Advertisement
Published in:
2018 | OriginalPaper | Chapter
ADFL: An Improved Algorithm for American Fuzzy Lop in Fuzz Testing
Authors : Chenxin Wang, Shunyao Kang
Published in: Cloud Computing and Security
Publisher: Springer International Publishing
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
Abstract
Fuzz testing is an effective software testing technology being used to find correctness problems and security issues in software. AFL (American Fuzzy Lop) is one of the most advanced fuzzy testing tools. However, it is difficult for AFL to explore deeper parts of the program. This paper proposes an improved method called ADFL for low hit branch of the tested program to solve this problem. The method first optimizes the selection strategy for seed files, and secondly generates test cases with hits and low hits at higher frequencies during the mutation phase. The experimental results show that compared with the latest version of AFL, the coverage of ADFL is significantly increased in 24 h than AFL. ADFL can cover more branches than AFL in each benchmark program and improve branch coverage of program refactoring by 19.7% and 74.5%. Moreover, ADFL can indeed find more bugs, especially for deep nested test programs.