Research And Implementation Of Cooperative Spectrum Sensing Technology Based On Compressed Sensing

In order to solve the problem of shortage of spectrum resources,cognitive radio technology came into being and attracts the close attention.Spectrum sensing is the core technology of cognitive radio.This thesis studies the traditional spectrum sensing technology.In order to improve the current technology of spectrum sensing,it presents a new cooperative spectrum sensing algorithm using new theoretical tools:random matrix and compressed sensing.In this thesis,the classical MED algorithm is improved.This thesis applies research results about the maximum eigenvalue of covariance matrix in random matrix theory,combining it with energy detection.Then,a new joint energy-eigenvalue detection(EMED)algorithm is proposed.This algorithm takes the ratio of two values as the test statistic,numerator is the maximum eigenvalue of covariance matrix obtained from signals at SU receivers,and denominator is the the average energy of the received signals.The threshold is obtained from the limits of the maximum eigenvalue,the average energy distribution and the false alarm probability.The threshold computing does not require any prior knowledge of any PU signal,which overcomes the shortcomings of traditional energy detection algorithm and MED algorithm.Besides,the proposed algorithm is not sensitive to the noise.When the noise variance is not stable,EMED algorithm has better performance and is more robust.The numerical simulation shows that the EMED algorithm has the best performance and the highest reliability compared with the ED algorithm and the MED algorithm in the presence of noise uncertainty.Based on the EMED algorithm,with the compressed sensing theory combined,a joint energy-eigenvalue detection algorithm based on non-construction compressed sensing(CS-EMED)is proposed.CS-EMED algorithm takes compressed observation signals as the received signal.A befitting observation matrix is necessary to ensure the information of the signal is not damaged.Hence the statistical characteristics of the covariance matrix and the eigenvalue remain unchanged,and the signal can achieve the expected detection performance without signal reconstruction.Simulation results show that,CS-EMED dose not have any noticeable loss of performance and has low complexity compared with the EMED algorithm based on OMP signal reconstruction,Beyond that,a cooperative spectrum detection algorithm based on SU credibility coefficient is proposed.Each SU takes the CS-EMED algorithm.Then they transform test statistics and judgment results to fusion center.After the fusion center finishes the global decison,it is compared with the local results from SU to calculate the credibility coefficient of each SU.The credibility threshold is calculated in FC,which is the average of coefficients without the maximum and the minimum.The SU which has lower credibility threshold than the threshold cannot participates in cooperative spectrum detection.The algorithm effectively reduces the complexity of collaborative cognitive system and improves the performance of spectrum detection.The simulation results show that the performance of the proposed algorithm in this thesis is superior to that of classical soft-decision fusion strategy,and the system overhead of cognitive network is effectively saved.