Practical spectrum sensing with frequency-domain processing in cognitive radio

The imperfections of practical receivers such as nonwhite noise floor, noise power uncertainty and spurs normally have large influence on the performance and stability of spectrum sensing. However, in most of the literatures about spectrum sensing, part of or all of theses imperfections are ignored and only addictive white Gaussian noise (AWGN) is assumed in many cases. In this paper, two spectrum sensing algorithms are proposed using frequency-domain processing. The first one is a blind detection and the second one takes the known power spectrum density (PSD) of the target signal as a template. Analysis and simulations show that they are equivalent or similar to some reported spectrum sensing algorithms. The advantage of the proposed ones is that the imperfections can be easily mitigated at nearly no cost of extra complexities, which makes it more feasible to low-cost implementations.