Dynamic Task Scheduling Optimization for Multi-Core Embedded System Using Computational Intelligence

• Main Authors: Jian-Ming Chen, 陳建銘
• Other Authors: Ying- Sun Tsung
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/32729300751528652251

碩士 === 國立東華大學 === 電機工程學系 === 94 === This thesis proposes an intelligent algorithm of task scheduling for heterogeneous multi-core processor system. The algorithm can help system to well utilize the computation ability of each processor. This algorithm can be embedded in operating system kernel. Once the algorithm was involved in kernel of multi-core system, the system will distribute the task to available processors according to each processor’s loading and the computation ability for current task, so as to meet the real-time requirement. Based on the two intelligent algorithms, Genetic Algorithm (GA) and Particle Swarm Optimization (PSO), this thesis proposes the improved schemes to enhance the ability of searching optimal solution. According to the specification of both algorithms to improve their efficiency and searching ability for finding optimal solution. Finally, we verified task scheduling for Multi-core embedded system using intelligent algorithm. This thesis will analyze the computation ability of each processor so as to establish a performance index beforehand. While same data need to process, system will probe each processor’s loading and its computation ability for data type and make a judgment to partition the task, so as to integrate all processors and bring up top performance. Keyword：Heterogeneous, Multi-core Real-time Embedded System, Dynamic Task Scheduling Optimization, Intelligent Algorithm, Genetic Algorithm, Particle Swarm Optimization.