Identification of Active Jamming Based on Swin Transformer Model and Splitting Features

With the continuous development of Digital Radio Frequency Memory (DRFM) technology, radar working condition is seriously threatened by various activate jamming, echo of true target will be mixed or covered by jamming. In this condition, splitting features extracted by modulating splitting code into the process of pulse compression present greatly difference between true target and jamming, and then this paper proposes a jamming identification method based on splitting feature and Swin Transformer (shifted window Transformer) neural network which can effectively distinguish the typical jamming, achieve classification task, and improve detection performance and recognition accuracy. Finally, the verification result of measured data shows that true target and jamming can be recognized perfectly.

I. Introduction

Since most anti-jamming measures are aimed at one or several specific jamming, jamming identification has become necessary under the environment of complex electromagnetic jamming. At present, the main method of jamming identification, to classify and identify, is extracting difference features of various jamming and designing reasonable classifier^[1].