Intrusion Detection Framework for In-vehicle Network Combining Time Features and Data Features

In recent years, the security of in-vehicle networks has received a lot of concern. The lack of in-vehicle network security mechanisms makes the in-vehicle network extremely vulnerable to attackers. Intrusion detection systems (IDS) have proved to be an efficient way to protect in-vehicle networks. Researchers have proposed many in-vehicle network intrusion detection methods, such as deep learning-based, statistical-based, and fingerprint-based detection methods. However, their methods either require a large amount of computation and are difficult to deploy, or have poor detection ability and high detection latency. In this paper, we first proposed an improved time-interval-based detection method and then designed an intrusion detection framework based on time features and data features according to the feature analysis. We experimented proposed method to compare with the conventional time-interval-based method on the same dataset. The proposed method achieved 99.57%, 97.57%, 97.98%, and 0.977 in accuracy, precision, recall, and f1-score, respectively. Compared with the conventional method, the proposed method improved 1.6%, 12.8%, 3.5%, and 8.1% in the four metrics, respectively.