MisbehaviorX: Comprehensive V2X Misbehavior Detection Dataset Enabled by the V2X Application Spoofing Platform

This dataset contains a comprehensive V2X misbehavior dataset simulated using VASP, an open-source framework. VASP allows the simulation of diverse types of V2X attacks and works as a sub-module for Veins, a well-established open-source framework for running vehicular network simulations. Veins runs on an event-based network simulator OMNeT ++, and road traffic simulator SUMO. Data are collected from the Boston traffic network, which is a good candidate to represent real-world traffic mobility. We run VASP simulation for 3,000 simulated seconds to collect benign traces without any attacks. Such simulation provided us with 1,018,098 benign BSMs from 475 different vehicles. Similarly, we ran a VASP simulation for 1360 simulated seconds to collect malicious traces with 68 distinct attacks. While running the attack, we selected the attack policy as persistent, where the attacker vehicle always transmits attack messages and 25% malicious vehicles.

本数据集包含了一套全面的V2X恶意行为数据集，该数据集通过开源框架VASP进行模拟。VASP框架能够模拟多种类型的V2X攻击，并作为Veins框架的子模块运行，Veins框架是一个成熟的开源车辆网络仿真平台。Veins框架运行于基于事件的网络仿真器OMNeT++和道路交通仿真器SUMO之上。数据源自波士顿交通网络，该网络是代表现实世界交通流动性的理想选择。我们对VASP进行了3000秒的仿真模拟，以收集无攻击的良性痕迹，从而获得了来自475辆不同车辆的1,018,098个良性BSMs（基本安全消息）。同样地，我们进行了1360秒的VASP仿真模拟，以收集包含68种不同攻击的恶意痕迹。在执行攻击过程中，我们选择了持续性的攻击策略，即攻击车辆始终传输攻击消息，且恶意车辆占总车辆数的25%。