基于多目标进化算法的内建自测试（BIST）优化设计技术研究

项目名称： 基于多目标进化算法的内建自测试（BIST）优化设计技术研究

项目编号： No.60861003

项目类型： 地区科学基金项目

立项/批准年度： 2009

项目学科： 金属学与金属工艺

项目作者： 谈恩民

作者单位： 桂林电子科技大学

项目金额： 24万元

中文摘要： 内建自测试（built-in self-test,BIST)技术是解决高密集成电路测试问题的常见方法，面临着测试时间、测试功耗、矢量长度、故障覆盖率、硬件占用等诸多因素的相互制约。首先，基于多目标遗传算法和加权CA(cellular automata,细胞自动机）测试生成结构进行BIST优化设计，测试对象是ISCAS'85基准电路。验证结果表明，CA结构可以作为BIST较为理想的加权测试生成器，其规则可以采用遗传算法（NSGA-II）进行寻优，达到多目标优化设计的效果。而且由于CA结构的简洁和一致性，这种加权测试生成器在保证了矢量长度、测试功耗等指标得到优化的前提下，其带来的硬件占用率也会是比较小的。进一步地，采用遗传算法（NSGA-II）进行SoC（system on a chip,片上系统）测试的多目标优化设计，给出了优化的模型，并结合有代表性的ITC'02标准电路进行了验证，得到了有价值的二维空间描述的优化测试方案，也证明了多目标遗传算法用于SoC测试设计上的有效性。这些研究工作在技术上是一种集成创新，处于国内先进水平，对于SoC的测试具有较大的理论价值和指导意义。

中文关键词： 内建自测试；遗传算法；多目标优化；NSGA-Ⅱ#65307;SoC

英文摘要： Built-in self-test (BIST) is a popular method in solving the testability of high integrated circuits, however, it is involving of some troubles such as the co-influence of test times,test power consumption,test length,fault coverage,hardware overhead, etc. Firstly,a BIST is designed upon ISCAS'85 benchmark circuits,based on multi-objective genetic algorithm and weighted cellular automata (CA). The experimentation indicated that CA can be used as a perfect test pattern generator of BIST, to bring an optimization on those multi-objectives, with its rules optimized by the genetic algorithm (NSGA-Ⅱ. Also,for the simple architecture of CA, the hardware overhead would be smaller. Then,based on NSGA-Ⅱthe multi-objective optimizing design for system-on-a-chip (SoC)is also proposed. With the establishing of an optimization model,some valuable two-dimensional test schemes were obtained, by the experimention of some ITC'02 benchmark circuits, proving that the multi-objective genetic algorithm is effective on the test design for SoC. As a creative research, these achievements are valuable and important in the development of SoC.

英文关键词： built-in self-test;genetic algorithm;multi-objective optimizing;NSGA-ⅡSoC