WiFi-Sense - Person Detection & Localization

A WiFi-based passive radar system using ESP32-S3 that detects human presence, motion, and position in a room using RSSI and CSI (Channel State Information) - no additional sensors required.

How It Works

When a person moves through a room, they disturb the WiFi signal paths between the ESP32 and the router. This project measures two types of signal disturbance:

• RSSI (Received Signal Strength) - a single aggregate signal level value
• CSI (Channel State Information) - per-subcarrier amplitude/phase data (52+ data points per measurement), far more sensitive than RSSI alone

A digital signal processing pipeline extracts motion information:

Raw Signal -> Moving Average Filter -> Variance Calculation -> Integrator -> Detection Level

Hardware

• MCU: ESP32-S3-WROOM-1
• WiFi Router: Any 2.4GHz WiFi access point
• No additional sensors or hardware required

Software Requirements

• Arduino IDE 2.x with ESP32 Arduino Core v3.x
• Board: ESP32S3 Dev Module

Required Libraries (install via Arduino Library Manager):

• ESPAsyncWebServer by me-no-dev
• AsyncTCP by me-no-dev
• ArduinoJson by Benoit Blanchon

Built-in (no install needed):

• LittleFS
• WiFi
• esp_wifi (CSI API)

Quick Start

Standalone (1 ESP32-S3)

1. Open wifi-sense/wifi_sense.ino in Arduino IDE
2. Set WIFI_SSID and WIFI_PASSWORD to your router credentials
3. Select board: ESP32S3 Dev Module
4. Upload sketch
5. Upload wifi-sense/data/ folder to LittleFS:
   • Install the ESP32 LittleFS Upload Plugin
   • Arduino IDE: Tools > ESP32 Sketch Data Upload
6. Open Serial Monitor (115200 baud) to see the device IP
7. Open browser to that IP address

Place the ESP32-S3 on one side of the room and the router on the other side.

Web Interface

• Room view with presence/motion indicator, live RSSI/CSI metrics, motion history chart

Motion Detected

Empty Room

Person Detected

Support:

• Adjustable detection thresholds via sliders
• Configurable room dimensions
• Real-time WebSocket updates (~10 Hz)

Project Structure

wifi-sense/           # Single-node presence detector
  wifi_sense.ino      # Main sketch
  signal_processor.h/cpp # DSP: moving average, variance, detection
  csi_collector.h/cpp    # ESP32-S3 CSI data collection
  web_server.h/cpp       # HTTP server + WebSocket
  data/                  # Web GUI (upload to LittleFS)
    index.html
    style.css
    app.js

Tuning

Detection Sensitivity: Adjust via web UI or in code:

• motionThreshold (default 15) - higher = less sensitive to motion
• presenceThreshold (default 5) - higher = less sensitive to presence
• scanIntervalMs (default 100) - sampling rate in milliseconds

Signal Processor Config (in setup()):

processor.init(
    64,    // sample buffer size (more = smoother, slower response)
    16,    // moving average filter size
    15,    // variance threshold
    3      // variance integrator limit (samples to accumulate)
);

For best results:

• Place ESP32 and router with clear line-of-sight across the room
• Avoid placing nodes near metal objects or other 2.4GHz interference
• Allow 30-60 seconds for the signal processor to calibrate on startup
• Start with low thresholds and increase until false positives stop

CSI vs RSSI

This project uses both simultaneously. CSI is the primary detection source when available, with RSSI as fallback.

License

MIT License. Based on concepts from the MotionDetector project.