Machine Learning-Based Malware Detection System for Securing Android-Based Internet of Vehicles

Abstract

The idea of the Internet of Vehicles (IoV) emerged out of the introduction of Internet of Things (IoT) technology into vehicular networks. Android-based platforms are commonly used in IoV due to their adaptability and smart object integration. However, because of their connectivity, vehicles are more vulnerable to threats from the internet, especially malware attacks, which may compromise passenger privacy, safety, and the car itself.

In response, a novel machine learning-based method for malware detection in Android-based IoV systems is proposed in this paper. Our method's essential elements are feature engineering, which extracts relevant data properties, and model selection, which chooses the best detection methods. We extensively test our approach on TUANDROMD datasets that reflect permission and on API-based characteristics of various malware variants to show its effectiveness.

The simulation results are cross-validated using k-fold cross-validation and compared with other machine learning systems and benchmark studies. The suggested method adds to the security and dependability of interconnected vehicle networks by providing a proactive defence mechanism against dynamic malware attacks. Results indicate significant improvements in malware detection accuracy, with an F1-score of 99.55% and a mean accuracy of 99.36% with a standard deviation of 0.006032.