DBSCAN (Density-Based Spatial Clustering of Applications with Noise) is an unsupervised clustering algorithm that groups data points based on their density in a feature space, identifying core points with sufficient neighbors within a specified radius while labeling outlier points as noise. In WiFi and CSI-based sensing, DBSCAN is particularly valuable because it does not require a predefined number of clusters, making it well-suited for dynamic scenarios such as people counting and crowd estimation where the number of distinct groups or individuals is unknown in advance. A key advantage relevant to device-free sensing is its robustness to noise and irregular cluster shapes, which arise naturally from multipath propagation and environmental variability; variants such as HDBSCAN, which extends the approach to hierarchical density levels, have also been explored to improve adaptability across varying density distributions.

Source Papers

  • A Survey on Wireless Device-free Human Sensing: Application Scenarios, Current Solutions, and Open Issues — A Survey on Wireless Device-free Human Sensing: Application
  • Channel State Information from Pure Communication to Sense and Track Human Motion: A Survey — Channel State Information from Pure Communication to Sense a
  • MMCOUNT: Stationary Crowd Counting System Based on Commodity Millimeter-Wave Radar — MMCOUNT: Stationary Crowd Counting System Based on Commodity
  • Privacy-preserving WiFi fingerprint-based people counting for crowd management — Privacy-preserving WiFi fingerprint-based people counting fo