A Spectrum Selection Method Based on Recurrent Neural Network for OTHR

A spectrum selection method for OTHR is presented. We first establish electromagnetic environment database using the data that frequency monitoring system have observed. Then we define spectrum selection consisting of two steps: spectrum state prediction and frequency band selection. Finally, we introduce a method based on recurrent neural network (RNN) for spectrum prediction which has a good effect on the short-term fluctuations of the band status sequence. Experiment with observed data are provided to demonstrate the effectiveness of the proposed method.