Search

As the automotive industry grows, modern vehicles will be connected to 5G networks, creating a new Vehicular-to-Everything (V2X) ecosystem. Network Slicing (NS) supports this 5G-V2X ecosystem by enabling network operators to flexibly provide dedicated logical networks addressing use case specific-requirements on top of a shared physical infrastructure. Despite its benefits, NS is highly vulnerable to privacy and security threats, which can put Connected and Automated Vehicles (CAVs) in dangerous situations. Deep Learning-based Intrusion Detection Systems (DL-based IDSs) have been proposed as the first defense line to detect and report these attacks. However, current DL-based IDSs are processing and memory-consuming, increasing security costs and jeopardizing 5G-V2X acceptance. To this end, this paper proposes a lightweight intrusion detection scheme for 5G-V2X sliced networks. Our scheme leverages DL and Knowledge Distillation (KD) for training in the cloud and offloading knowledge to slice-tailored lightweight DL models running on CAVs. Our results show that our scheme provides an optimal trade-off between detection accuracy and security overhead. Specifically, it can reduce security overhead in computation and memory complexity to more than 50% while keeping almost the same performance as heavy DL-based IDSs.

Building accurate Machine Learning (ML) attack detection models for 5G and Beyond (5GB) vehicular networks requires collaboration between Vehicle-to-Everything (V2X) nodes. However, while operating collaboratively, ensuring the ML model's security and data privacy is challenging. To this end, this article proposes a secure and privacy-preservation on-demand framework for building attack-detection ML models for 5GB vehicular networks. The proposed framework emerged from combining 5GB technologies, namely, Federated Learning (FL), blockchain, and smart contracts to ensure fair and trusted interactions between FL servers (edge nodes) with FL workers (vehicles). Moreover, it also provides an efficient consensus algorithm with an intelligent incentive mechanism to select the best FL workers that deliver highly accurate local ML models. Our experiments demonstrate that the framework achieves higher accuracy on a well-known vehicular dataset with a lower blockchain consensus time than related solutions. Specifically, our framework enhances the accuracy by 14% and decreases the consensus time, at least by 50%, compared to related works. Finally, this article discusses the framework's key challenges and potential solutions.

Deploying Connected and Automated Vehicles (CAVs) on top of 5G and Beyond networks (5GB) makes them vulnerable to increasing vectors of security and privacy attacks. In this context, a wide range of advanced machine/deep learning-based solutions have been designed to accurately detect security attacks. Specifically, supervised learning techniques have been widely applied to train attack detection models. However, the main limitation of such solutions is their inability to detect attacks different from those seen during the training phase, or new attacks, also called zero-day attacks. Moreover, training the detection model requires significant data collection and labeling, which increases the communication overhead, and raises privacy concerns. To address the aforementioned limits, we propose in this paper a novel detection mechanism that leverages the ability of the deep auto-encoder method to detect attacks relying only on the benign network traffic pattern. Using federated learning, the proposed intrusion detection system can be trained with large and diverse benign network traffic, while preserving the CAVs’ privacy, and minimizing the communication overhead. The in-depth experiment on a recent network traffic dataset shows that the proposed system achieved a high detection rate while minimizing the false positive rate, and the detection delay.