SoC Test Scheduling Based on Scan Chain Balance of Pair Cores

Since the test application time of a core is proportional to the length of its longest scan chain, this paper presents a technique to pair-wise match the cores in SoC to balance the scan chains of pair cores. The longest scan chain of pair cores is shorter than each longest scan chain of the single core. A division and union strategy is proposed, and by using an evolutionary algorithm named particle swarm optimization, the cores are efficiently matched, then are ordered and allocated to the test access mechanism bus to get an optimal test plan. Experimental results on ITC'02 SoC benchmarks reveal that the proposed test scheduling technique achieves the minimum test application time compared to the previous approaches that are based on the scan chain balance of single core.