A nonparametric approach for spectrum sensing using bootstrap techniques

This paper deals with the blind spectrum sensing problem for arbitrary noise. The majority of current methods consider the Gaussian noise. However, this assumption cannot model the impulsive noise due to the artificial source. In this paper, we remove the requirement on Gaussianity and propose a detection method based on the bootstrap technique. By using multiple receiving antennas, the proposed detector exploits the eigenstructure of sample covariance matrix. Since there is no closed-form expression for the joint distribution of eigenvalues, the nonparametric bootstrap resampling is applied to estimate the null distribution of the test statistic. Simulation results show that the proposed detector performs well in different noise types and a performance gain can be expected when the noise is non-Gaussian.