Research On Blind Sensing Algorithm Based On Covariance Detection In Cognitive Radio

With the rapid development and widespread application of communication technology and sensor technology,the available spectrum resources are increasingly exhausted.Cognitive radio technology is a method of reusing available spectrum holes,which can effectively improve the throughput of the network and improve the utilization of the spectrum.Spectrum sensing plays an important role in cognitive radio technology,requiring cognitive users to detect the status of authorized users in real time,so that authorized users can effectively use spectrum resources without interference.In this thesis,the blind perceptual algorithm is deeply studied.From the perspective of covariance perception,the perceptual algorithm for optimizing eigenvalues and the covariance matrix detection algorithm combined with goodness of fit test are proposed.The main contributions of the thesis are as follows:First,this thesis analyzes typical perceptual methods based on eigenvalues and covariance matrices.The traditional eigenvalue sensing method fails to make full use of the information of the covariance matrix eigenvalue,which limits its detection performance under low SNR.This paper proposes an optimized sensing scheme.The detection scheme constructs a new statistic based on the maximum,minimum and average eigenvalues of the covariance matrix,and the acquired eigenvalue information is more complete.Compared with the existing scheme,the proposed algorithm has better detection performance and does not increase computational complexity.Secondly,combined with the goodness-of-fit algorithm,this paper proposes an optimized covariance matrix sensing method.Compared with the eigenvalue perception algorithm,the detection technique based on covariance matrix has better perceptual performance.However,multiple approximations in the threshold solution process result in deviations between the theoretical threshold and the actual threshold,which affects its detection effect.In order to optimize this problem,this paper applies the nonparametric statistical method to the detection scheme of the covariance matrix.A simple matrix is used to form a new matrix to calculate the statistical information of the matrix elements;further,a new decision statistic is constructed and the statistical information is processed accordingly;then,the goodness of fit test is used to determine the state of the authorized user.Compared with the inherent covariance matrix sensing technology,the proposed improved covariance matrix sensing algorithm is more stable and effective,but the complexity is slightly increased...