Compiler Version Identification Based on Multi-Feature Fusion

Advancements in computing, information, and communication have led to iterative updates in compilers, enhancing code performance and making it more akin to natural language. These improvements have facilitated the widespread application of electronic computers but also introduced cybersecurity challenges, particularly in the software supply chain. This paper presents a novel framework for compiler classification using a radial basis function support vector machine (SVM). Key features such as register usage, opcode frequency, byte histograms, byte entropy histograms, and control flow graph (CFG) characteristics are extracted and reduced using principal component analysis (PCA). These reduced features are then fused for classification. The proposed methodology demonstrates superior performance in differentiating compiler versions. Experimental results show that the framework achieves an accuracy of 99.54%, outperforming existing methods. This high accuracy highlights the potential of the proposed approach in enhancing software supply chain security by accurately identifying compiler versions and pinpointing vulnerabilities.