信息网络安全 ›› 2025, Vol. 25 ›› Issue (4): 587-597.doi: 10.3969/j.issn.1671-1122.2025.04.007

• 专题论文:智能系统安全 • 上一篇    下一篇

面向智能系统开源模糊测试框架优化技术研究

韦超仁1, 夏万煦1, 屈刚2, 白万荣3, 杨立群4()   

  1. 1.北京航空航天大学国家卓越工程师学院,北京 100191
    2.国家电网有限公司华东分部,上海 200120
    3.国网甘肃省电力公司电力科学研究院,甘肃 734000
    4.北京航空航天大学网络空间安全学院,北京 100191
  • 收稿日期:2024-11-23 出版日期:2025-04-10 发布日期:2025-04-25
  • 通讯作者: 杨立群 lqyang@buaa.edu.cn
  • 作者简介:韦超仁(2002—),男,四川,硕士研究生,主要研究方向为模糊测试、大语言模型|夏万煦(2002—),男,重庆,博士研究生,主要研究方向为逆向工程、大语言模型|屈刚(1977—),男,新疆,高级工程师,博士,主要研究方向为网络安全及其自动化|白万荣(1985—),男,甘肃,高级工程师,硕士,主要研究方向为网络安全|杨立群(1990—),男,河北,副教授,博士,CCF会员,主要研究方向为网络信息安全、工业互联网和工控安全。
  • 基金资助:
    国家自然科学基金(62302025);国家自然科学基金(2333205);国家电网公司总部科技项目(5108-202303439A-3-2-ZN);中央高校基本科研业务费(501QYJC2024139001)

Research on the Optimization Technology of Open Source Fuzzing Framework for Intelligent Systems

WEI Chaoren1, XIA Wanxu1, QU Gang2, BAI Wanrong3, YANG Liqun4()   

  1. 1. National Superior College for Engineers, Beihang University, Beijing 100191, China
    2. East Branch of State Grid Corporation of China, Shanghai 200120, China
    3. State Grid Gansu Electric Power Research Institute, Gansu 734000, China
    4. School of Cyber Science and Technology, Beihang University, Beijing 100191, China
  • Received:2024-11-23 Online:2025-04-10 Published:2025-04-25

摘要:

随着智能系统中应用软件的普及,保障软件的安全性对提升智能系统的可靠性至关重要。现有的模糊测试技术虽然能够在一定程度上揭示软件安全缺陷,但同时也面临着测试效果差和测试效率低的问题。针对上述问题,文章提出一种基于变异敏感的模糊测试方法(Seq2Seq-Fuzzer)。首先,提出4种基于改进LSTM和Transformer的Seq2Seq模型,通过构建基于objdump、readelf等程序的字节向量数据集,对所提的模型进行训练。然后,使用Seq2Seq模型对模糊测试器AFL进行优化,预测有效的变异策略和变异位置对,解决AFL模糊测试随机性大、效率低的问题。最后,对所提的AFL优化方法进行评估。实验结果表明,在对objdump、readelf和nm的测试中,Seq2Seq-Fuzzer的代码覆盖率较AFL最高提升了56.8%,并成功发现了21个针对objdump的程序的崩溃。

关键词: 智能系统, 模糊测试, Seq2Seq, Transformer, LSTM

Abstract:

With the widespread adoption of application softwares in intelligent systems, ensuring software security is crucial for enhancing the reliability of these systems. Although existing fuzz testing techniques can reveal software security vulnerabilities to some extent, they are often hindered by issues related to testing effectiveness and efficiency. To address these challenges, this paper proposed a mutation-sensitive fuzz testing method (Seq2Seq-Fuzzer). First, we introduced four Seq2Seq models based on improved LSTM and Transformer architectures, and trained the proposed models using byte vector datasets constructed from programs such as objdump, readelf, and others. Next, we appled the Seq2Seq model to optimize american fuzzy lop (AFL) by predicting effective mutation strategies and mutation position pairs, aiming to address the high randomness and low efficiency inherent in AFL fuzz testing. Finally, we evaluated the proposed AFL optimization method. Experimental results show that, in tests on objdump, readelf, and nm, the code coverage of Seq2Seq-Fuzzer surpasses that of AFL by up to 56.8%, and it successfully identifies 21 crashes in programs related to objdump.

Key words: intelligent systems, fuzz testing, Seq2Seq, Transformer, LSTM

中图分类号: