Reaearch On Scheduling Strategy For Open Hybrid Real-Time Systems

Real-time systems are mainly designed to satisfy the timing requirements from the real world applications. They are widely adopted in many areas, such as industrial controls, aerospace electronics, military equipments, and so on. A real-time system concerns not only the logic correctness of the computing results, but also the time when the results come out. With the development of real-time system applications, the cases that different kinds of hard real-time, soft real-time and non-real-time applications coexist in one system become more and more popular. This situation makes the application requirements become more complex and the concept of the open hybrid real-time system is propositioned, in which the problem "scheduling" is one of hotspots. At present, the system info for real-time scheduling has not been established completely. On the basis of existent results of research, this paper will choose a typical model of scheduling frameworks for open hybrid real-time systems and make some improvement and expansion in order to perfect the scheduling theory for open hybrid real-time systems.Firstly this paper analyses and summarizes the research status quo of scheduling theory for open hybrid real-time systems home and abroad. Based on that, we choose a typical scheduling scheme for open hybrid real-time systems—the two-level scheduling scheme and make improvement from three aspects:(1) Perfect scheduling strategy for non-real-time applicationsThrough analysis, we find the scheduling strategy for non-real-time applications in original hierarchical scheduling scheme is too simple: it may make real-time applications unschedulable if non-real-time applications contain non-preemptive sections. In order to avoid that, this paper proposes four pointed scheduling rules. Then by integrating the improved scheduling algorithm for non-real-time applications, we solve problems existing in non-real-time applications scheduling.(2) Improve scheduling strategy for real-time applications In two-level scheduling scheme hard and soft real-time applications are scheduled nondistinctively as the same type real-time applications, but the Quality of Service (QoS) cannot be guaranteed. This method has two flaws: The first, it can not differentiate scheduling priorities of hard and soft real-time applications and their requirements to time limit, that is to say, it neglects characteristic differences between hard real-time applications and soft ones, so adaptability of this scheme is not strong and it does not suit a more complex real-time environment. The second, the worst case execution time of soft real-time applications cannot be predicted exactly, so it is not worth while to cost much spending in order to assure all soft real-time applications not to miss their deadlines, and doing that may cause resource wasting. Therefore, improvement to the real-time scheduling strategy should follow two principles listed: one is to differentiate scheduling priorities of hard and soft real-time applications and guarantee schedulability of hard real-time applications; the other is that when scheduling soft real-time applications, we only need meet scheduling success ratio on the whole and achieve the throughput requested by users on the basis of guaranteeing schedulability of hard real-time applications.(3) Amend the original admission rulesAmendment to the original admission rules derives from improvement of scheduling strategy for non-real-time applications and real-time ones, while the improvement makes the original admission rules unsuitable. In this section we amend the original admission rules and enable them to suit the new scheduling scheme.At last we verify the improved real-time scheduling strategy from two aspects of theory and experiment. The results of proof and simulation indicate that the perfection and improvement we make to the two-level scheduling scheme have achieved objectives proposed in this paper, so that scheduling theory for open hybrid real-time systems is perfected.