Research Of Blind Spectrum Sensing Algorithm Based On Non-parametric Statistics

The rapid development of wireless communication technology and the fixed frequency spectrum management policy cause the shortage of spectrum resources.Cognitive radio is a newly developing technology that can improve the spectrum utilization dramatically.The main purpose of cognitive radio is to detect the presence of primary user within the desired frequency band and then enable secondary users to access the vacant channel rapidly without causing interference to primary user.Therefore,a fast and accurate spectrum sensing is a prerequisite and fundamental task in cognitive radio.In this thesis,the spectrum sensing technologies of cognitive radio are studied in detail,and the spectrum sensing is transformed as a non-parametric testing problem.The main works done by the author are listed below:1.Goodness of fit(GoF)test based anti-noise-uncertainty blind spectrum sensing.The performance of the existing spectrum sensing algorithms based on GoF test are excellent,however,they are sensitive to the noise uncertainty.The first component of Cramer-von Mises,insensitive to variances shift,is employed as test statistics in GOF test and a fast spectrum sensing based on this component of Cramer-von Mises is proposed.The detection probability and false alarnm probability of the proposed method are analyzed,and its theoretical threshold is derived.Simulation results demonstrate that the proposed lower complexity anti-noise-uncertainty method obtains the best performance compared with energy detector and the eigenvalue function based anti-noise-uncertainty method.2.Fast cooperative spectrum sensing under multi-antenna Gaussian channels.Multi-antenna based spectrum sensing algorithms are widely used in cognitive radio networks on account of improving the system reliability.Utilizing the difference between the received signal and the noise statistical covariances,two kinds of novel spectrum sensing algorithms,binomial distribution based detection(BD)and wilcoxon signed rank test based detection(WSD),are proposed based on the sample covariance matrix which is calculated from a limited number of received signal samples.BD and WSD algorithms do not need any priori information of the primary signal and the noise.In addition,their thresholds are found via the statistical theory.Both analysis and simulation results show that those two algorithms can obtain better performance compared with energy detection(ED),maximum-minimum eigenvalue(MME)and covariance absolute value(CAV).3.Robust cooperative spectrum sensing under low antenna correlation Rayleigh channels.In diversity-based multiple antenna cognitive radio systems,the performance of the existing cooperative spectrum sensing algorithm on covariance matrix degrades seriously due to the channel independence.In this thesis,spectrum sensing problem is reformulated as a multinomial distribution test problem,and the chi-square goodness of fit test is used to examine the above problelm and a spectrum sensing algorithn based on chi-square test is presented.Theoretical analysis and simulation show that the performance of the proposed algorithm is robust to the antenna correlation and noise uncertainty.All in all,the proposed methods and proposals effectively overcome the existing problems of spectrum sensing in cognitive radio,and its effectiveness is proved by simulations.