Machine Learning‐Based Intrusion Detection and Quantum Cryptography in Vehicular Networks
Abin John Joseph, R. Asaletha, V. J. Manoj, R. Nishanth · International Journal of Robust and Nonlinear Control · 2025
ABSTRACT Vehicle Ad‐hoc Network is prominent in the context of cloud security during Vehicle‐to‐Vehicle communication. Recently, the automotive industry has revolutionized wireless network communication due to the integration of vehicles into the digital ecosystem. The vehicle‐based cloud computing offers different types of communications such as Vehicle‐to‐Vehicle, Vehicle‐to‐Infrastructure, and Vehicle‐to‐Device, in which the vehicles communicate with each other over sensing abilities. In the past decades, secure communication has been the major challenging task during Vehicle‐to‐Vehicle communication. To address this security issue, the proposed model named Light Gradient‐Boosting Machine Optimized Federated‐Based Quantum Key Cryptography is introduced in this research article. The data were collected from three data sources that comprised details about Global Positioning System data, vehicles safety messages, and data transmitting as well as the recipient vehicle. Then, the gathered data are transmitted into the preprocessing stage through the base station for enhancing data quality and detection ability. The machine learning‐based model named Optimized Boruta with Light Gradient Boosting Machine model is proposed for categorizing data into attacked and attack‐free for identifying types of data. The Boruta algorithm is used for selecting features in the preprocessed data, and Light Gradient‐Boosting Machine categorized the intrusions according to the relevant features. Finally, the cloud security system is designed using Federated‐Based Quantum Key Cryptography model for accessing authenticated data. The experimental evaluations are conducted based on various measures and different types of analyses for demonstrating the efficacy of the proposed model. The proposed model's experimental outcomes, such as 96.9%, 0.36 J, 98.82%, 98.5%, 97.2%, 97.85%, 1568 bytes/s, 48%, 320 time units, 1.24, and 98.2%, are attained from packet delivery ratio, energy consumption, detection accuracy, precision, sensitivity, F 1‐score, throughput, packet drop rate, network lifetime, fairness index, and security rate, respectively. The proposed model provided an efficient outcome compared to existing research articles.