iDetect for vulnerability detection in internet of things operating systems using machine learning.

Area of Science:

• Computer Science
• Software Engineering
• Cybersecurity

Background:

• Internet of Things (IoT) devices are increasingly prevalent, operating in diverse environments susceptible to security threats.
• Vulnerabilities in IoT Operating Systems (IoT OSs) can undermine higher-level security measures.
• Identifying and mitigating security flaws in IoT OS source code is critical for device security.

Purpose of the Study:

• To develop a Machine Learning (ML) based tool, named iDetect, for detecting vulnerabilities in C/C++ source code of IoT OSs.
• To train and evaluate ML models for identifying Common Weakness Enumeration (CWE) vulnerabilities within IoT OS codebases.
• To compare the performance of ML models against traditional Static Analysis Tools (SATs) for IoT OS vulnerability detection.

Main Methods:

• A labeled dataset was created using source code from 16 IoT OS releases and the Software Assurance Reference Dataset (SARD), annotated with CWEs.
• Three ML models—Random Forest (RF), Convolutional Neural Network (CNN), and Recurrent Neural Network (RNN)—were trained for vulnerability detection.
• The Random Forest model was selected for iDetect due to its superior performance in binary and multiclass classification tasks.

Main Results:

• Machine learning models, particularly Random Forest, demonstrated high accuracy in identifying vulnerabilities in IoT OS source code.
• iDetect achieved a macro-averaged F1 score of 98.5% for multiclass classification and 97.8% for binary classification on unseen data.
• iDetect outperformed three Static Analysis Tools (SATs) in detecting vulnerabilities within the evaluated IoT OS code snippets.

Conclusions:

• Machine Learning, specifically the Random Forest algorithm, is effective for developing automated tools like iDetect to enhance IoT OS security.
• The iDetect tool provides a robust solution for identifying C/C++ source code vulnerabilities in IoT OSs, surpassing traditional methods.
• Automated vulnerability detection in IoT OSs is essential for securing the expanding landscape of connected devices.