《计算机应用》唯一官方网站

• •    下一篇

TDRFuzzer:基于自适应动态区间策略的工业控制协议模糊测试方法

宗学军1,2，韩冰1,2，王国刚1,2，宁博伟2,3，何戡1,2，连莲1,2   

  1. 1.沈阳化工大学 信息工程学院 2.辽宁省石油化工行业信息安全重点实验室（沈阳化工大学） 3.沈阳工业大学 人工智能学院
  • 收稿日期:2024-09-18 修回日期:2024-12-13 发布日期:2025-01-14 出版日期:2025-01-14
  • 通讯作者: 宗学军
  • 作者简介:宗学军(1970—)，男，辽宁沈阳人，教授，硕士，主要研究方向：工业信息安全；韩冰(1998—)，男，辽宁朝阳人，硕士研究生，主要研究方向：工业信息安全、漏洞挖掘；王国刚(1977—)，男，山东青岛人，教授，博士，主要研究方向：工业自动化优化控制、工业信息安全；宁博伟(1998—)，男，辽宁大连人，博士研究生，主要研究方向：工业信息安全；连莲(1981—)，女，辽宁丹东人，教授，博士，主要研究方向：控制理论与控制工程、工业信息安全。
  • 基金资助:
    辽宁省科学技术计划项目(2023JH1/10400082)；辽宁省人工智能创新发展计划项目(2023JH26/1030008)；辽宁省科技创新平台建设计划项目([2022]36号)；辽宁省自然科学基金资助项目(2023-MSLH-273)。

TDRFuzzer: fuzzy testing method for industrial control protocols based on adaptive dynamic interval strategy

ZONG Xuejun1,2, HAN Bing1,2, WANG Guogang1,2, NING Bowei2,3, HE Kan1,2, LIAN Lian1,2   

  1. 1.College of Information Engineering, Shenyang Chemical University 2.Liaoning Key Laboratory of Information Security for Petrochemical Industry (Shenyang Chemical University) 3.College of Artificial Intelligence，Shenyang University of Technology
  • Received:2024-09-18 Revised:2024-12-13 Online:2025-01-14 Published:2025-01-14
  • About author:ZONG Xuejun, born in 1970, M. S., professor. His research interests include industrial information security. HAN Bing, born in 1998, M. S. candidate. His research interests include industrial information security, vulnerability mining. WANG Guogang, born in 1977, Ph. D., professor. His research interests include industrial automation optimization control, industrial information security. NING Bowi, born in 1998, Ph.D. candidate. His research interests include industrial information security. LIAN Lian, born in 1981, Ph. D., professor. Her research interests include control theory, control engineering, industrial information security.
  • Supported by:
    Liaoning Provincial Science and Technology Plan Project (2023JH1/10400082); Liaoning Provincial Artificial Intelligence Innovation Development Plan Project (2023JH26/1030008); Liaoning Provincial Science and Technology Innovation Platform Construction Plan Project ([2022] No. 36); Liaoning Provincial Natural Science Foundation Project (2023-MSLH-273)

PDF

摘要： 针对模糊测试在工业控制协议应用中存在测试用例接受率(TCAR)低、多样性不足等问题，提出一种基于自适应动态区间策略的工业控制协议模糊测试方法。该方法将循环神经网络(RNN)加入Transformer的自注意力(Self-Attention)机制，构建协议特征提取模型，RNN通过滑动窗口提取数据的局部特征，输入自注意力机制进行全局特征提取，保证测试用例的接受率。注意力块间添加残差连接传递权重分数，提高计算效率。生成过程定义动态区间策略，调节模型在任意时间步的采样范围，增加测试用例的多样性。测试过程中构建字段自适应重要性函数，定位变异关键字段。基于上述方法，设计了模糊测试框架TDRFuzzer，采用Modbus TCP、S7 comm和Ethernet/IP 3种工业协议进行实验评估：相较于GANFuzzer、WGANFuzzer、PeachFuzzer 3种模型，TDRFuzzer模糊测试框架的TCAR指标显著提高，且漏洞检测率(VDR)分别提高了0.073、0.035、0.150个百分点。表明TDRFuzzer具备更强的工控协议漏洞挖掘能力。

关键词: 模糊测试, 工业控制协议, 漏洞挖掘, Transformer, 循环神经网络

Abstract: Aiming at the problems of low acceptance rate of test cases and lack of diversity in the application of fuzzy testing in industrial control protocols, a fuzzy testing method for industrial control protocols based on adaptive dynamic interval strategy is proposed. The method adds Recurrent Neural Network (RNN) to Transformer's Self-Attention mechanism to construct a protocol feature extraction model. RNN extracts local features of the data through a sliding window, and inputs the Self-Attention mechanism to carry out global feature extraction to ensure the acceptance rate of test cases. The residual connection was added between the attention blocks to transfer the weight scores and improve the computational efficiency. The generation process defines a dynamic interval strategy to adjust the sampling range of the model at any time step to increase the diversity of test cases. The testing process constructs the field adaptive importance function to locate the variant key fields. Based on the above methodology, the fuzzy test framework TDRFuzzer was designed and experimentally evaluated using three industrial protocols, Modbus TCP, S7 comm, and Ethernet/IP: compared to the three models GANFuzzer, WGANFuzzer, and PeachFuzzer, the Test Case Acceptance Rate (TCAR) of the TDRFuzzer testing framework is significantly increase, and the Vulnerability Detection Rate (VDR) increases by 0.073, 0.035, and 0.150 percentage points, respectively. This indicates that TDRFuzzer has stronger vulnerability mining capability for industrial control protocols.

Key words: fuzzing test, Industrial Control Protocol (ICP), vulnerability mining, Transformer, Recurrent Neural Network (RNN)

中图分类号: 

版权所有 © 2021 四川计算机应用杂志社有限公司
蜀ICP备 05010208 号-3　新出网证(川)字026号
技术支持：北京玛格泰克科技发展有限公司
访问总数：
今日访问：