A Blind Spectrum Sensing Method Based on Deep Learning.

Area of Science:

• Wireless Communications
• Signal Processing
• Artificial Intelligence

Background:

• Low utilization of spectrum resources is a significant challenge in wireless communications.
• Existing spectrum sensing methods struggle with low signal-to-noise ratios (SNR) and lack of prior licensed user signal information.

Purpose of the Study:

• To propose a blind spectrum sensing method using deep learning to enhance spectrum resource utilization.
• To address the limitations of current methods in low SNR conditions with missing prior signal information.

Main Methods:

• A novel blind spectrum sensing approach combining convolutional neural networks (CNN), long short-term memory (LSTM), and fully connected neural networks (FCNN).
• Experimental validation comparing the proposed deep learning method against traditional energy detection techniques.

Main Results:

• The proposed deep learning-based spectrum sensing method demonstrates superior performance compared to energy detectors, particularly at low SNR.
• Analysis reveals the impact of varying LSTM layers on detection performance and provides insights into the effectiveness of the deep learning approach.

Conclusions:

• The developed deep learning spectrum sensing method offers a robust solution for improving spectrum efficiency, especially in challenging low SNR environments.
• The findings highlight the potential of combining CNN, LSTM, and FCNN for advanced blind spectrum sensing applications.