V2V-SynTrust Dataset: Anomaly Detection Dataset for Secure Vehicular Communication in Intelligent Transportation Systems

This dataset contains vehicle-to-vehicle communication records for anomaly detection research. The dataset comprises vehicle identifiers and timestamps to preserve vehicle-wise temporal continuity and support leakage-proof data splits. Kinematic and spatial behaviour is captured through speed, acceleration, and GPS coordinates, which evolve continuously over time under physical constraints and may be perturbed during location-based attacks. Communication characteristics are represented by message type, message transmission frequency, packet loss rate, and received signal strength, reflecting both normal and adversarial network conditions. In addition, a synthetically generated trust score is included to emulate the output of a generic trust management mechanism and is manipulated under trust-related attacks. Each record is labelled with a multi-class attack category and a binary malicious indicator, supporting both fine-grained attack classification and binary anomaly detection. The dataset contains 50,000 samples for 17 classes, benign traffic and several attack scenarios, and has a binary labelling scheme which differentiates benign and malicious messages. Instead of being based on unconstrained random value generation, all features are generated according to predefined rules and physical constraints. These constraints provide temporal continuity for vehicle-wise, keeping feasible coupling speed, acceleration, and GPS trajectory, or perturbing thunderstorm, only for communication-and-trust-specific attributes.