Cognitive Radio Spectrum Sensing Technology

Abstract

Spectrum sensing has important research implications for alleviating the conflict between static spectrum allocation strategies and dynamic spectrum demand. This paper provides a brief summary and comparison of some traditional detection techniques in spectrum sensing. This paper first conducts experiments on single-node spectrum sensing and discovers that the detection performance is severely impacted by the signal-to-noise ratio. For collaborative spectrum sensing, this paper first compares the traditional methods and then uses different machine learning techniques to detect the channel occupancy status based on ROC curves. The experimental results demonstrate that the machine learning-based approaches perform better in terms of channel detection.