Syntactic-Semantic Detection of Clone-Caused Vulnerabilities in the IoT Devices

Area of Science:

• Computer Science
• Software Engineering
• Cybersecurity

Background:

• Code cloning is a significant security risk in the development of diverse Internet of Things (IoT) devices.
• Identifying vulnerabilities arising from code duplication is crucial for enhancing IoT software security.

Purpose of the Study:

• To propose a novel clone detection method for identifying clone-caused vulnerabilities in IoT software.
• To improve the accuracy and efficiency of vulnerability detection in large-scale IoT binary code.

Main Methods:

• A hybrid approach combining syntactic and semantic code analyses.
• Construction of attributed abstract syntax trees with semantic attribute vectors.
• Encoding trees into semantic vectors using a Deep Graph Neural Network within a Siamese neural model.
• Integration of the BinDiff algorithm for automated clone candidate selection.

Main Results:

• The developed method demonstrated superior efficiency compared to existing tools like BinDiff, Gemini, and Asteria.
• The hybrid analysis effectively corrects similarity decisions, improving accuracy in clone detection.
• The method is applicable to large sets of binary code, aiding in automated vulnerability analysis.

Conclusions:

• The proposed clone detection method significantly enhances IoT software security by identifying vulnerabilities stemming from code cloning.
• The hybrid syntactic and semantic analysis, powered by Deep Graph Neural Networks, offers a robust solution for binary code analysis.
• This approach provides a more efficient and accurate means of securing the rapidly expanding landscape of smart devices.