Research Of Minimizing Energy Consumption Scheduling Algorithm For Parallel Real-Time Tasks In Multicore Systems

With the development of embedded technology and the mobile computing, control computing needs, embedded real-time operating systems are widely used in consumer electronics, communication equipments, medical equipments and industrial control applications.As the embedded equipment is strictly restricted by the size, weight, the contradiction between the boosting requirements system performance and the slow progress of technologies of batteries and heat dissipation performance becomes increasingly prominent. Therefore, the energy-efficient scheduling for embedded real-time systems has become an important research topic in the embedded computing field.Because of the unique advantages of multi-core in the ratio of performance to energy consumption, parallel processing, using multi-core architecture for embedded systems is an important development direction. At the same time the application of chip multi-core DVS technology presents new challenges to the energy-efficient scheduling problem for parallel real-time tasks.Firstly, we develop rigid parallel real-time tasks scheduling problem based on multi-core architecture, comparison efficiency of classical level-packing scheduling strategy and Bottom-Left algorithm. We point out that the BL algorithm is more powerful to meet the real-time scheduling and and improving saving energy consumption. On this basis, we present two kinds of polynomial complexity two level frequency set algorithm in interlayer and inside-layer, purpose to ensure this real-time task scheduling and get the minimum energy consumption.Secondly, we study the moldable parallel real-time tasks scheduling problem for multi-core platforms. Through the observation of some task scheduling instances, we prove three theorems about the scheduling performance. We then consider the EDF algorithm in moldable parallel real-time tasks scheduling, and present two frequency assignment algorithms, to solve energy consumption optimization problem.Finally, we present two linear programming algorithms for the above energy minimization problems, to compare with the proposed heuristic algorithms. We conduct extensive simulation experiments and the results show that the excellent performance and high flexibility of the proposed heuristics.