Ballistic Target Recognition Based on 4-D Point Cloud Using Randomized Stepped Frequency Radar
Chuncheng Zhao, Lei Wang, Yimin Liu
Abstract
Ballistic target recognition (BTR) is critical to the ballistic missile defense system. The challenge of this task is to distinguish warheads from numerous unknown confusing targets within a short observing time. The micromotion feature is proved to be effective for this task. However, traditional methods need a long observing time to acquire enough information for the recognition because of using low-dimensional features. In addition, these model-driven methods cannot handle irregular ballistic targets, such as debris. In this article, we propose a BTR scheme, which characterizes the micromotion features with a higher dimensional representation, i.e., the time–range–velocity–power 4-D point cloud, using the randomized stepped frequency radar. The higher dimensional information contained in the 4-D point cloud can reduce the required observing time. Besides, this scheme combines the model-driven method with a data-driven deep neural network to meet the challenge of model mismatch caused by irregular targets. As a result, the proposed BTR scheme is time efficient and robust, which has been proved on an electromagnetic simulation dataset.