Research On Scheduling For Open Real-Time Systems

Real-time systems are mainly designed to satisfy the timing requirements from the real world applications. A real-time system concerns not only the logic correctness of the computing results, but also the time when the results come out. Nowadays, real-time systems have been widely used in military and civilian industry. However, with the development of the real-time system application, 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 causes the proposition of the concept of open real-time system.In this thesis, the scheduling problems for open real-time systems are discussed. A new integrated scheduling framework for open real-time system, named InSFORT, is presented. Basing on InSFORT, adaptive real-time scheduling approaches are developed. In the environment of Linux operating system, the design and implement of InSFORT is carried out.InSFORT consists of three levels. It aims to solve the task scheduling problems in open real-time systems. This framework integrates different kinds of scheduling approaches that have some common characteristics, which enables the concurrent running in the case that different kinds of real-time and non-real-time tasks coexist. Furthermore, the quality of service problem is also discussed, and a corresponding solution is presented. Compared with the previous approaches, the advances of this framework are that it can support more kinds of tasks, and its extensibility is better.According to the adaptive scheduling requirements of the open real-time system, adaptive scheduling approaches are presented. Firstly, an approach for the adaptive scheduling of the hard real-time tasks in InSFORT is presented. Secondly, for soft real-time applications, another adaptive scheduling approach is presented. The second one is based on fussy control technology. It can control the deadline missing ratio near the expected value while sampling the applications' performance. These adaptive scheduling approaches are developed for InSFORT, with the merits that they can schedule different kinds of real-time tasks and adapt to the system's dynamic uncertainty through tuning the schedulers.If a scheduling approach would be applied in the real world, it must be implemented in an operating system. For this aim, also in order to test the theories and approaches presented, a scheme is designed for implementing InSFORT in a kind of real-time Linux kernel. This scheme is based on a double-kernel structure and basicpriority scheduling. It associates the InSFORT levels and the functions of the Linux kernel, consequently, it can support the scheduling of open real-time systems. According to the design, most of the software implementation works are done.In order to validate the InSFORT and to see its efficiency in the application, experiment schemes are designed, and then performed using the implemented system in Linux environment. The results from the experiments show that the expected performance is achieved, and hard real-time, soft real-time and non-real-time tasks in the open real-time system can be integrated and scheduled. Some other performance marks can also satisfy application requirements. Furthermore, by another simulation experiment, the adaptive scheduling approach based on fussy control is validated. The results indicate that it is effective and suitable for the open real-time systems.