Robust spectrum sensing algorithm for cognitive radio networks

Energy detection (ED) is a widely used spectrum sensing scheme for cognitive radio networks since it is very easy to implement and does not require any knowledge of the primary signal. However, ED requires accurate noise power to perform the detection and its performance is very sensitive to noise uncertainty. To overcome the limitation of ED, this paper proposes a novel spectrum sensing algorithm based on maximizing the geometric mean of the individual likelihood elements. The proposed method does not need any knowledge of the primary signal or noise power, so it is a blind spectrum sensing algorithm and its performance is robust to noise uncertainty. Simulation results show that the proposed algorithm outperforms the ED with 1 dB noise uncertainty and has almost identical detection probability as the ideal ED for the same false alarm probability.