摘要
对存在的多维Fuzzing技术中使用的遗传算法不能表示多种输入类型元素,不能充分使用已得到知识从而大大降低了基于知识的多维Fuzzing技术中提出的多维Fuzzing技术挖掘的漏洞的范围和能力,设计了一个包含选择、交叉、变异、修补等操作的可以表示大多数输入元素类型的遗传算法,提出一种多个输入元素的小染色体级连成一个大染色体,大染色体的遗传算子操作分解到各个小染色体之间操作的编码及操作方案,针对字符串型输入元素,提出一套可变长染色体的实值编码及操作方法.漏洞挖掘实验结果显示应用论文设计的遗传算法的多维Fuzzing技术具有更好的漏洞挖掘能力和更好的漏洞挖掘效率.
The genetic algorithm in existing multi-dimensional Fuzzing technologies can not represent most types of input elements and can not use got knowledge thoroughly,which limit its discovered vulnerability′s scope and vulnerability mining ability,and this paper designs a genetic algorithm including select,crossover,mutate and mend operations and which can represent most input elements′ type,proposes an encoding and operation scheme where several input elements′ small chromosomes cascades a big chromosome and the operations on big chromosomes are divided into the operations on the small chromosomes,and proposes a suite of encoding and operations on variable-length chromosomes of input elements of string type.Experiment results on vulnerability mining show that the multi-dimensional Fuzzing technology which uses the proposed genetic algorithms works better than other multi-dimensional vulnerability mining ability both on the ability of vulnerability mining and the efficiency of vulnerability mining.
出处
《小型微型计算机系统》
CSCD
北大核心
2011年第5期998-1004,共7页
Journal of Chinese Computer Systems
基金
国家自然科学基金项目(60972161)资助
解放军电子工程学院博士生创新基金(CX2007016)资助
关键词
多维Fuzzing技术
遗传算法
演化测试
漏洞挖掘
multi-dimension fuzzing technology
genetic algorithm
evolutionary testing
vulnerability mining
