Research On The Technology Of Cooperative Spectrum Sensing Based On Non Reconstruction Compressed Sensing

Cognitive radio achieves a dynamic access of spectrum, which greatly improves the frequency spectrum utilization. Spectrum sensing is a mean of detecting spectrum holes, in which signal detection method of a single node cognitive user directly affect the final verdict. Meanwhile,undetection caused by the deep fading, can lead to that cognitive user has occupied the primary user band, thus cognitive users need high detection sensitivity. The cooperative spectrum sensing technology can merge information from multiple cognitiv e user in a specific way, it can improve the detection effect, reducing cognitive user detection sensitivity requirements.Compressed sensing apply random matrix theory to the signal measured, breaking the Nyquist sampling theorem limit, the sampling frequency can be reduced, reducing the number of samples.In this paper, compressed sensing technology is applied into cooperative spectrum sensing, which reduces both the sampling frequency and the amount of computation data,so that shortens the inspection time.Firstly, energy awareness model under compressed sampling is established, then the effects of compressed sampling and channel fading is researched. Inference was obtained based on Restricted Isometry Property: energy under compressed sampling approximately obey the chi-square distribution, but the degree of freedom of which has been reduced. Further, the effect what compressed sampling and fewer number of samples through reducing sample observation time have on detection is analysised and compared. We can discovery that the energy detection method is sensitive to the number of samples and the noise uncertainty, unable to take fully advantage of information obtained by compressed sensing.To reduce the effect of fading channel, a cooperative spectrum sensing information hard merger method based on double threshold detection is proposed,called Double Threshold Or fusion. Then, the article points its detection probability and false alarm probability, derives constraint on the optimal number of users. Theoretical studies and simulations show that the Double Threshold Or fusion both significantly reduces the false alarm probability while maintaining the high detection probability, and can effectively overcome channel fading.This article for the current problem of characteristic value detection that to determine the detection threshold,due to the use of sampling covariance matrix instead of the statistical covariance matrix, proposes a accurate threshold value setting mode adapted to a low false alarm probability requirment. Later, the relationship between the false alarm probability and the threshold is detailed derivated.Finally, characteristic value detection method combined with the compressed sensing is discussed. Probability density distribution of eigenvalues rate,cognitive users using different measurement matrix, using the same measurement matrix and using the same optimization measurement matrix, are analyzed resp ectively. The article proves the feasibility of characteristic value detection applied in compressed sensing and indicates an optimal direction of measurement matrix at the same time.