Modeling and optimization of spectrum sensing for unknown signals based on Bartlett´s power spectrum estimator

In this paper a new and accurate performance analysis of the Bartlett method-based spectrum sensor is presented for the application of cognitive radios. To do this, first the attributes of the quadratic form representation of Bartlett power spectrum estimator are studied, and then a closed form for the cumulative distribution function (CDF) is derived and exploited to obtain the probabilities of false alarm and detection. We also investigate the case of joint detection of multiple frequencies where we apply the Neyman-Pearson approach to obtain an optimized detector for both cases of single and multiple frequency sensing. The accuracy of the presented mathematical analysis is verified by Monte-Carlo simulations.