The Research On Spectrum Sensing Algorithm In Cognitive Radio

The modern communication technology has made great advance and the trend accelerate the growth of wireless devices,meanwhile,scarce spectrum resources have increasingly intensified as the fixed spectrum allocation policy contributes the low efficiency of the spectrum.Cognitive radio has emerged as a promising method to solve the problem and spectrum sensing is an essential problem to monitor the primary user.In this paper,we will discuss several spectrum sensing technology in cognitive radio.Firstly,the background and related research at home and abroad will be introduced and several spectrum sensing algorithms will be shown in detail,including primary user transmitter sensing and receiver sensing.Although there are so many sensing methods,they have their own advantages and disadvantages.Secondly,focus on two classical single-node sensing algorithm including energy sensing and covariance absolute value,we discuss how to achieve cooperative sensing.Combined with unsupervised learning such as k-means clustering and Gaussian mixture model,Energy detector does not suffer the noise uncertainty problem.Next,the parameter optimization of cooperative sensing is shown while the local sensing is covariance absolute value detector.Thirdly,we improve the traditional maximum and minimum eigenvalue algorithm.On the one hand,the new threshold based on the distribution of minimum eigenvalue is introduced and on the other hand,the signals are discomposed into I and Q components.Simulation shows that the performance of the proposed method improves obviously by increasing the computational complexity.Besides the eigenvalues,eigenvectors are also utilized to propose the leading eigenvector matching algorithm.The test statistic of the method is the similarity between two leading eigenvectors and the closed-form expression of the threshold is also derived.Simulation presents that the proposed algorithm outperforms the maximum and minimum eigenvalue algorithm.Last but not least,the cooperative spectrum sensing without quiet period is introduced.Aforementioned spectrum sensing algorithms require the quiet periods and it is inevitable to decrease the quality of service.So the scheme based on the estimation of the number of the signals is introduced and simplified to reduce the computational complexity.Simulation shows that the algorithm based on the exponentially embedded family and improved Bayesian information criterion outperform the ones based on Akaike's information criterion,Bayesian information criterion and minimum description length,but the performance of these algorithms decrease while the signal noise ratio is low.