Open‐set recognition of compound jamming signal based on multi‐task multi‐label learning

In the increasingly intricate electromagnetic environment, the radar receiver may simultaneously encounter multiple intentional or unintentional jamming signals, which results in temporal and spectral overlap of received signals and forms a composite jamming signal. The nature and extent of interference contained in the received signal are often unknown, while they significantly affect the accuracy of radar detection. AnOpen‐Set Compound Jamming Signal Recognition Framework based on Multi‐Task Multi‐Label (MTML‐OCJR) is proposed. Based on the time–frequency characteristic of compound jamming signals, the proposed framework employs multi‐label classification to identify components of compound jamming signals while incorporating an unknown signal detection task into the classification process. Time–frequency image reconstruction combined with extreme value model estimation is used to detect unknown types of jamming signals, enabling simultaneous signal recognition and anomaly detection. The obtained results show that the proposed approach has superior recognition performance for composite jamming signals in closed‐set environments and high anomaly detection ability for unknown signals in open‐set environments. This method has the potential to significantly enhance the effectiveness and reliability of jamming systems in battlefield scenarios.