摘要： 针对嵌入式系统中的单MPU和单ASIC体系结构问题，提出一种改进粒子群算法，将该算法应用到数字音视频解码器的软/硬件划分中，一次运行可以获得较多Pareto最优解。讨论目标函数、系统约束、粒子比较准则、拥挤距离函数、变异算子和粒子适应度等问题的处理。实验结果表明，该算法改善了传统算法产生未成熟收敛、较少Pareto最优解和Pareto最优解前端分布不均匀的问题，增强算法的自适应性及结果的全局最优性。

关键词: 嵌入式系统, 软/硬件划分, 软硬件协同设计, 粒子群优化算法, Pareto最优解

Abstract: Aiming at the configuration problems of single MPU and ASIC in embedded system, this paper proposes an improved Particle Swarm Optimization(PSO) which is applied to HardWare/SoftWare(HW/SW) partitioning of digital audio/video decoder, each run of the algorithm can produce many Pareto-optional solutions, and the problems of target function, system constraint, Particle comparison criterion, congestion distance function, mutate operator and particle Fitness are discussed. Experimental result shows that the algorithm improves immature convergence, less Pareto-optimal solutions and Front-end distributed heterogeneous of Pareto-optimal solutions of traditional algorithm, strengthens adaptability and global optimality of results.

Key words: embedded system, HardWare/SoftWare(HW/SW) partition, HW/SW co-design, Particle Swarm Optimization(PSO) algorithm, Pareto-optimal solutions

中图分类号: 

• TP302