Distributed Spectrum Sensing Based On Improved Clostationary Characteristic

With the increase of wireless communication services,the wireless spectrum resources are becoming scarce.To solve this problem,the traditional static spectrum allocation concept was abandoned and Cognitive Radio technology came into being.With Cognitive Radio technology,the unoccupied spectrum resources can be accessed by users without interfering with the normal communication of authorized users,and it greatly improves the spectrum utilization.Spectrum sensing technology is one of the most important research fields in cognitive radio system.The detection of potential spectrum hole in wireless communication environment was realized with spectrum sensing,and it provides important guarantee for the dynamic distribution of spectrum.In this paper,focusing on spectrum sensing technology,we discuss the improved cyclostationary detection in single-node spectrum sensing system and the distributed cooperative signal detection based on average denoising.The main innovations and work of this thesis are as follows:In this thesis,an improved cyclostationary detection algorithm based on the correlation between the signal and noise is proposed.After analyzing the traditional algorithm,we find an approach to enhance the anti-noise performance of the feature.In the improved algorithm,new feature based on the correlation between signal and noise is used.And simulation experiment proved its valid.In this thesis,a spectrum detection algorithm based on machine learning is proposed.In the traditional detection algorithm based on threshold detection strategy,the threshold value is calculated from the incomplete sampled data,and it is not optimal.In this thesis,the detection task is transformed to classification problem,which can be solved by machine learning.Through training the SVM and neural networks using large sample data,the resulting model has a higher detection performance than traditional threshold detection.In this thesis,a cooperative signal detection algorithm based on distributed feature enhancement algorithm is proposed.With traditional fusion method,bad data points may have a high weight,which will further reduce the detection performance.In this thesis,a distributed feature enhancement algorithm based on average denoising is proposed,which can preserve the information of the signal in the distributed cooperative signal detection system.The improved method has a higher detection performance than traditional method.