Abstract
In cognitive system, spectrum sensing technique exploits unused frequency band for opportunistic wireless transmission, It is known that pilot data transmission is a necessary function in digital communication for channel realization. Then we can determine whether the primary user is on or off by detecting the existence of the pilot signal [1]. However, the received signal is always correlated when it passed through frequency selective fading channel. The widely used energy detector is seems less efficient. In the thesis, we design feature detectors based on pilot data with better performance, and discuss their performance analysis. First, pilot based receivers are proposed in order to provide the pilot information. Then in different cases depending on how we know about the pilot information, some detectors, estimator-correlator, LMP detector and eigenvalue detector, are designed. Simulation results are performed for comparison between proposed detectors and conventional methods. Finally, asymptotical behaviors of these detectors are analyzed by applying random matrix theory and Gauss quadrature rule.