Radar systems use radio waves to detect and track objects, playing a crucial role in modern defense, aviation and surveillance, but their effectiveness is often challenged by environmental chaos, meaning objects such as buildings, trees or ground. unnecessary signal interference radar detection. Researchers at Northwestern University, led by Professor CAI WEN and the Xi’an Space Radio Technology Institute in China have developed an innovative approach to improving radar mobile target detection. Their study, published in the peer-reviewed journal Remote Sensing, introduces a novel detection network that uses an AI AID learning method that quickly adapts to new situations and adopts a focus enhancement method to improve detection.
Due to the strong and heterogeneous chaotic echoes, traditional radar systems strive to detect moving targets in complex environments, which are reflections from non-target objects, making it difficult to identify actual moving targets. This makes it difficult to distinguish between weak signals and background noise. To solve this problem, the research team proposed a detection network that first used simulated radar data for offline training, thus reducing the need for extensive online training. Then, a small amount of real-time data, i.e., real-time, constantly updated information, is used to fine-tune the network to ensure adaptability to the reality. Professor Wen explained: “The use of small sample transfer learning allows the system to quickly adapt to new chaotic environments while maintaining high detection accuracy.”
The key innovation of this study is the integration of attention mechanisms, which helps focus on the most important parts of radar signals to improve detection in specific radar data fields, thereby aiding in the analysis of motion patterns. This mechanism helps the network prioritize basic functions, thereby improving its ability to distinguish mobile targets from background chaos. The research team conducted extensive simulations to verify its approach, showing that the attention mechanism significantly enhances debris suppression, reducing interference to unwanted signals even when the target signal is very weak compared to background noise. . Professor Wen said: “Our simulations show that attention mechanisms improve classification accuracy, the system’s ability to correctly identify targets, and even in challenging situations, the system can detect targets more efficiently.”
Compared with conventional methods, the proposed network achieves the required processing power, which is the computing power required to process large amounts of data quickly while maintaining robust detection performance. Traditional spatiotemporal adaptive processing techniques require a large number of independent training samples that are often unavailable in a variety of and unpredictable environments. The new approach reduces dependence on these samples, performs real-time detection, and can immediately identify mobile targets even without delay, and is more feasible for radar systems that propagate in the air and space.
The findings of this study pave the way for more efficient and reliable radar detection systems and use potential applications in defense, aviation and remote sensing. By combining small sample transfer learning with attention mechanisms, this approach provides a powerful alternative to existing detection methods. Future research may focus on further optimizing realistically deployed networks and extending their capabilities to different radar platforms.
Journal Reference
Zhu J., Wen C., Duan C., Wang W., Yang X. “Radar moving target detection based on small sample transfer learning and attention mechanisms.” Remote Sensing, 2024; 16: 4325. doi: https://doi.org/10.3390/rs16224325
About the Author
Professor Kane He received his bachelor’s degree from the School of Electronic Engineering at Xidian University in July 2009 and received his PhD in Engineering in December 2014 in Xidian University’s National Critical Signal Processing. From November 2019 to 2023, he served as a postdoctoral fellow in the Department of Electrical and Computer Engineering, McMaster University, Canada. He has been an assistant professor at Northwestern University’s School of Information Science and Technology since November 2016 and was promoted to associate professor in 2019.
He has led more than 10 national and provincial projects, including the National Natural Science Foundation of China and several industrial projects. He has also participated in many research projects such as the Pre-Defense Research Program, the National Basic Research Program (973 Program) and the National Key Research and Development Program. He has published more than 80 science fiction/EI index papers in top international academic journals and conferences, including IEEE TSP, IEEE TAES and IEEE TGRS. In these publications, ESI highly cites five papers, three of which are IEEE Transactions Hot Papers. He has written three academic monographs and owned 10 authorized invention patents.
Professor Cai Wen has served as conference chair and TPC member at several prominent international conferences, and has served as reviewer and team leader for several national projects. He is currently a member of the editorial board of the Naval Aeronautics and Astronautics University and Modern Radar Magazine. He is also a senior member of the China Electronics Research Institute and the China Radar Industry Association. He is the recipient of China’s “International Postdoctoral Exchange Program” and Northwestern University’s “Young Academic Talent Support Program”. His research interests focus on radar signal processing, integrated sensing and communications (ISAC) and artificial intelligence (AI).
