Researches On Spectrum Sensing Technology And Interference Detection Technology For Cognitive Radio System

With the rapid development of wireless communication, radio spectrum resources become increasingly scarce. Spectrum utilization is extremely low due to traditional spectrum allocation policy, which can not meet the rapidly growing needs of users. In order to meet increasingly strong demand for spectrum in the wireless communication, cognitive radio technology has been proposed. As a smart spectrum sharing technology, cognitive radio can sense the wireless communication environment with the support of artificial intelligence. That can solve the spectrum resources shortage problem in wireless communication effetely, and make radio communication system more intelligent. Spectrum sensing, which is an important prerequisite to achieve cognitive radio, is the key technology of cognitive radio technology. Research about cognitive radio spectrum sensing technology is addressed in this dissertation.Firstly, some spectrum sensing methods in the cognitive radio are analyzed, and then several commonly used methods are studied in detail. On this basis, we proposed novel spectrum sensing methods in single-antenna channel, multi-antenna channel and emulation attack environment respectively.In single-antenna environment, based on t-distribution and order statistic test, a blind spectrum sensing method is proposed with noise variance unknown. Numerical results show that the proposed order statistic based spectrum sensing method outperforms the Anderson-Darling spectrum sensing method.In a multiple input multiple output cognitive radio channels, a characteristic function based blind spectrum sensing method is proposed which is based on the non-parametric hypothesis testing, and that achieves the signal estimation by utilizing the standard of the whole characteristics of the signals. Empirical characteristic function of the observed sample vectors from multiple antennas is calculated, and then spectrum sensing is formulated as a characteristic function testing problem. So we can complete this testing through measuring the distance between the empirical characteristic function and the known characteristic function. Theoretical analysis and simulation results show that the proposed blind spectrum sensing method outperforms other existing spectrum sensing methods.In the presence of the primary user emulation attack, a new spectrum sensing is proposed based on asymptotic random matrix theory. The proposed algorithm achieves the signal estimation by calculating the eigenvalues of the covariance matrix of the received sample. By analyzing the covariance matrix of the received signal, we find that the eigenvalues of interference signal and spread spectrum signal are separated, so that statistics has nothing to do with interfering users, and then the performance of spectrum sensing in the presence of the primary user emulation attack environment is improved. Both theoretical analysis and simulation results show that the proposed method outperforms the traditional energy cooperative spectrum sensing.