Post-combining based cyclostationary feature detection for cognitive radio over fading channels

Cognitive radio can improve the spectrum utilization by allowing cognitive users (CUs) to access the licensed bands. To coexist without causing harmful interference, cognitive users have to detect the presence of primary users in the vicinity. Cyclostationary feature detection (CFD) is proposed for performing this task due to its reliability and robustness. Given multiple branches of observations either from multiple receiving antennas or from cooperative CUs, pre-combining or post-combining techniques can be applied to fusing these data. In this paper, the analytical performance of post-combining based CFD subject to independent and identically distributed Rayleigh fading is investigated. We derive approximated detection probabilities in a series form for post addition combining and post selection combining. Numerical results demonstrate that the theoretical detection performance can be well approximated by our proposed in the low average signal-to-noise ratio region which is critical for cognitive radio applications.