Multi-Classifier Fusion for Open-Set Specific Emitter Identification

To safeguard the privacy and security of IoT systems, specific emitter identification is utilized to recognize device identity with hardware characteristics. In view growing demand for identifying unknown devices, this paper aims discuss open-set identification. We firstly build up a problem formulation SEI by discussing working mechanisms radio signals recognition. And then it pointed out that feature coincidence an intractable challenge in SEI. The reason, accounting phenome, pretrained fingerprint extractors are incapable clustering features differentiating them from known ones. Considering leads error recognition we propose fuse multi-classifiers decision layer improve accuracy recall. Three distinct inputs four different fusion methods adopted implement multi-classifier fusion. datasets collected at Huanghua Airport demonstrate proposed method can avoid space achieve higher