Specific Emitter Identification for Open-Set Satellite TT${\&}$C Signals Based on Multi-Task Learning

Specific emitter identification (SEI) has attracted wide attention in the field of radar and communication counter-measures. Different from traditional SEI techniques, identifying satellite individual suffers from composite modulation (CM) of telemetry, tracking, and command (TT&C) signals, especially in an open-set scenario. This paper proposes a novel open-set high-orbit satellite recognition based on multi-task learning (MTL). Based on the in-depth analysis of CM of TT&C signals and corresponding unintentional modulation (UM) features of specific high-orbit satellite transmitters, a complex-valued multi-scale attention progressive layered extraction (CMSA-PLE) model is proposed to extract CM and UM features by constructing exclusive and shared expert modules. Additionally, an open-set recognition algorithm, namely prototype cosine OpenMax (PCOpenMax), is proposed to effectively detect unknown satellites using the Weibull model. It can, therefore, effectively classify newly authorized data. Experimental results on simulation and measured data demonstrate that the proposed CMSA-PLE enables to recognize high-orbit satellite individuals in the open-set scenario.