基于时间Petri网和THBA的3D NoC测试规划

Research on test planning of 3D NoC based on timed Petri net and THBA

针对三维片上网络（3D NoC）中IP核并行测试任务复杂、测试效率较低的问题,结合3D NoC测试的特点建立了一种时间Petri网模型,将变迁激发序列作为并行测试任务规划方案。为了获得最短的测试时间,设计拆分编码,在测试路径分配基础上进行顺序调度优化,采用两级递阶蝙蝠算法对变迁激发序列集进行搜索,将测试资源合理有效的分配给各IP核。仿真结果表明,所提模型可以有效地描述3D NoC测试规划问题,算法能够以较大的收敛概率迅速收敛到最优解,缩短了测试时间,提高了测试效率。

For resolving the problems of complex parallel test and low parallel efficiency of IP core in the three-dimensional network-onchip（3D NoC）,this paper proposes a timed Petri net model and planned the scheme by combining with characteristics of D NoC test.This scheme takes the transition firing sequence as a parallel test task.To obtain the shortest test time,the split coding is designed and the scheduling optimization on the basis of test path allocation is implemented.The two-level hierarch bat algorithm is also adopted to search the transition firing sequence.Then,the reasonable allocation of test resources for each IP core can be achieved.Simulation experimental results show that the proposed model can effectively describe 3D NoC test planning problems.The algorithm can converge to the optimum solution rapidly with the bigger convergence probability,which shortens the test time and improves the test efficiency.