Novel autocorrelation based spectrum sensing methods for cognitive radios

In cognitive radios, energy detector is often considered for spectrum sensing in the literature. However, its performance deteriorates rapidly when noise power is fluctuating. In order to solve this problem, several autocorrelation based detection methods such as statistical covariances based (CAV) detection method and eigenvalue based (MME) detection method have been proposed. However, CAV detector is derived based on the assumption the noise is real and MME detector on the other hand is a little conservative. In this paper, two novel autocorrelation based spectrum sensing methods, which are correlation coefficients (CCE) based detection method and nonparametric autocorrelation (NAC) based detection method, have been proposed. CCE detector is suitable for complex Gaussian noise and NAC detector can distinguish correlated signals from arbitrary independent noise without knowing the noise type. Both CCE detector and NAC detector are also nonparametric. Simulation experiments are provided to show the validity of the proposed CCE detector and NAC detector.