Research Multiprocessor Real-time Scheduling Policy

With the popularization and development of multi-processor, real-time systems are also more and more used in a multi-processor system environment, real-time multiprocessor task scheduling problem more and more people pay attention, real-time scheduling theory and technology has not only stay in the laboratory, but also more and more widely used in embedded systems, automation and control, and other fields, but not yet ripe for multiprocessor real-time task scheduling, still can not meet the complex multi-processor system environment. Therefore, there is a need for long-term attention and research on the subject.This paper first introduces real-time systems and real-time scheduling research background and current development status, and summarized the basic theoretical knowledge of the real-time scheduling, a comprehensive analysis of the currently more widely used single-processor and multi-processor systems, real-time scheduling algorithm, as well as mixed real-time scheduling theory and algorithms, research and analysis of two real-time scheduling method and algorithm for multiprocessor focus on scheduling theory and performance of fair scheduling algorithm based on the proportion of global scheduling(PFair) carried out research, analysis two PFair algorithm:frequent context switching and task migration, as well as scheduling time-consuming. These two issues the PFair scheduling algorithm correlation algorithm improvements, use the the of processor grouping algorithm and boundary scheduling of thinking formed a boundary fair scheduling algorithm (BFair), to verify the feasibility and effectiveness of the algorithm through experimental tests. While, the face of the current coexistence of periodic and aperiodic tasks mixture of real-time systems environment, to use improved BFair algorithms and server thinking to form hybrid real-time task scheduling strategy for multi-processor and inspection scheduling policy by the real-time task instance viable. Concluded improved algorithms and scheduling policies exist some shortcomings as well as the follow-up work in the future, and give criticism and experts teachers.