Research On The Spectrum Sensing Algorithm Based On Sparse Characteristic In Frequency Domain

With the rapid development of wireless communication technology,the static spectrum allocation strategy and the static usage method couldn't satisfy the high-speed and high-quality communication demands.The cognitive radio system provides a new scheme for dynamic spectrum allocation and utilization.Since proposed in 1999,cognitive radio has received abroad attention and research.As the core technology of cognitive radio,spectrum detection algorithm can identify the "spectrum hole" constantly,provide access for the secondary users,and then improves the efficiency of spectrum utilization.The sparse characteristics of signal in some domain,one side can help to concentrate signal energy more valid,the other hand,also can effectively reduce the sampling-rate and sampling-difficulty,which respectively correspond to the high detection probability and the requirements for real-time detection of spectrum detection algorithm.Therefore,this paper attempts to combine the sparse features and detection algorithms,focus on the detection algorithm based on sparse features in frequency domain,and analysis the detection performance of the algorithm.Firstly,this paper introduces the background and significance of the research,does analysis of the advantages and defects of the existing detection algorithms.After that,we point out the detection algorithm based on the maximum of power spectrum estimation and the detection algorithm based on MWC-SBL method.Then,the basic theories used in these two algorithms are summarized,including power spectrum estimation,MWC sampling system,Sparse Bayesian Learning,etc.Beside,the corresponding simulation results are also given.Secondly,the detection algorithm based on the maximum of power spectrum estimation is analyzed in depth,and the maximum of power spectrum estimation is selected as statistical decision variable.After the analysis of variable mean,variance and correlation of statistics statistical decision,we use the Chi-Square distribution to model the distribution of statistical decision variables,under H0 and H1 assumption.We also derive the theory expression of detection probability,false alarm probability and the detection threshold.In the part of simulation,we verify the correctness of chi-square distribution and the advantages of the selection of maximum value as statistical decision variable,then analysis the impact of sampling data length and the segment numbers of power spectrum on detection performance,comparing with the performance of energy detection algorithm.Subsequent simulations show that the choice of different window functions has an impact on the detection probability and the accuracy of chi-square distribution modeling,and thetwo aspect gives an opposite trend.Although the detection algorithm based on the maximum value of power spectrum estimation has a unique advantage for sparse narrow band signal,but because of the sampling limit for sparse multi-band signals,it can not achieve the requirement of real time detection.Finally,the detection algorithm,based on MWC-SBL system,is studied.The algorithm,based on compressed sensing,using sparse Bayesian learning algorithm to fully mine the information of observed data,provides another scheme for broadband detection.The support set is defined as statistical decision variable,and the judgment of signal presence or absence of each frequency band is carried out through the detection of the support set.After the description of algorithm is completed,the statistical expressions for detection probability and false alarm probability are defined,according to the relation between the recovered support set and the original support set.Then the following simulation focus attention on detection performance with fixed position for single signal and multi-signal,detection performance with random position for single signal and multi-signal,the improved detection performance under different equality principles,mean square error of the algorithm,and comparing with the detection performance of algorithm based on MWC-OMP.Simulation results show that the MWC-SBL based detection algorithm is better than the MWC-OMP based detection algorithm in most performance indexes.