A People-Counting and Speed-Estimation System Using Wi-Fi Signals.

Area of Science:

• * People counting and movement analysis
• * Wireless sensing technologies
• * Crowd dynamics and flow management

Background:

• * Accurate crowd monitoring is crucial for safety and efficiency.
• * Existing methods for people counting have limitations.
• * Wi-Fi signals offer a pervasive and non-intrusive sensing modality.

Purpose of the Study:

• * To develop a Wi-Fi-based system for real-time people counting and walking speed estimation.
• * To accurately detect individuals entering and leaving a space.
• * To analyze crowd flow dynamics using wireless signals.

Main Methods:

• * Wi-Fi signal processing including subcarrier selection and Hampel filtering for outlier removal.
• * Double thresholding for detecting entry/exit events.
• * Dynamic Time Warping (DTW) for matching signals to a pre-collected database.
• * Variance thresholding for state identification and nonlinear fitting for speed calculation.

Main Results:

• * High accuracy in people counting: 100% for one person, 81-95% for multiple people in a lab.
• * Office environment accuracy: 98% for 1-2 people, up to 75% for five people.
• * Walking speed estimation accuracy: 75% for single person with error < 0.2410 m/s.

Conclusions:

• * The proposed Wi-Fi sensing system effectively counts people and estimates walking speeds.
• * The system demonstrates robust performance across different environments and crowd densities.
• * This technology has significant potential for crowd control and flow analysis.