Research On Distributed Spectrum Sensing Algorithm Based On Bayesian Compressed Sensing

Cognitive Radio(CR)technology enables different wireless communica-tion systems to share spectrum resources,solving the problem of spectrum re-source shortage caused by the growth of wireless communication services.In CR,spectrum sensing is a key technique and is a prerequisite for other tech-niques in CR.wide-band spectrum sensing enables CR users to sense multi-ple spectrum holes at the same time for fast switching,which improves spec-trum utilization.However,due to the limitation of the Nyquist sampling the-orem,the traditional narrow-band spectrum sensing method will face tremen-dous hardware pressure when applied to wide-band signals.To this problem,Compressed sensing technology is viewed as a good solution,because it can sample compressively and accurately recover the original signal.Compressed sensing theory has been a hot area of research,scholars have proposed a variety of reconstruction algorithms to improve the reconstruction accuracy of com-pressed sensing,as well as distributed compressed sensing theory to solve multi-user compressed perception problem.Bayesian compression sensing algorithm based on probability model has good reconstruction performance.However,applying Bayesian compressed sensing model to multiuser cooperative spec-trum sensing still has many problems to be solved.In this thesis,we focus on the Bayesian compression sensing algorithm to investigate and discuss the distributed compressed spectrum sensing in CR networks in order to solve some existing problems in the compressed sensing of broadband spectrum.The content of this thesis is also the result of the au-thor's participation in the project of Natural Science Fund("Research on Sens-ing and Control of Wideband Spectrum in Polarization Domain",Project No.61372116).The main contents of this thesis are as follows:1.The existing research status of multi-user spectrum sensing technology and distributed compression sensing technology in CR networks are investi-gated.Based on the application characteristics of multi-user spectrum sensing in different scenarios,we analyze The shortcomings the existing distributed compression sensing algorithm,indicate the challenges in multi-user coopera-tive compressed spectrum sensing in CR distributed networks.This paper fo-cuses on issues when the Bayesian compressed sensing algorithm is applied to multi-user cooperative spectrum sensing:First,how to make full use of the cor-relation between multi-user signals to improve the reconstruction performance;Second,how to sense spectrum in a distributed network without a fusion center;Third,how to make cooperative compression spectrum sensing algorithm adapt to the dynamic change of CR network.2.In order to make full use of the correlation between multiuser signals,this paper focuses on the characteristics of multiuser signals in CR networks and proposes a common sparsely distributed probability signal model.Based on this signal model,combined with the user's sensing information,a joint Bayesian regression model is established to reconstruct the original signal.Simulation shows that the distributed Bayesian method based on common sparse distribu-tion signal model has better performance at low signal-to-noise ratio and low sampling rate.3.Aiming at the problem of cooperative spectrum sensing in a distributed network without a fusion center,this paper studies the Bayesian cooperation reconstruction process,and divides the joint reconstruction task into multiple sub-tasks and allocates them to CR users in the network.Each user need not transmit the sensing information to the fusion center,but rather to help local reconstruction through the sharing of information between networks.The sim-ulation results show that users have similar perceived performance using dis-tributed collaborative methods compared with centralized methods.4.In order to adapt to the dynamic change of CR network for coopera-tive compressed spectrum sensing algorithm,this paper proposes an adaptive sampling rate adjustment scheme.Through adaptive node selection,rate ad-justment,the spectrum sensing process can be adjusted as the network changes.Simulation results show that the adaptive node selection scheme can make the sampling rate of nodes in the network quickly converge to a better rate.