Spectrum Situation Completion Based on Model-Enhanced Generative Learning

Three-dimensional (3D) spectrum situation can be exploited to address the problem of spectrum resource under-utilization in integrated space and terrestrial information networks. In order to improve the accuracy of 3D spectrum situation completion under limited unmanned aerial vehicle (UAV) trajectory, this paper proposes a 3D spectrum situation completion scheme based on model-enhanced generative learning, which effectively solves the problem of unknown prior information of the complete spectrum situation and effectively digs the environmental characteristics of the 3D electromagnetic spectrum space. Furthermore, this paper proposes an improved generative adversarial networks structure and a series of data processing methods to reduce the scheme's completion error and training time. Simulation results demonstrate that the 3D spectrum situation completion scheme proposed in this paper can significantly outperform the conventional interpolation-based algorithm in terms of completion accuracy.