A Energy Prediction Based Spectrum Sensing Approach for Cognitive Radio Networks

The spectrum scarcity is a serious problem for future development of wireless communication. The current researches on cognitive radio (CR) technologies and secondary usage of the spectrum have received a lot of attention. However, in CR networks, secondary users (SUs) should spend a large number of times for spectrum sensing. In this paper, we propose a exponential moving average (EMA) based energy prediction method to enhance the energy detection for spectrum sensing. In our algorithm, the energy prediction can predict the energy level in the frequency bands. Based on these predicted information, SUs can skip the sensing duty when the predicted energy level is larger than the preset threshold (the existence of primary user(PU)). The simulation results indicate that our method can reduce the whole sensing time of SUs effectively, as compared with the traditional energy detection method.