SSD_AMC_Open-set_CODE

Abstract: Increased crowding on the radio frequency spectrum has resulted in a greater risk of radar interference, creating the demand for cognitive radars that can dynamically adapt to avoid interference. A technique that benefits cognitive radar performance is automatic modulation classification, which is the task of identifying the modulation scheme used to encode received digital communications without prior knowledge. Current approaches fail to address the challenges of wideband radio frequency environments that simultaneously contain multiple transmitters and previously unseen modulation schemes. To address the issue of numerous transmitters, this research proposes a novel singleshot detector architecture for detecting and classifying radio frequency communications in a single forward pass through the model. The second issue of previously unseen modulation schemes is also addressed through the incorporation of open-set recognition. The results demonstrate that the proposed model achieves high accuracy in detection, classification, and open-set recognition. This research helps adapt automatic modulation classification to more realistic scenarios by addressing the joint detection and classification in wideband operation with previously unseen modulation schemes. In turn, this framework can improve the performance of downstream tasks such as spectrum sensing for cognitive radar