A Spectrum Occupancy Prediction Based on Deep Reinforcement Learning for Cognitive Radio
Biao Zhang, Lan Zhang, Tianyi Liang, Huijie Zhu · 2023
In this paper, we present an innovative method for forecasting spectrum occupancy, named the Deep Reinforcement Learning Spectrum Occupancy Prediction (DRL-SOP). This approach leverages the combined strengths of Convolutional Neural Networks (CNN) and Long Short-Term Memory (LSTM) networks, creating an advanced network capable of accurately identifying spectral features and analyzing past spectrum usage data. By processing unprocessed spectrum sensing information, we develop an intricate spectrum waterfall chart, greatly improving our predictions of how primary users (PU) occupy the spectrum. The aim of this paper is to detail the techniques used and the results of our simulations, providing a novel insight into spectrum prediction within fluctuating wireless networks.