
Have a safe journey with the Internet of Vehicles
Technology News
|
By
Wisse Hettinga
The Internet of Vehicles (IoV) is a network system that enables wireless communication and information exchange between vehicles and other traffic participants
Researchers from the School of Information Engineering, Henan University of Science and Technology, Luoyang in China focussed on a Intrusion Detection Method for the Internet of Vehicles.
Their conclusions:
“Network security is a key challenge facing the Internet of Vehicles today. Intrusion detection, as an important technology for defending against network attacks and protecting data security, is imperative to apply in the Internet of Vehicles.
In this paper, we propose an intrusion detection method for the Internet of Vehicles based on federated learning and a memory-augmented autoencoder. We added a memory module to the traditional autoencoder model to store the latent features of the normal behavior of the Internet of Vehicles, based on the reconstruction loss as an intrusion judgment indicator, so that various network attacks against the Internet of Vehicles can be effectively detected. In addition, our method performs federated learning training between the roadside unit and the cloud, by locally training the intrusion detection model on the roadside unit, and then uploading the model parameters to the cloud server for aggregation based on performance contribution. This method does not need to upload the Internet of Vehicles data to the cloud, avoiding the leakage of user privacy. Experimental results show that our method has higher accuracy, recall, and F1 score than the existing state-of-the-art methods, and has stronger robustness and faster training speed.
We use a memory-augmented autoencoder to model the communication behavior characteristics of the Internet of Vehicles. However, since the traffic of the Internet of Vehicles is a time series, the memory-augmented autoencoder ignores the time dependence, which may have a certain impact on the detection performance. In future research work, we will study this problem, and further extract temporal features to mine deeper information, so as to improve intrusion detection performance. At the same time, with the increasing data dimension and scale in the Internet of Vehicles, the intrusion detection model is becoming more and more complex. How to reduce the training cost of the federated learning intrusion detection model and make it more suitable for the environment with limited communication resources such as the Internet of Vehicles is also worth further research”.
Find the full report on MDPI
