中文摘要：

针对基于可重用组件的MPSoC软硬件划分问题,提出了一种采用自动波竞争神经网络的优化算法。先将软硬件划分问题转化为图论中的多约束最短路径问题,然后重新设计神经网络中的自动波机制,从组件库中为系统中的每个任务模块选择合适的软件构件或IP核,在系统成本和实时性约束下,使得MPSoC功耗最优。该算法具有并行化、无参数、易于硬件实现的特点,可获得MPSoC软硬件划分问题的最优解。

英文摘要：

A hardware-software partitioning algorithm using autowave competition neural networks is proposed for MPSoCs based on reusable components.The partitioning problem is formulated as a multiconstrained shortest path problem.The autowaves are designed specially to allocate a software component or IP core to each task module from a components library,such that the power consumption of the MPSoC is minimized subject to cost and timing constraints.This algorithm could obtain the globally optimal solution,and it is parallel,non parameter.and easy to implement by VLSI.