Sensors & Transducers (Feb 2024)

Deep Learning and the Modern Radar

  • Webert MONTLOUIS

Journal volume & issue
Vol. 264, no. 1
pp. 1 – 10

Abstract

Read online

For decades, radar Systems have been essential for detecting and tracking moving targets, such as aircraft, ships, and vehicles. Their tasks increase in difficulty with the development of more complex targets. A radar system operates by transmitting a high-frequency electromagnetic signal that travels through the air and bounces back when it hits an object. By measuring the time it takes for the signal to return, the radar can determine the distance to the object and its speed. Traditionally, radar signals are processed using classical signal processing techniques, such as filtering, matched filtering, and pulse compression. These methods are generally effective but require significant computational resources, particularly when dealing with large datasets or complex signals. As a result, conventional radar systems that rely on direct sampling of Radio Frequency (RF) signals can be too computationally intensive to obtain accurate range estimates within the required time frame. To address these limitations, recent research has focused on applying machine learning techniques for radar signal processing to process the radar return at machine speed. One such approach is the use of neural networks to perform sequence-to-sequence classification on radar signals. The neural network can be trained using a dataset of radar signals, allowing it to learn patterns and features that are indicative of different types of targets. In addition, the introduction of high-speed and highly maneuvering targets necessitates that the response time of the data and signal processor be fast and optimized to meet the close look track of the system requirements. This study’s results show that using a trained neural network can significantly improve the accuracy and efficiency of radar range estimation. The neural network approach can accurately detect targets within the radar system’s field of regard and estimate their ranges with an overall accuracy of 99 %, making it a promising alternative to traditional radar signal processing techniques. The use of machine learning techniques, particularly neural networks, has the potential to revolutionize the way radar signals are processed and analyzed. The promising results of this study highlight the importance of exploring innovative approaches to radar range estimation and signal processing, particularly in applications where real-time detection and tracking of targets are critical.

Keywords