Research On Spectrum Sensing Methods And Strategies In Cognitive Radio

With the rapid development of wireless communication,the new wireless communication network equipment has exploded an explosive growth.The demand of users for communication quality is also increasing,which makes the current spectrum resource more and more scarce.The technology of cognitive radio could provide strong technical support for solving the shortage of spectrum resources and improving the utilization rate of spectrum,which will be an important part of the next generation communication network.In order to find and identify idle spectrum resources,the primary task of cognitive radio is spectrum sensing.At present,there are still some problems in the research of spectrum sensing,which is mainly manifested that some of the original research results are difficult to meet the new challenges posed by the current wireless communication services.For the above issues,some new theoretical methods could provide new ideas.In this paper,spectrum sensing methods and strategies in cognitive radio are studied.The main research work is as follows:(1)Cooperative spectrum sensing method based on Cholesky decomposition of covariance matrix.In order to solve the problem that traditional covariance matrix is susceptible to noise fluctuation in cooperative spectrum sensing,a cooperative spectrum sensing method based on Cholesky decomposition of covariance matrix is proposed by combining random matrix theory with multi-user cooperative spectrum sensing model.Firstly,the covariance matrix model is established according to the characteristics of cooperative spectrum perception,based on which features of Wishart matrix are pointed out.Then,the covariance matrix is decomposed by Cholesky based on the random matrix theory and moment matching method.The maximum eigenvalue of the decomposed matrix is taken as the detection statistics,thus the detection threshold is deduced and analyzed.Finally,the influence of noise fluctuation on detection threshold is calculated theoretically.The effectiveness of the algorithm is illustrated by an example and the complexity of the algorithm is analyzed.Simulation results show that the method can reduce the computational complexity while alleviating the impact of noise fluctuation on the algorithm.So it has good robustness.(2)In order to solve the problem of poor performance in spectrum sensing when using traditional machine learning methods,deep learning theory is applied to the field of spectrum sensing.By analyzing the connection between deep learning and pattern recognition,a deep learning network model for spectrum sensing is established.According to the characteristics of two-dimensional covariance matrix gray-scale,a spectrum sensing method based on convolution neural network is proposed.Firstly,by analyzing the covariance matrix model of received signal,the issue of spectrum sensing is transformed into the issue of image processing.Then,according to the gray image of the covariance matrix of the signal,the training data is learned in layers by using the convolutional neural network to extract more abstract features.Finally,the test data is input to the trained convolution neural network model,and the spectrum sensing is completed.Simulation results show that this method has higher recognition accuracy than other machine learning methods in spectrum sensing,and it can effectively implement spectrum sensing tasks in real-time data testing.(3)Coalitional game based distributed cooperative spectrum sensing method.In multi-node distributed cooperative spectrum sensing,the perception ability of different sensing nodes is different due to multipath fading and shadow effect.To solve the problem of combination among these sensing nodes,the game coalition theory of microeconomics is applied to distributed cooperative spectrum sensing,and a distributed cooperative spectrum sensing strategy based on coalition game is proposed.Firstly,the cooperative spectrum sensing system model and related parameters are analyzed.The coalition utility function is established to determine whether the sensing node joins the coalition cooperative spectrum sensing.Subsequently,the perceptual nodes are divided into several coalitions by the coalition formation rules,and then,information exchange is carried out under the single user priority criterion.After several iterations,a stable coalition partition structure is formed.Finally,the spectrum sensing is completed through integrating the perceptual information in the coalition.Simulation results show that the strategy improves the flexibility of coalition formation,and it could help to obtain higher coalition effective value.The detection probability of cooperative spectrum sensing is improved,while its false alarm probability is reduced,indicating that the proposed method has better spectrum sensing performance than other methods.(4)Distributed cooperative spectrum sensing strategy based on reinforcement learning and consensus fusion.Due to the inherent cooperative and distributed characteristics of cooperative spectrum sensing,it is easy to be interfered by malicious users in the process of sensing.These disturbances would cause instability of the system.In order to solve the problem of malicious users’ interference,the reinforcement learning and consensus fusion model are combined.A distributed cooperative spectrum sensing strategy based on reinforcement learning and consensus fusion is proposed.Firstly,a distributed cooperative spectrum sensing model is established based on reinforcement learning and consensus fusion.Then,cooperative users are selected from adjacent nodes by reinforcement learning for consensus fusion,and the reputation generated by the interaction between adjacent users is used as the reward of reinforcement learning.Finally,the interference of malicious users to the sensing network is reduced by multiple iterations,and the spectrum sensing of the whole network is realized by consensus fusion.Simulation results show that the proposed method can effectively identify malicious users and make cooperative spectrum sensing network intelligent and stable.In this paper,some new methods and theories are applied to slove the spectrum sensing problem.Correspondingly,some research results have been achieved,which would have guiding significances in further research of spectral perception.