摘要
鉴于当前fuzzing技术发展中出现的各种策略的性能无法形式化衡量的问题,设计一种评价框架,通过基于跟踪器生成的污点数据图与解析器生成的语义树之间的映射关系,计算某种策略的效能值与熵值来对此策略性能进行量化,根据评价结果从策略集中筛选出较优策略。通过实例的测试,证明了方法的可行性。对测试目标实施最优策略将能提高对测试目标的代码覆盖率与漏洞检测能力。
Considering the lack of evaluation methods aiming to evaluate the performance of many kinds of fuzzing strategies in the developing of fuzzing techniques at present, this paper proposed an evaluation framework, which showed the mapping between tainted data graph and syntax tree based tracer developed. According to the relation of mappings, it proposed an evaluation method to calculate the power and entropy values of a strategy and the optimal one would be chosen from strategy sets. Compared with the results on the current fuzzing platform, it proves the validity and effectiveness of method. The application of optimal strategy on the SUT will improve the capability of code coverage and vulnerability detection.
出处
《计算机应用研究》
CSCD
北大核心
2013年第12期3804-3806,共3页
Application Research of Computers
基金
中国人民公安大学博士生创新基金资助项目
