Few-Shot Open-Set Modulation Recognition Based on Signal Constellation and Meta-Learning

To address existing limitations in data scarcity and open-set scenarios in Automatic Modulation Recognition (AMR), this work introduces an approach that applies few-shot meta-learning techniques. This method effectively utilizes signal constellation information in conjunction with meta-learning principles, aiming to enhance signal modulation classification and adeptly handle challenges posed by unknown modulation types. The proposed model exhibits impressive results in few-shot open-set AMR scenarios, thereby bridging a significant research gap in the field.