Blind Spectrum Sensing for Cognitive Radio Channels with Noise Uncertainty

In this letter, a blind spectrum sensing method is proposed which does not need any information of primary users and the noise power. For a given number of observation samples of primary channels, the spectrum sensing is reformulated into a Student's t-distribution testing problem. The analytical results of the blind spectrum sensing are given. It is shown that over flat fading channels in noise of uncertainty power the blind spectrum sensing greatly outperforms the energy detection at about 4 dB gain.