Machine Learning-Based Collection and Analysis of Embedded Systems Vulnerabilities

Abstract

The security of embedded systems is deteriorating in comparison to conventional systems due to resource limitations in memory, processing, and power. Daily publications highlight various vulnerabilities associated with these systems. While significant efforts have been made to systematize and analyze these vulnerabilities, most studies focus on specific areas within embedded systems and lack the implementation of artificial intelligence (AI). This research aims to address these gaps by utilizing support vector machine (SVM) to classify vulnerabilities sourced from the national vulnerabilities database (NVD) and specifically targeting embedded system vulnerabilities. Results indicate that seven of the top 10 common weakness enumeration (CWE) vulnerabilities in embedded systems are also present in the 2022 CWE Top 25 Most Dangerous Software Weaknesses. The findings of this study will facilitate security researchers and companies in comprehensively analyzing embedded system vulnerabilities and developing tailored solutions.