Mixture Detectors for Improved Spectrum Sensing

The energy detector and the sphericity test are two widely used spectrum sensing techniques that utilize different properties of the signal received at the secondary user terminal. In this paper we use meta analysis to combine these two techniques and derive two novel mixture detectors that outperform both techniques. Since the spectrum sensing capability of the energy detector is limited by the uncertain knowledge of the noise power, first, we analyze the performance of the energy detector with estimated noise power. We derive analytical expressions for the false alarm and the detection probabilities when the secondary user terminal is equipped with multiple antennas. Next, we apply meta analysis to combine the outputs of the energy detector and the sphericity test to derive two mixture detectors, namely, Fisher's method and the weighted $z$-transform method. Furthermore, we extend our analysis to consider cooperative spectrum sensing where multiple secondary user terminals cooperatively detect the presence of primary users. Based on the mixture detectors, we propose two new cooperative spectrum sensing techniques and derive simple analytical expressions for false alarm probabilities. Extensive numerical examples are used to illustrate the accuracy of our analysis and to highlight the performance gains obtained by the mixture detectors.