A Novel Detection Approach of Unknown Cyber-Attacks for Intra-Vehicle Networks Using Recurrence Plots and Neural Networks

Abstract

Proliferation of connected services in modern vehicles could make them vulnerable to a wide range of cyber-attacks through intra-vehicle networks that connect various vehicle systems. Designers usually equip vehicles with predesigned counter-measures, but these may not be effective against novel cyber-attacks. Intrusion Detection Systems (IDSs) serve as an additional layer of defence when conventional measures that are implemented by the designers fail. Several intrusion detection techniques, such as rule-based IDS, have been proposed in the literature but these techniques have limited capability in detecting novel cyber-attacks. This paper proposes a new Machine Learning (ML)-based IDS for detecting novel cyber-attacks in intra-vehicle networks, specifically in Controller Area Networks (CANs). The proposed IDS generates high-level representations of CAN messages transmitted on the bus exploiting their temporal properties as well as the intra and inter message dependencies through the use of Recurrence Plot (RP), which are then fed into a bespoke Neural Network, designed and trained to detect novel intrusions. Evaluations of the performance of the proposed IDS in comparison with that of the state-of-the-art existing IDS schemes namely Decision Tree (DT), Random Forest (RF) and other well-known approaches demonstrate the superiority of the proposed IDS in this paper.

CF-AIDS: Comprehensive Frequency-Agnostic Intrusion Detection System on In-Vehicle Network

“…A DNN-based system proficient in capturing attacks was then used. Similarly, Omar et al [17] have introduced two DL model-based IDS for CAN, utilizing two representations of CAN data:) raw data, and converted raw data in the form of images. These representations incorporate deep learning techniques, including LSTM and ConvLSTM, along with extensive CAN features.…”

“…However, challenges persist in feature extraction and data input for deep learning models. Most studies directly feed raw data into these models [17], [18], potentially limiting their scopes. This approach, where data are sequentially inputted one by one, can overwhelm deep-learning IDSs in complex scenarios, especially given the extensive data generated in in-vehicle networks.…”

Fed-Deep CTRL (Light GBM)-Scalable, Interpretable, and Privacy-Preserving Federated Learning Framework for Wireless Intrusion Detection System

“…Some of the papers that look at network intrusion detection systems using the idea of horizontal federated learning are (Seo et al, 2021;Al-Jarrah et al, 2023;Li et al, 2020, Preuveneers et al, 2018. Upon understanding the various works using machine learning and deep learning approaches, the accuracy results for classification were higher, however, they failed to address the problem of training overhead, scalability of the decentralized training model, and the classification happening considering the analysis of the traffic.…”

Automotive Electric Power Steering Control With Robust Observer Based Neuroadaptive Type-2 Radial Basis Function Methodology

“…Considering the global characteristics of the function approximators such as neural networks (NNs) [14], [15] fuzzy logic systems (FLSs) [16], [17], and fuzzy NNs (FNNs) [18], [19], the adaptive approximation-based methods are greatly capable of managing the operational uncertainties in nonlinear systems [20]. By learning to approximate the nonlinear functions, the NNs, FLSs and FNNs can effectively enhance the control performance against system uncertainties [21], [22].…”