Detecting and Mitigating DDoS Attacks in 5G-V2X Networks: A Deep Learning-based Approach

In this paper, we introduce an OpenAirInterface-based testbed focused on the development of cybersecurity solutions in the context of 5G vehicle-to-everything (V2X) networks. Through our testbed, we show the ability of allocating resources dynamically according to the traffic observed from all users, including robot cars. We give special attention to distributed denial of service (DDoS) attacks, with one or more attackers overwhelming a mobility server that acknowledges each mobility decision taken by each robot car. By not being able to send such acknowledgment, the robot cars cannot move correctly and may cause an accident. In order to tackle this issue, we also propose in this paper a deep learning (DL)-based model that not only detects DDoS attacks accurately, but also triggers the creation of a sinkhole-type slice with the smallest possible amount of network resources, where all attackers are isolated and mitigated, hence allowing benign users to communicate normally with the mobility server. We showcase the efficiency of our DL-based approach by reducing the attackers' throughput by a factor of 70, while benign users remain stable with a high throughput.