基于调和距离量子多目标进化算法的NoC测试规划优化

发布时间：2018-07-20 19:12

【摘要】：如何实现测试时间和测试功耗协同优化是目前片上网络(Network-on-Chip,NoC)测试中亟待解决的问题.提出一种基于调和距离量子多目标进化算法(Harmonic distance quantum-inspired multiobjective evolutionary algorithm,HQMEA)的NoC测试规划优化方法.采用重用NoC作为测试存取机制(Test access mechanism,TAM)的并行测试方法,对NoC中的内核进行测试,节省测试资源,提高测试效率.提出的算法在量子多目标进化算法(Quantum-inspired multiobjective evolutionary algorithm,QMEA)的基础上,采用多进制概率角编码替代二进制概率幅编码,更好的适应NoC测试规划问题;采用调和距离替代拥挤距离(Crowding distance)能更好的衡量拥挤程度;采用混沌策略动态更新旋转角,能很好地兼顾了算法的探索和发掘能力.在ITC’02test benchmarks测试集上进行对比实验,结果表明相比量子多目标进化算法,提出的算法不仅提升了算法的收敛性,而且保证了Pareto解集良好的分布性.
[Abstract]:How to realize the cooperative optimization of test time and test power consumption is an urgent problem in Network-on-ChipNOC testing. A test planning optimization method based on Harmonic distance quantum-inspired multiobjective evolutionary algorithm (HQMEA) is proposed. The test access mechanism is used as the parallel test method to test the kernel of NOC, which saves the test resources and improves the test efficiency. On the basis of Quantum-inspired multiobjective evolutionary algorithm (QMEA), the proposed algorithm uses multi-ary probability angle coding instead of binary probabilistic amplitude coding to better adapt to the NOC test planning problem. Using harmonic distance instead of crowding distance can better measure the degree of congestion, and dynamic updating of rotation angle by chaotic strategy can give good consideration to the exploration and discovery ability of the algorithm. Compared with the quantum multi-objective evolutionary algorithm, the proposed algorithm not only improves the convergence of the algorithm, but also ensures the good distribution of the Pareto solution set.
【作者单位】：西安电子科技大学机电工程学院;桂林电子科技大学电子工程与自动化学院;桂林航天工业学院;
【基金】：国家自然科学基金(61561012;21662018) 广西自然科学基金(2014GXNSFAA118370;2014GXNSFAA118393) 广西自动检测技术与仪器重点实验室(YQ16106)
【分类号】：TP18

【相似文献】