Secure Handover in 5G-V2X: Detecting and Localizing DoS Attack to Ensure Reliable Communication

A fifth-generation vehicle-to-everything (5G-V2X) communication network has been created by combining the capabilities of vehicular networks and 5G technologies. This advanced network provides assurance of increased comfort, reliability, and security for vehicle user data. However, because of the high mobility, one of the major difficulties in 5G-V2X is the Handover (HO) process when a vehicle switches from one base station to another. This procedure is vulnerable to a variety of attacks, including impersonation, DoS, and Man in the Middle attacks. This study focuses specifically on the impersonation attack during the HO procedure, which can lead to critical DoS situations with potential real-life implications. The attack scenario involves rogue Femto Access Points (FAPs) exploiting the HO event. In this scenario, a malicious vehicle poses as a legitimate FAP, disrupting communication for nearby vehicles. To mitigate this attack, a novel attack detection model is proposed, relying on metrics like HO Failure Ratio (HFR) and Radio Link Failure (RLF). The adverse effects of this attack are demonstrated within the 5G-V2X context through measurable parameters such as packet loss rate, throughput, block rate, and HFR. Regarding the localization aspect, an ML-based multiclass classification mechanism is adopted. This mechanism leverages features particularly relevant to the handover scenario. A variety of classification methods, including k-Nearest Neighbor (kNN), Decision Tree (DT), Random Forest (RF), Logistic Regression (LR), and Naive Bayes (NB), are employed for localization purposes. To enhance localization accuracy, an Ensemble Learning approach is implemented by involving the integration of KNN, LR, and RF techniques. Overall, the proposed approach effectively detects and localizes attackers during HO procedures, and also outperforms existing approaches.