Intelligent Radar Jamming Recognition in Open Set Environment Based on Deep Learning Networks

Jamming recognition is an essential step in radar detection and anti-jamming the complex electromagnetic environment. When radars detect unknown type of jamming that does not occur training set, existing algorithms fail to correctly recognize it. However, these can only this as one already exists our library. To address issue, we present two models for open set (OSR) accurately classify known distinguish case small samples. The OSR model based on confidence score from by assessing reliability sample output probability distribution setting thresholds. Meanwhile, OpenMax belonging all classes but also classes. Experimental results show exhibit high accuracy play a vital role sensing environments.