Abstract:The progression of innovation and technology and ease of inter-connectivity among networks has allowed us to evolve towards one of the promising areas, the Internet of Vehicles. Nowadays, modern vehicles are connected to a range of networks, including intra-vehicle networks and external networks. However, a primary challenge in the automotive industry is to make the vehicle safe and reliable; particularly with the loopholes in the existing traditional protocols, cyber-attacks on the vehicle network are rising drastically. Practically every vehicle uses the universal Controller Area Network (CAN) bus protocol for the communication between electronic control units to transmit key vehicle functionality and messages related to driver safety. The CAN bus system, although its critical significance, lacks the key feature of any protocol authentication and authorization. Resulting in compromises of CAN bus security leads to serious issues to both car and driver safety. This paper discusses the security issues of the CAN bus protocol and proposes an Intrusion Detection System (IDS) that detects known attacks on in-vehicle networks. Multiple Artificial Intelligence (AI) algorithms are employed to provide recognition of known potential cyber-attacks based on messages, timestamps, and data packets traveling through the CAN. The main objective of this paper is to accurately detect cyberattacks by considering time-series features and attack frequency. The majority of the evaluated AI algorithms, when considering attack frequency, correctly identify known attacks with remarkable accuracy of more than 99%. However, these models achieve approximately 92% to 97% accuracy when timestamps are not taken into account. Long Short Term Memory (LSTM), Xgboost, and SVC have proved to the well-performing classifiers.

Anomaly Detection System of Controller Area Network (CAN) Bus Based on Time Series Prediction

Application of Controller Area Network (CAN) Bus Anomaly Detection Based on Time Series Prediction.

A Survey of Anomaly Detection in In-Vehicle Networks

Anomaly Detection in Intra-Vehicle Networks

A Vehicle Can Bus Anomaly Detection Method for Periodic Attacks Based on the Entropy Model

Event-Triggered Interval-Based Anomaly Detection and Attack Identification Methods for an In-Vehicle Network

An Effective Abnormal Behavior Detection Approach for In-Vehicle Networks Using Feature Selection and Classification Algorithm

Application-Layer Anomaly Detection Leveraging Time-Series Physical Semantics in CAN-FD Vehicle Networks

Deep Learning-Based Anomaly Detection for Connected Autonomous Vehicles Using Spatiotemporal Information

In-Vehicle CAN Bus Tampering Attacks Detection for Connected and Autonomous Vehicles Using an Improved Isolation Forest Method

TCE-IDS: Time Interval Conditional Entropy- Based Intrusion Detection System for Automotive Controller Area Networks

Spatio-Temporal Network Traffic Estimation and Anomaly Detection Based on Convolutional Neural Network in Vehicular Ad-Hoc Networks

Intrusion detection and defence for CAN bus through frequency anomaly analysis and arbitration mechanism

Multiple Observation HMM-Based CAN Bus Intrusion Detection System for In-Vehicle Network

Attack Detection for Intelligent Vehicles Via CAN- Bus: A Lightweight Image Network Approach

Practical IDS on In-vehicle Network Against Diversified Attack Models

Attacker Identification and Intrusion Detection for In-Vehicle Networks.

Intrusion Detection System for CAN Bus In-Vehicle Network based on Machine Learning Algorithms

Classification of Normal and Malicious Traffic Based on an Ensemble of Machine Learning for a Vehicle CAN-Network

Domain Adversarial Neural Network-Based Intrusion Detection System for In-Vehicle Network Variant Attacks

An Adaptive Intrusion Detection Algorithm for In-vehicle CAN Bus Based on Periodicity of Message