Radar-Based Surveillance for Intruder Identification with Embedded AI

Abstract

Traditional surveillance systems often underperform in low-visibility or obstructed environments, posing a risk in military zones and other sensitive areas. This research introduces a robust, real-time intruder detection system combining Frequency-Modulated Continuous Wave (FMCW) radar and face recognition technology. The designed system integrates radar technology to identify motion, while a camera module is employed to carry out facial recognition. Embedded systems handle preprocessing and classification using machine learning techniques. When an unknown individual is detected, the system triggers local alerts and sends notifications via IoT platforms. The system is designed to ensure operational reliability in diverse environments, offering an efficient and cost-effective alternative to conventional surveillance.

Introduction

???? Objective

To develop an intelligent, non-intrusive, and real-time surveillance system that detects unauthorized individuals using FMCW radar and machine learning (ML), especially for low-visibility and high-security environments where traditional CCTV systems fail.

???? Problem with Traditional Surveillance

• CCTV and infrared sensors:

  • Depend on lighting and visibility

  • Prone to obstructions and privacy issues

• Growing need for systems that work in:

  • Nighttime, fog, or occluded areas

  • Sensitive zones like military bases or research labs

? Proposed Solution

Use Frequency-Modulated Continuous Wave (FMCW) radar to detect human motion by analyzing micro-Doppler signatures, combined with ML classification to:

• Identify individuals (known/unknown)

• Trigger alerts through IoT platforms (e.g., Telegram, Firebase, Blynk)

• Log and visualize data via a real-time dashboard

???? Core Technical Concepts

???? FMCW Radar

• Emits a chirp signal with linearly increasing frequency

• Measures range (R) and velocity (v) using:

  • Beat frequency (fb): related to distance

  • Doppler shift (fd): related to velocity

• Micro-Doppler signatures capture unique human movements

???? Machine Learning

• Feature extraction via STFT or wavelet transforms

• Classifiers used: SVM, KNN, CNN

• Learns to distinguish between:

  • Authorized individuals

  • Unknown intruders

???? Literature Review Highlights

?? Methodology

A. System Workflow

1. Motion Detection: FMCW radar senses human movement

2. Signal Processing: Generates spectrograms using FFT

3. Feature Extraction: Micro-Doppler-based features

4. ML Classification: Classify as known/unknown using KNN, SVM, CNN

5. Alert & Logging: Trigger buzzer + IoT alert, log events, update dashboard

B. Hardware Components

• FMCW Radar (TI IWR1443): Motion + micro-Doppler capture

• Raspberry Pi 4: Signal processing + ML inference

• ESP8266: IoT alerts via Firebase, Telegram, etc.

• Buzzer / LED: Local alerts

• UPS: 24/7 reliable operation

C. Software Stack

• Language: Python 3.8+

• Signal Processing: NumPy, SciPy, OpenCV

• ML Frameworks: TensorFlow, PyTorch, Scikit-learn

• IoT APIs: Blynk, Telegram, Firebase

• Database: SQLite / Firebase Realtime DB

• Dev Tools: VS Code, Thonny IDE

???? Testing & Results

???? Functional Flow Summary

???? Future Extension

• Integration of camera module post motion detection to:

  • Capture face images

  • Fuse radar + vision data for enhanced identity detection

Conclusion

Provides real-time, intelligent, and privacy-friendly surveillance Works in low-light or obstructed environments Is scalable and low-cost, using embedded systems Enhances safety in restricted and high-security areas

References

[1] Dey, N., Ashour, A. S., & Borra, S. (2018). Machine Learning in Bio-Signal Analysis and Diagnostic Imaging. Academic Press. [2] Texas Instruments. (2021). IWR1443BOOST: mmWave Radar Sensor Evaluation Module Datasheet. Retrieved from: https://www.ti.com/ [3] Zhang, J., & Lin, Y. (2023). A Comparative Review of Person Detection Technologies in Intelligent Security Systems. International Journal of Advanced Computer Science and Applications, 14(1), 75–84. [4] Python Software Foundation. (2023). Python 3.8 Documentation. Retrieved from: https://docs.python.org/3.8/ [5] Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., ... & Duchesnay, É. (2011). Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research, 12, 2825–2830.

Copyright

Copyright © 2025 Prakash OS, V. K. Shashanka. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.