The Polar CenterNet for Radar Signal Identification
Lin Wang, Siya Mi, Ye Tian, Fengsheng Wang, Shen He · 2022
The radar signal sorting is important in the reconnaissance phase of electronic battlefield, and its performance affects the subsequent decision for the situational awareness seriously. With the wide adoption of various types of radars, the electromagnetic environment is more complex than ever. In this paper, a dynamic multi-parameter processing technique based on CenterNet is proposed. The proposed method can map the pulse description words (PDW) as dots in PDW polar diagram, then the radar pulse can be sorted by identifying the PDW dots using the improved CenterNet. The detected pulse signals are identified to prove the proposed method, and 97.74% accuracy can be achieved. Moreover, the proposed method provides a PDW visualization approach for the complex application scenarios.