Data Mining Algorithm Of Spectrum Sensing Data

With the rapid development of wireless communication technologies,the ever increasing demand for limited spectrum resources will eventually cause spectrum scaricity problem.Using data mining technology to analyze the spectrum sensing data,find out the potential spectrum usage rule in the massive data,and predict the spectrum usage in the future,so as to provide reliable basis for the dynamic spectrum access.The theory of data mining algorithm and radio spectrum sensing technology are studied.The principle of the least squares support vector algorithm(LSSVM)and the optimization method of the model are deeply studied.The main work and innovation of this paper are as follows:Firstly,a feature weighted LSSVM algorithm based on immune algorithm has been comprehensively presented.Compared with the traditional least squares support vector machine algorithm,the feature weighting vector is introduced,and the new parameters are formed with the super parameters of the original model.In order to improve the accuracy of the model,use immune algorithm to find the optimal feature weighting vector.Second,this thesis proposed a feature weighted LSSVM algorithm based on simulated annealing.In order to weight the different feature dimensions,the new alogrithm changing the original supper parameters and then used the coupled simulated annealing algorithm to find the optimal supper paramters.Besides,the dissertation compare the off-line prediction and online prediction algorithm.Thirdly,this thesis designed a complete simulation system which simulate the satellite communication system for spectrum data sensing and predicting.The system includes the geographic movement module,user access module and signal generation module.The simulation results showed that the two proposed algorithm can improve the prediction accuracy of the spectrum data.This thesis studied the data mining technology of spectrum sensing data,improved the method of least squares support vector machine algorithm,and gave the prediction results to the dynamic spectrum access user as a reference,which is meaningful in both theoretic and application.