| Cognitive Radio(CR)is a technology to solve the problem of imbalance of existing spectrum resource allocation and low spectrum utilization.The spectrum sensing plays an important role in cognitive radio technology.However,the traditional spectrum sensing algorithm has problems such as large interference to primary users and high perceived delay.This thesis focuses on the research of cognitive radio spectrum sensing technology based on artificial intelligence.A nonlinear energy detection algorithm based on improved particle swarm k-Nearest Neighbor(k NN)-Support Vector Machine(SVM)and an Extreme Gradient Boosting(XGBoost)cooperative spectrum sensing algorithm based on multiuser compressed sensing is proposed.The main work of the thesis is as follows.(1)This thesis briefly introduces the research background of cognitive radio,the research status of spectrum sensing and the challenges it faces,and introduces the basic theory of spectrum sensing and typical spectrum sensing algorithms.(2)In the nonlinear energy detection system based on machine learning,a nonlinear energy detection algorithm based on improved particle swarm k NN-SVM is proposed.Firstly,the boundary vector set in the training samples is extracted by the idea of k NN,and then the decision function is generated by SVM training instead of the decision threshold of the energy detection algorithm,the algorithm overcomes the influence of the energy detection algorithm's detection performance by noise power fluctuation,low signal-to-noise ratio of primary user signal and too few sampling points.The simulation results show that the proposed algorithm can effectively improve the detection performance of the energy detection algorithm.(3)In order to overcome the shortcomings of high sampling rate of wideband signals and high probability of false alarm of single secondary users,a XGBoost cooperative spectrum sensing algorithm based on multiuser compressed sensing is proposed.Firstly,the wide-band signal is sampled at low speed by using the observation matrix in the optimized compressed sensing and the training sample statistic is calculated,and then the optimized XGBoost algorithm is used to predict the spectrum state,the judgment result is made according to the fusion rule in the fusion center.The simulation results show that the proposed algorithm can effectively improve the cooperative spectrum sensing performance. |
