Research On Spectrum Sensing Technology In Cognitive Radio

The rapid growth in wiles communications has contributed to a huge demand on the deployment of wireless services in the frequency spectrum. However, the fixed spectrum assignment policy enforced today results in poor spectrum utilization. Cognitive radio has emerged as a promising technology to improve the spectrum utilization dramatically, and it is the prerequisite and foundation of cognitive radio technology accurately and effectively to sensing the target spectrum.The paper is supported by Natural Science Foundation of China (60672132,60872149) and the project "the research on FH(Frequency Hopping) communication platform of wideband cognitive radio". The spectrum sensing technologies of cognitive radio are studied in detail in this paper, and the main works done by the author are listed below:(1) For the problem of the energy detection’s inability of weak signals and the likelihood ratio test (LRT)’s high computational complexity in Cognitive radio, the paper simplifies LRT algorithm under low signal-to-noise ratio and calculates the performance parameters of the simplified LRT algorithm. And then presents a spectrum joint detection algorithm based on simplified LRT, which uses energy and the simplified LRT under low signal noise ratio (SNR), and uses energy detection only under the other condition. Both theoretical analysis and simulation results show that the algorithm detects the primary user signal effectively under low SNR condition, and the algorithm has lower complexity than the well-known LRT with close performance.(2) In cognitive wireless network, the Blind covariance detection algorithms, such as MME and CAV, have the shortcoming that the performance parameters are obtained using non-asymptotic method. To deal with this problem, the paper presented a new blind detection algorithm using Cholesky factorization. The algorithm utilizes the differences between the off-diagonal elements of PU signal and noise signal after decomposition of receiving covariance matrix to distinguish PU signal and noise. According to random matrix theory, the performance parameters are derived using non-asymptotic method. The proposed method overcomes the noise uncertainty problem and performs well without informations about the channel, primary user and noise. Numerical simulation results demonstrate that the performance parameters expressions are correct and the new detector outperforms the other blind covariance detectors.(3) The auto-correlation of PU signal is very different from noise in the cognitive wireless networks, and most existing covariance blind detector utilized only the nature of elements themselves or eigenvalues without the nature of eigenvectors. The cognitive relays with multiple correlated antennas in the multiuser cooperative sensing scenario are considered in this paper. Frist, a multi-user spectrum sensing model with multiple correlated antennas is established, and only two CR users with best sensing channel are selected to participate in the detection algorithm. Then a blind detection algorithm and a semi-blind detection algorithm based on eigenvector are presented utilizing the difference of auto-correlation between the primary user signal and noise signal. The blind algorithm is based on the correlation among the main eigenvectors of the receive covariance matrices from the different cognitive relays, whereas the semi-blind algorithm is based on the correlation of the main eigenvectors between the receive covariance matrix and the known covariance matrix. And the closed expression of false alarm probability and threshold are derived in the paper. Numerical simulations show that the proposed algorithms performs better than the similar detectors.(4) Multiuser covariance blind detection algorithms in cognitive wireless network usually assume that the noise covariance is equal in different sensing channels, but In a real scenario, the noise covariance in the sensing channels must be different. Moreover, most blind detection algorithms assume that the PU has single antenna only, but there are always multi-antennas on PU in practice. The PU with multi-antennas and the sensing channels with different covariances in the multiuser cooperative sensing scenario are considered in this paper. Utilizing the main eigenvector’s nature in statistical covariance theory, BN-FTM (Feature Template Matching Based Noise) algorithm,which uses the main eigenvector as the feature template, is proposed for the complex noise scenario. The BN-FTM algorithm can detect PU signal in complex noise scenario without the priori information of PU and the sensing channel. Simulations are presented to show that the proposed algorithm is capable to achieve an improvement in detection performance.(5) As the key technology in4G mobile communications, OFDM is optimized for CR system wireless transmission technology. The research on OFDM spectrum sensing algorithm utilizes only the time diversity and frequency diversity to improve detection performance without utilizing spatial diversity. In this paper, a cooperative sensing algorithm for OFDM is proposed. this algorithm applies the OFDM feature detection algorithm to cooperative detection with utilizing the spatial diversity to enhance the system detection performance, then the system weight vector is optimized and the closed form expression of optimal weight vector is obtained using the maximum improved deflection coefficient. Theoretical analysis and simulation results demonstrate the remarkable improvement of proposed algorithm on detection performance compared with the classical cooperation detection algorithm.