An Intrusion Detection System for Smart Autonomous E-Bikes

As the world continues to face environmental challenges, cost-effective ecological mobility solutions are sought. In this context, E-bikes have been gaining popularity. In this context, a Smart Autonomous E-Bike prototype has been implemented at our institution. This Bike can be seen as an IoT (Internet of Things) network of synchronized embedded devices and sensors/actuators that communicate and exchange data. With multiple connected modules that must function safely, securing a Smart Bike becomes crucial to avoid breaches that can cause severe consequences, such as accidents or disclosure of sensitive information. Hence, this paper presents an effective Smart Autonomous Bike Intrusion Detection System (IDS) specifically designed for smart autonomous bikes to address the aforementioned concerns. This IDS is capable of detecting and mitigating various types of cyber threats targeting different modules within the smart bike architecture, including path planning, platooning and Advanced Rider Assistance Systems (ARAS) mechanisms. The proposed system is capable of detecting both network attacks and physical tampering on the hardware components of the bike. An open-source Network-Based IDS (Snort) is first used to detect network intrusions, and an anomaly detection system is used for physical attacks. As an enhancement, the proposed IDS employs an additional hybrid Machine Learning (ML) techniques for anomaly and signature classification to analyse incoming external network traffic to detect intrusion attempts, anomalous activities, and potential security breaches. The enhanced module of the proposed IDS is trained using a comprehensive dataset (Canadian Institute for Cybersecurity Intrusion Detection System (CICIDS)-2017) and assessed using different evaluation parameters.