AEWACS: Airborne Early Warning and Control System Using Pulse-Doppler Radar

Abstract

This project presents the design and development of a compact Airborne Early Warning and Control System (AEWACS) using Pulse Doppler Radar for real-time drone detection and classification. Conventional AEWACS platforms used in defense systems are large and expensive, making them unsuitable for smaller and mobile applications. To address this limitation, the proposed system uses the Texas Instruments IWR6843 mmWave radar, operating in the 65 GHz frequency band, to transmit and receive electromagnetic signals. The reflected signals are analyzed using Doppler shift principles to estimate target range, velocity, and motion characteristics. Signal processing techniques are used to separate moving targets from background clutter. The processed radar data is then provided to a machine learning model trained to classify drones and birds. The system is designed to be compact, energy-efficient, and modular, allowing future integration with UAV platforms. This approach enables a cost-effective and flexible solution for applications such as border surveillance, infrastructure protection, and public safety monitoring.

Introduction

The rapid increase in the use of drones and UAVs across applications such as surveillance, agriculture, delivery, and disaster management has also raised security concerns due to their potential misuse for unauthorized surveillance and restricted-area intrusions. Traditional radar systems are not well suited for detecting small, low-flying drones, creating the need for compact and efficient radar-based detection systems. This project proposes a compact AEWACS system that employs a Pulse Doppler radar using the TI IWR6843 mmWave sensor to detect and classify aerial targets such as drones and birds. The system combines radar signal processing with machine learning to improve detection accuracy.

Previous research has explored radar-based drone detection using micro-Doppler signatures, passive radar systems, and deep learning techniques. While these methods have shown promising results, they often depend on prior knowledge, environmental conditions, or computationally intensive models. The proposed system integrates signal processing and a Convolutional Neural Network (CNN) to provide an effective and practical solution.

The system architecture consists of a waveform generator, radar transmitter and receiver, antenna arrays, and a signal processing unit. Frequency-modulated chirp signals are transmitted, reflected by aerial targets, and processed to extract target information such as range, velocity, and direction. Fast Fourier Transform (FFT) is used to identify Doppler shifts for moving targets, while Cell Averaging Constant False Alarm Rate (CA-CFAR) detection dynamically distinguishes targets from background noise, reducing false alarms.

Since real radar datasets are difficult to obtain, synthetic radar data is generated by simulating FMCW radar chirp signals, target movements, and environmental noise. The resulting in-phase and quadrature (I/Q) samples are converted into Range-Doppler maps, which serve as input to a CNN model. The CNN extracts spatial features through convolution and pooling layers before performing binary classification to distinguish drones from birds.

The system was evaluated using a synthetic dataset of 2,000 Range-Doppler images, with 1,600 images for training and 400 for testing. Experimental results achieved 86.5% accuracy for bird detection, 82.5% for drone detection, and an overall classification accuracy of 84.5%, demonstrating the effectiveness of combining radar signal processing with deep learning. A graphical user interface displays detected targets, their range, velocity, and classification results in real time. The hardware prototype, built using the TI IWR6843AOP mmWave radar module, validates the practical implementation of the proposed compact AEWACS system for real-time drone detection and monitoring.

Conclusion

This paper presented the design and implementation of a compact AEWACS system for drone detection using Pulse-Doppler radar and machine learning techniques. Synthetic radar datasets were generated to simulate aerial targets and train a CNN model for object classification. Signal processing techniques such as FFT and CFAR were used to detect moving targets and generate Range-Doppler maps. Experimental results demonstrated that the proposed system achieved an overall classification accuracy of 84.5%. The hardware prototype implemented using the TI IWR6843 radar module successfully validated the radar detection pipeline. The proposed approach demonstrates the potential of integrating radar sensing and machine learning for compact airborne surveillance systems

References

[1] B. Taha and A. Shoufan, \'Machine learning?based drone detection and classification,\' IEEE Access, 2019. [2] A. Coluccia et al., \'Drone?vs?Bird Detection Challenge,\' IEEE Open Journal of Signal Processing, 2024. [3] F. Gao et al., \'Multistatic Passive Radar for Drone Detection,\' IEEE Access, 2018.

Copyright

Copyright © 2026 Ashik Nassar, Mohammed Shemeem, Muhammed Faris, Sayed Muhammed Suhail Mujthaba, Thomas P Rajan, Rincy Raphael. 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.