📡 EvilRoom - Wi-Fi CSI-Based Real-Time Human Zone Detection

Detect human presence and location in real-time using Wi-Fi Channel State Information (CSI) with ESP32 and Machine Learning - no cameras or wearables required!

🎯 Overview

This project uses Channel State Information (CSI) from an ESP32 device to detect and classify human presence into four predefined zones within a room using machine learning techniques. The system operates entirely through Wi-Fi signal analysis, making it a non-intrusive privacy-friendly solution.

🧠 What is CSI?

Channel State Information (CSI) represents how Wi-Fi signals propagate from transmitter to receiver, capturing fine-grained information about the wireless channel. When humans move through the environment, they cause changes in signal propagation that can be detected and analyzed to determine location and movement patterns.

✨ Features

• ❌ No cameras or wearables - completely privacy-friendly
• 📍 Real-time zone detection with 80.5% accuracy
• 🔄 Live monitoring with terminal-based visualization
• 🤖 Machine learning powered by XGBoost with PCA and SMOTE
• 📊 Data balancing to handle uneven zone occupancy
• ⚡ Low latency predictions via serial communication

🛠️ Tech Stack

Component	Technology
Hardware	ESP32 in AP mode
Language	Python 3.8+
ML Model	XGBoost Classifier
Preprocessing	PCA + SMOTE
Communication	Serial (PySerial)
UI	Terminal-based real-time display

🗂️ Project Structure

Seefromwall/
├── ZONE1.csv                 # Training data for Zone 1
├── ZONE2.csv                 # Training data for Zone 2
├── ZONE3.csv                 # Training data for Zone 3
├── ZONE4.csv                 # Training data for Zone 4
├── trainmodel.py             # Model training script
├── live_predict.py           # Real-time prediction script
├── csi_zone_model.pkl        # Trained XGBoost model
├── csi_zone_pca.pkl          # PCA transformer
├── csi_label_encoder.pkl     # Label encoder
├── requirements.txt          # Python dependencies
└── README.md                 # Project documentation

📊 Model Performance

Our trained model achieves impressive results:

• 🎯 Accuracy: 80.5%
• 📈 F1-Score: 0.80 (macro average)

Confusion Matrix

	Pred Zone 1	Pred Zone 2	Pred Zone 3	Pred Zone 4
Zone 1	151	10	24	9
Zone 2	6	180	10	22
Zone 3	20	23	130	9
Zone 4	2	6	7	150

🚀 Quick Start

Prerequisites

• ESP32 development board
• Python 3.8 or higher
• USB cable for ESP32 connection

1. Install Dependencies

pip install -r requirements.txt

Or install manually:

pip install numpy pandas scikit-learn xgboost joblib imbalanced-learn pyserial

2. Flash ESP32 with CSI Firmware

Use a CSI-enabled firmware (ESP32 CSI tool or modified ESP-IDF example). This project uses:

https://github.com/espressif/esp-csi — specifically csi_recv_router

The ESP32 was flashed with this to capture CSI data, which was then transmitted via serial to Python.

3. Collect Training Data (Optional)

If you want to retrain the model for your specific environment:

python collect_csi.py  # Collect data for each zone

4. Train the Model

python trainmodel.py

This generates:

• csi_zone_model.pkl - Trained XGBoost classifier
• csi_zone_pca.pkl - PCA transformer
• csi_label_encoder.pkl - Label encoder

5. Run Live Prediction

python live_predict.py

📺 Live Monitoring Interface

The system provides a real-time terminal interface showing current zone detection:

## 📡 Real-Time CSI Zone Monitor

⚪ ZONE 1
🟢 ZONE 2 ← Active
⚪ ZONE 3  
⚪ ZONE 4
--------
⏱ Updated at: 01:42:01

📡 Sample CSI Data Format

CSI_DATA,0,ee:8a:c6:a9:19:f4,-67,11,1,7,0,0,1,0,0,0,1,-96,0,6,0,8162657,0,126,0,128,1,"[126,96,7,0,28,...,30,17]"

The CSI array contains 128 complex values representing the channel response across different subcarriers.

🔧 Configuration

Serial Port Configuration

Update the COM port in live_predict.py:

SERIAL_PORT = 'COM5'  # Windows
# SERIAL_PORT = '/dev/ttyUSB0'  # Linux
# SERIAL_PORT = '/dev/cu.usbserial-*'  # macOS

Zone Mapping

Modify zone definitions in your training data collection phase to match your room layout.

🧪 How It Works

1. Data Collection: ESP32 in AP mode captures CSI packets from Wi-Fi signals
2. Preprocessing: Raw CSI data is parsed into 128-element arrays
3. Feature Engineering: PCA reduces dimensionality while preserving important features
4. Data Balancing: SMOTE handles class imbalance across different zones
5. Classification: XGBoost classifier predicts the most likely zone
6. Real-time Inference: Live CSI packets are processed and displayed instantly

🔮 Future Enhancements

• 🧠 Deep Learning: Implement LSTM/CNN for temporal sequence modeling
• 👥 Multi-person Tracking: Detect and track multiple individuals
• 🌐 3D Visualization: Web-based room visualization with real-time avatars
• 📱 Mobile Dashboard: WebSocket-based mobile/web interface
• 🏠 Smart Home Integration: MQTT integration for home automation
• 🎯 Improved Accuracy: Advanced feature extraction techniques

📊 Research Applications

This project has applications in:

• Smart Buildings: Occupancy detection and space utilization
• Healthcare: Elderly monitoring and fall detection
• Security: Intrusion detection systems
• IoT: Context-aware smart home automation
• Research: Human-computer interaction studies

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request. For major changes, please open an issue first to discuss what you would like to change.

1. Fork the Project
2. Create your Feature Branch (git checkout -b feature/AmazingFeature)
3. Commit your Changes (git commit -m 'Add some AmazingFeature')
4. Push to the Branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• ESP32 community for CSI extraction tools
• scikit-learn and XGBoost developers
• Research papers on CSI-based human detection

👨‍💻 Author

Madhav Shah

---

Video PoC

🛡️ Disclaimer

This project is for educational and research purposes only. Ensure proper privacy compliance and obtain consent when deploying in real environments.

---

⭐ If you found this project helpful, please give it a star! ⭐