Research On Wideband Spectrum Sensing Based On Compressive Sensing

Cognitive Radio(CR),as an intelligent spectrum sharing technique,can detect and access the allocated but unoccupied spectrum by sensing surrounding wireless environment,reusing the non-renewable spectrum resource.Spectrum sensing is the precondition and prerequisite for the realization of CR technique.The spectrum sensing technology with good performance can not only improve the efficiency of spectrum utilization,but also avoid the harmful interference to the primary users.Wideband spectrum sensing can provide more spectrum access opportunities for cognitive users,and thus it is widely concerned by researchers.The wideband spectrum sensing method based on Compressive Sensing(CS)utilizes the sparseness of wideband spectrum to realize the sub-Nyquist sampling of wideband signal,reducing the sampling pressure of the RF front end,which provides an effective solution for breaking through the bandwidth bottleneck of wideband spectrum sensing.Based on the systematic study of CS theory and spectrum sensing algorithms,this paper focuses on the key technologies of wideband compressive spectrum sensing.The main contents and contributions of this dissertation are summarized as follows:1.In practical application,the random measurement matrix needs a large storage and its hardware implementation is difficult.To address this problem,this paper proposes a deterministic measurement matrix,which basically accesses the optimal coherence.Firstly,one class sparse square matrix is constructed by using finite fields.Then the Hadamard matrix or DFT matrix is nested into the square matrix to construct an asymptotically optimal deterministic measurement matrix.Through the theoretical analysis,it is pointed out that the coherence of this matrix is redundant.A modified method is proposed to reduce the coherence of the measurement matrix,which can basically access the greatest lower bound of the coherence(Welch bound).In addition,an optimization method for measurement matrix is proposed to overcome the disadvantage of existing optimization algorithms,which focus on matrix coherence but ignore global coherence.The proposed algorithm can reduce the coherence and the global coherence of the measurement matrix at the same time.2.Most of the sparsity order estimation algorithms need to reconstruct the signal many times,resulting higher computational complexity.To address this problem,this paper presents a sparsity order estimation algorithm based on the energy of measured signal.Asymptotic random matrix spectrum analysis theory is used to derive the asymptotic eigenvalue probability distribution function of the measured signal's convariance matrix.Based on this,the relationship between the sparsity order and the energy of measured signl is deduced.This algorithm only needs to calculate the energy of the measured data,so the computational complexity is low.At the same time,the algorithm takes into account the eigenvalue distribution of the covariance matrix,so it has a better sparsity order estimation precision.3.When the sparse block boundaries are unknown,the performance of existing block sparse reconstruction algorithms is seriously deteriorated.To address this problem,this paper proposes a weighted double backtracking matching pursuit algorithm to reconstruct the block sparse spectrum.This algorithm uses block sparsity to improve the greedy algorithm from two aspects.On the one hand,the algorithm proposes weighted proxy to select the candidates,which can increase the probability of selecting correct supports and improve the convergence speed.The calculation method of the weight is also given.On the other hand,double backtracking method is adopted to eliminate the support points with low trust degree to ensure the accuracy of the candidate support set.Compared with existing block sparse reconstruction algorithms,the proposed algorithm can maintain good reconstruction performance when the block boundaries are unknown.The simulation results show that the wideband compressive spectrum sensing based on the proposed algorithm has better detection probability and false alarm probability compared with other methods.4.The noise folding phenomenon seriously affects the wideband compressive spectrum sensing performance.To address this problem,a spectrum reconstruction algorithm based on selective measurement is proposed to realize the wideband spectrum sensing with suppressing noise.The proposed algorithm uses the measurement to estimate the probability that each position of the wideband signal spectrum contains the signal component,which is used to guide the optimization of measurement maxtrix.The optimized measurement matrix can selectively sense the sparse signal and suppress the noise to improve the SNR of the measurement,resulting in the improvement of sparse reconstruction performance.Finally,it is pointed out that increasing the measurement times can futher enhance the performance of denoising reconstruction.5.Most of the cooperative compressive spectrum sensing algorithms are based on signal spectrum reconstruction,with the result that the computational complexity is large.To address this problem,this paper proposes a cooperative compressive spectrum sensing algorithm without reconstructing the wideband spectrum.The cooperative Seconary User separately measures the received signal and processes the measurement data.Then it is uploaded to the fusion center to obtain the characteristic statistics,which is used to estimate the spectrum energy of each subchannel.In this paper,the probability distribution of the characteristic statistic is derived.Combining with the maximum posterior probability criterion,the decision threshold is derived,which is used to determine whether the subchannel is occupied or not.The proposed algorithm does not need to reconstruct the wideband spectrum,so the computational complexity is low.And the performance of the proposed algorithm is not constrained by the compressive sensing reconstruction condition.The poor performance caused by the lack of measurement vector dimension can be improved by increasing the number of cooperative Secondary User or improving SNR,so the application environment of the proposed algorithm is more flexible.The simulation results show that the spectrum sensing performance of the proposed algorithm is better than the cooperative spectrum sensing algorithms based on joint sparse reconstruction.