Using Machine Learning Algorithms for Cybersecurity in CAVs

Abstract

The digital world is vulnerable to security threats, and cyber security helps mitigate these threats. Cyber security refers to the protection of data, networks, systems, applications and all types of data from cyber attacks which include viruses and various types of attacks. Autonomous and Connected Vehicle Networks (CAVs) are widely used, and because of their wireless and self-driving properties; They are highly vulnerable to previous threats.

This research studies the techniques of using artificial intelligence to protect networks of CAVs from cyberattacks. It uses machine learning algorithms to detect these attacks and compares the machine learning algorithms used for this in terms of accuracy and required operating time. The research uses the WEKA tool to make the comparison, as the experiments are carried out on a a new dataset, which is a dataset abbreviated from the KDD99 dataset.

Two machine learning algorithms, Decision Tree and Naive Bayes, were used as classification models, based on a modified training dataset of the KDD99 dataset to be suitable for CAVs. The accuracy and runtime of these two models are compared and analyzed when selecting each type of communication-based attack. The obtained results show that the decision tree model requires a shorter runtime, which is more suitable for detecting a CAV communication attack.