Design Of Spectral Feature Detection For Wideband Compressed Spectrum Sensing

In recent years,with the rapid development of mobile communication system and the Internet,the number of users in wireless communication has rapidly increased.The inefficient usage of the spectrum together with the increasing demand of emerging services for more radio spectrum suggests that a transition to a more intelligent and flexible spectrum management regime is needed.Therefore,Cognitive Radio?CR?and spectrum sensing techniques have been proposed to sense the surrounding environment so that spectral efficiency can be improved.Compared with the traditional narrow-band spectrum sensing,wideband spectrum sensing?WSS?technology can sense a wide frequency range at the same time,and it has been applied in many areas.However,many related hardware can not meet the demand due to its high sampling frequency in WSS.Compressed sensing?CS?technology solves this problem and alleviates the sampling bottleneck.The basic theory of CS and the related reconstruction algorithms have been introduced in this thesis.According to the theory which has been introduced before,this thesis presents a weighted compression sampling matching pursuit?W-CoSaMP?algorithm based on spectral feature.The existence of interference emanating from low-regulated transmissions is also taken into account in this CS model.The basic strategy of this algorithm is to compare the spectral shape of the primary users as priori knowledge with the autocorrelation matrix of the received signal.This comparison is made in terms of autocorrelation by means of a correlation matching,thus avoiding the computation of the power spectral density of the received signal.What's more,a weighted formulation of the l₁-minimization is introduced to help give high priority to atoms containing more energy.In addition,shadowing and fading in real wireless communication have been discussed.Take Rayleigh fading for example,a multi-antenna receiver is used.And an angle-of-arrival?AOA?reconstruction is introduced to expand spectrum sensing to the spatial domain.Finally,the simulation of the proposed algorithm is carried out by MATLAB software.The results show that if Rayleigh fading is considered,the normalized RMSE of the estimated power level,frequency and AOA are small when the multi-antenna receiver is introduced compared with no multi-antenna receiver.The proposed algorithm reduces the influence of Rayleigh fading and it can detect the spectrum.