Research On Spectrum Sensing Algorithms Based On Feature In Multi-antennas System

The rapid development of wireless communication business has brought great changes to the world.The urgent demand for wireless communication business and the exceeding scarcity of the available spectrum resource form the great contradiction.As an effective approach to improve the spectrum efficiency,cognitive radio is attracting more and more people to pay close attention to it.Cognitive radio allows the cognitive users reuse the spectrum source of the primary users under the premise that the cognitive users don't cause interference to the primary users.As the first part of cognitive radio,spectrum sensing significantly determiners the performance of the cognitive radio.After analysis of the existing spectrum sensing algorithms,it is found that algorithms which are based on the decomposition of covariance matrix have not made full use of the eigenvalues of the received signals.At the same time,the theoretical analysis of the algorithms is based on infinite sampling points,so that the algorithms have higher requirements for test time in practical application.In this paper,we apply multi-antennas system to the cognitive users,then according to the joint distribution of the received signal eigenvalues we design a variety of spectrum sensing algorithms under the background of rapid detection of small sample points.The new algorithms not only meet the requirements of detection performance,but also reduce the test time which improve the spectrum efficiency.In addition,state-of –the-art sensing methods only exploit three dimensions of the spectrum space: frequency,time and geography whereas the angle dimension,that is,spatial spectrum sensing has not been exploited well enough.In this paper,we apply the multiple signal classification(MUSIC)Angle of Arrival(AoA)estimation method into spectrum sensing.Note that MUSIC method needs to know the number of signals,which is not available in sensing scenario.Hence,we use eigenvalue weighting scheme to design the weighted multiple signal classification(WMUSIC)method which does not need the number of arriving signals.Utilizing the maximum-minimum spectrum ratio of the WMUSIC method,we finally propose a blind WMUSIC based detection algorithm.Taking the advantage of the high peak and resolution of MUSIC method,the proposed method can achieve high probability of detection as well as offer the AOA information for spectrum access,which improve the spectrum efficiency.Simulation results are presented to verify the efficiency and robustness of the proposed algorithm.