Enhanced Mixture Detectors for Spectrum Sensing in Cognitive Radio Networks

Abstract

Since the energy detection method (ED) is suitable for detecting independent signals and easy to implement, it is widely used in spectrum sensing. While maximum eigenvalue-based detection (MED) method outperforms the ED method in correlated signals scenarios. However, both algorithms are greatly affected by the uncertainty of the noise power. To render the detection algorithms more practical, two totally-blind methods, estimated-noise-power based energy detection (ENP-ED) method and the ratio of maximum eigenvalue to trace detection (MET) method are considered in this paper. According to meta-analysis, the ENP-ED method and MET method are combined by using Fisher's method and the weighted z-transform method to derive two totally-blind mixture detection methods. Furthermore, for simplicity, we directly combine ENP-ED and sphericity test by weight to derive an improved mixture detection method. Finally, simulation results show the effectiveness of the proposed algorithms.