Research On Adaptive Scheduling Algorithm On Real-Time Embedded Operating System

With the high-speed development of current numerical information technology and network, embedded systems have extensive penetration of scientific research, engineering design, military technology, various industries and commerce, culture, art and people's daily life, and so on. With the unceasing expansion of embedded system application, the function strengthening,various application areas of embedded systems increasingly high demands,so real-time operating system is gradually applied to the embedded system.Real-time embedded system requires to complete functional operation timely and correctly,and these tasks may also contradict to access shared resources. In addition, the processing capacity and memory capacity of the system is often limited, which results in the stringent requirements to this scheduling algorithm. The development process of actual real-time embedded systems is often ignored from the running perspective using mature scheduling theory, but, the existing real-time scheduling theory is too complex and not suitable for direct application in embedded systems.In response to these needs, this dissertation has studied existing classic real-time scheduling algorithms and traditional adaptive scheduling algorithms, and analyzed them applied to embedded systems inadequately, and we take the idea of automatic control theory into the scheduling algorithm from the task of the characteristics of the control system.This dissertation presents the automatic strategy of periodic and non-periodic,I/O consuming and processor consuming tasks. We take RM scheduling algorithm for periodic tasks,and take FIFO scheduling algorithm for non-periodic tasks. I/O consuming task can often run , but,generally it run only in a short ,because of it being obstructed when awaiting more I/O request. For such tasks, we use scheduling strategy to raise appropriately their level of priority, and the extension of its running time. But,processor consuming tasks spend most of their time on the implementation of code, and for such tasks,we take scheduling strategy to minimize their operational priorities and to reduce its running time. From these two aspects of the mission of the corresponding feedback control, we make the scheduler with a certain degree of adaptive function.Finally, with analysing data structure of the scheduling ofμC/OS-II kernel, we have realized the adaptive scheduling algorithm in it, and mainly realized the controller and basic scheduling. The police compared the forged signature with the original. We have compared improved algorithm with original algorithm by simulation results, and validated feasibility in the scheduling model for real-time embedded systems. The system has a good prospect in the field of real-time embedded system because of its main function features.