MoniCrowd: A Real-World WiFi Probe Labelled Dataset with Device-Level Ground Truth

This dataset, released under the MoniCrowd project, provides a real-world and controlled collection of WiFi probe request traces designed to support research in crowd sensing, device fingerprinting, and privacy-preserving wireless analytics. The dataset combines high-fidelity labelled data collected in a controlled laboratory environment, performed inside a custom-designed Faraday cage to ensure electromagnetic isolation and eliminate external interference, enabling the generation of reliable ground truth at the device level. Each device was individually characterized, with per-device packet capture (PCAP) traces and associated metadata including brand, model, operating system, and operating system version. The dataset is intended to support reproducible research in areas such as crowd monitoring, wireless device fingerprinting, anomaly detection, and machine learning-based identification methods. To the best of our knowledge, this is one of the largest publicly available datasets combining per-device PCAP traces and multi-level device ground truth. The dataset is accompanied by processing scripts and documentation to facilitate reuse and benchmarking by the research community.