Connected Vehicle Co-Simulation for Autonomous Vehicles in Airsim using Ns-3

Vehicle-to-everything (V2X) communications enables vehicles to communicate directly with each other, as well as roadside infrastructure. Vehicle-to-vehicle (V2V) communication, a subset of V2X communication, enables the vehicle to not solely rely on on-board sensors and allows the vehicle to share information directly to any nearby vehicles. Information shared between vehicles may include a vehicle\u27s position, velocity, and direction, as well as other data. As these are safety-critical applications, rigorous security assessments are needed, yet it can be very expensive, dangerous, and complex to test security vulnerabilities of autonomous vehicles. Therefore, we aim to leverage realistic open-sourced simulators to carry out testing for multiple features, such as security attacks as well as cooperative autonomous driving algorithms. Since there is no open-sourced simulator capable of visually and physically simulating a vehicle and accurately representing its network, this thesis aims to combine a vehicle simulator and network simulator in real-time. Specifically, we incorporate Network Simulator 3 (Ns-3) and Unreal Engine\u27s plugin, Airsim. To run this type of simulation accurately requires high computation power and time, and these requirements can cause delays between the two simulators. To handle the delays during simulation, we propose a system using a time step synchronization technique to pair Airsim and Ns-3. We further elaborate on our incorporation of delaying network packets that arrive earlier than the ideal packet delay. Additionally, we validate our system by demonstrating proof-of-concept attacks. Specifically, we simulate a replay attack and a jamming attack on our system, as well as show that a sybil attack is possible