Research On Eigenvalue Based Spectrum Sensing Algorithms For Cognitive Radio

The fifth generation(5G)mobile communication system will face new requirements and challenges in wider-coverage,massive-capacity,massive-connectivity,and low-latency due to the explosive growth of mobile data traffic,massive device connectivity.The main limitation in meeting these requirements comes from the unavailability of usable frequency resources.Using Cognitive Radio(CR)technology to improve the utilization of available frequency bands is a promising solution to address the problem of spectrum scarcity.Spectrum sensing,as the primary link in cognitive radio networks,allows secondary users to use vacant spectrum of licensed primary users for opportunity communication.In practice,low signal-to-noise ratio,multipath fading and time dispersion of wireless channels,and noise power uncertainty make spectrum sensing challenging.There are varieties of spectrum sensing methods,and the eigenvalue based algorithms have been widely studied for high detection performance at low signal-to-noise.The eigenvalue of the sample covariance matrix of the received signal reflects a lot of information about the signal and background noise.How to fully mine this information and use it in signal detection is the key of this kind of spectrum sensing method.The dissertation studied this issue in-depth and made contributions in the following aspects:(1)The maximum eigenvalue of the received signal covariance matrix captures the signal correlation characteristics well,while the minimum eigenvalue captures the noise characteristics well.How to make better use of this characteristic to improve the performance of the algorithm is the focus of the existing spectral sensing algorithms based on the maximum and minimum eigenvalues.Thus,in order to improve the existing spectral sensing algorithms based on maximum and minimum eigenvalues,this dissertation proposes several fusion algorithms,collectively known as the maximum minimum eigenvalues a combination algorithm,to improve the detection performance.In particular,the probability of detection and false alarm of ?-MMEP and ?-MMES algorithms are derived utilizing the random matrix theory.The simulation for multi-user,multi-antenna and multi-path scenarios shows the effectiveness of the algorithms and ?-MMEP and ?-MMES algorithms achieve the best detection performance.(2)The characteristics of signal and noise are reflected not only in the maximum and minimum eigenvalues,but also in other eigenvalues.How to make full use of all the eigenvalues of the received signal covariance matrix is also an effective way to improve the detection performance.In order to overcome the drawback of some existing detection algorithms,which utilize partial eigenvalues for spectrum detection,we proposed EES,MEW and EW spectrum detection algorithms in this dissertation by exploiting full advantage of full eigenvalues.The simulation results show that the full use of eigenvalues(joint distribution of eigenvalues)have high probability of detection in low SNR.Moreover,compared with traditional algorithms,the proposed joint eigenvalue based algorithms have relatively higher detection performance and stronger robustness.(3)Using the eigenvalues of the covariance matrix of received signal can certainly improve the detection performance for spectrum sensing.But whether there exists an optimal combination on these eigenvalues?In response to this question,the dissertation proposes the eigenvalue weighting detection algorithms based on the Neyman-Pearson criterion.Utilizing the approximation between eigenvalue and energy,the eigenvalue weighting issue can be transformed to an energy weighting function based optimal problem.By deriving the expression of detection threshold and probability of false alarm,the close form expression of optimal solution is obtained.Considering eigenvalues can further reflect the correlation of signals,we use the obtained optimal weighting coefficients to design new eigenvalue weighting based detection schemes,which further give promotions of the detection performance.Simulations verify the efficiency of the proposed algorithms.