Research On Wide-band Compressed Spectrum Sensing Algorithm

In cognitive radio(Radio Cognitive,CR)system,spectrum sensing tech-nology is the primary and key technology to detect spectrum opportunity.Com-pared with the traditional narrow-band spectrum sensing,wide-band spectrum sensing can find much more spectrum opportunities as soon as possible and thus enhance the transmission rate of cognitive radio networks by providing adequate frequency resources,thereby meeting the requirement of high com-munication data rate in the process of transmission from traditional voice com-munication to multimedia communication.In the process of transition from narrow-band spectrum sensing to wide-band spectrum sensing,the range of the spectrum to be detected will be changed from several MHz to several GHz.Due to the large width of the detection band,traditional sampling method will make the hardware face a high sampling rate,however,the introduction of the com-pressed sensing theory in the wide-band spectrum sensing process can solve this problem effectively.Wide-band compressed spectrum sensing,aiming at detect wide-band,can reduce the sampling rate,which is one of the main re-search directions in broadband spectrum sensing technology.In this thesis,we discuss and study two problems in the broadband spectrum sensing algorith-m,which are also the research results of the National Natural Science Fund Project(the research on the broadband spectrum sensing and control of the po-larization region,project number 61372116)and the Central University Basic Research Fund(multi-source group information representation and low ener0 gy content distribution network theory and key technology research,project number 2014ZD03-01).The research content and research results of this thesis are mainly:1 Summarize the existing spectrum sensing algorithms in detail.The re-search status,main features and application scenarios of various spectrum sens-ing techniques are analyzed in detail,and the reasons and necessity of spectrum sensing technology in the transition from narrow-band to broadband are ana-lyzed.Two problems in the implementation of existing wide-band spectrum sensing algorithm are proposed.First,the computational complexity of ex-isting wide-band spectrum sensing algorithms becomes too complicated when sparsity goes large.Second,existing wide-band spectrum sensing algorithm consider spectrum sensing independently between sensing periods,leading to redundancy samples and computation.2 For the first problem,the thesis studies how to reduce the computation-al complexity of wide-band compressed spectrum sensing algorithm.In the premise of the probability of channel occupancy,we first design compressed sampling matrix by using the channel occupancy probability.Then using this compression sampling matrix to sample the original signal is equivalent to di-viding the original signal into two sets of sub-signals with half the degree of sparsity and sampling,so the process of recovering the original signal is equiv-alent to recovering the two sets sub-signals.The computation complexity of the sub-signal recovery process is reduced by the sparse degree.According to such analysis,a low complexity wide-band compressed spectrum sensing algorithm based on channel occupancy probability is proposed.The simulation analysis shows that the algorithm can effectively reduce the computational complexi-ty and improve the real-time performance of the broadband spectrum sensing algorithm by twenty percent.3 For the second problem,the thesis studies how to use the correlation of spectrum occupancy states between sensing periods to avoid the sampling and computational redundancy of wide-band compressed spectrum sensing algo-rithm.In the actual CR application scenarios,the occupancy of each channel in broadband signal is often correlated with the adjacent detection period,which is the same as the detection results of the adjacent detection cycle.According to such a fact,this thesis first defines the correlation of spectrum occupancy states in broadband signal,and then designs a compressed sampling matrix based on such correlation.Further,a compressed wide-band spectrum sensing algorithm based on correlation of spectrum occupancy states is proposed.Sim-ulation analysis shows that the algorithm can dynamically adjust the number of samples,avoiding the waste of sampling,and the real-time performance of broadband spectrum sensing algorithm is improved by thirty percent.