Creating a realistic sybil attack dataset for inter-vehicle communication

Abstract

Sybil attacks, where malicious nodes use multiple identities, pose a critical threat to Vehicular Ad-Hoc Networks (VANETs). While Received Signal Strength Indicator (RSSI)-based detection offers a promising approach, existing research is hampered by the lack of realistic datasets that capture the complexities of real-world vehicular environments. Current datasets often simplify traffic conditions, neglecting the impact of congestion and packet drops on RSSI readings. This work addresses this gap by creating a large-scale, realistic VANET simulation environment for generating RSSI data. Using OpenStreetMap data for a region of Istanbul, we model realistic traffic flows calibrated with municipal data, incorporating packet loss due to congestion and interference. Critically, we employ a carefully calibrated log-normal shadowing model to capture the impact of environmental factors on RSSI. We implement four common Sybil attack methods with various power control strategies, creating a diverse set of attack scenarios. Rigorous validation against real-world traffic patterns, RSSI characteristics, attack distribution, and packet collisions demonstrates the realism of our dataset. We provide a sample dataset, along with our open-source simulation environment, enabling researchers to develop and evaluate robust Sybil attack detection mechanisms for real-world VANETs