Optimization Strategy for Spectrum Sensing in Cognitive Radio Networks Using Deep Learning Algorithms

As communication technology advances quickly, spectrum resources are becoming more and more limited, and cognitive radio (CR) network spectrum sensing technology has problems in balancing perception performance, perception effect, and complexity. Therefore, this article used deep learning algorithms to study the spectrum sensing of CR networks. Firstly, a Convolutional Neural Network (CNN) was used to construct a spectrum sensing model, which learned and extracted useful features from various spectrum data to better understand and predict the state of spectrum sensing. Then, the Q-learning algorithm was used to study the collaborative spectrum sensing strategy, determine the weight factors of each indicator in the spectrum fusion process, and help the spectrum sensing model run better. Experiments have shown that when the signal-to-noise ratio was between -15dB and -10dB, the detection probability of CNN can already reach over 80%. Moreover, as the number of statistics increases, the overall blocking probability of the perception strategy studied in this paper can ultimately be reduced to around 15.5%, which was much lower than the spectrum sensing strategies constructed by other methods. Deep learning algorithms can be used to study the spectrum sensing strategy of CR networks. By leveraging the powerful capabilities of deep learning models, comprehensive utilization of spectrum resources can be achieved, effectively improving spectrum utilization efficiency.