(total bandwidth server) and DMS(deadline monotonic scheduling algorithm).The approach can meet the deadline requirement of periodic tasks and cut short the response time of aperiodic tasks. In this approach, we transform aperiodic tasks into periodic tasks with hard deadline by TBS at first, and then we schedule all tasks through DMS.DMS is a static scheduling approach, so it has litter overhead and low computing complexity, it also has the ability of deal with tasks over loading. It is a good approach for hybrid tasks scheduling.