Design and Implementation of a Compact Radar-Based System Using Planar Antenna Arrays for IoT applications

Abstract

This capstone project presents the design, development, and implementation of a compact, low-cost Frequency-Modulated Continuous-Wave (FMCW) radar system optimized for Internet of Things (IoT) applications and contactless gesture recognition. To overcome the privacy risks and high computational demands of camera-based systems, a fully localized radio-frequency (RF) transceiver front-end was simulated and fabricated. The hardware chain maintains a highly compact physical footprint and integrates a planar microstrip patch antenna array, a Vivaldi antenna configuration, a Low-Noise Amplifier (LNA), a Power Amplifier (PA), a frequency mixer, and an RF directional coupler. Target reflections captured by the system were processed into raw radar data cubes to extract micro-Doppler signatures, Time-Velocity Profiles, and Range-Doppler Maps (RDMs). To enable intelligent edge-sensing, a Python-based machine learning framework was developed to classify discrete human hand gestures from the micro-Doppler data, achieving a robust classification accuracy of 92% on the test set. The experimental results demonstrate a successful integration of hardware miniaturization and smart signal processing, providing a scalable architecture for future gesture-controlled IoT interfaces.