Spoofing Attack Detection And Secrecy Capacity Estimation In Massive MIMO Based On Random Matrix Theory

Massive multiple-input-multiple-output(Massive MIMO)technology is considered as the core technology of the physical layer of the 5G network,which has attracted wide investigation among the scholars at domestic and aboard.With the increasing application of Massive MIMO technology,the security challenge in wireless networks has become a more prominent problem.The physical layer security of the Massive MIMO system has attracted much attention from researchers.Spoofing attack is a kind of active jamming attack that threatens physical layer security.At present,a series of methods have been put forward for spoofing attack,such as energy ratio detection,random symbol detection,generalized likelihood ratio detection,and minimum description length(MDL)based detection.Although these methods improve the detection performance of spoofing attacks in different degree,they still have shortcomings.For example,the previous channel state information(CSI)or the additional special signal in the transmission are needed,or the pilot sequence length is assumed to be far greater than the number of antennas,and the detection performance degrades in low SNR,etc.In order to realize the secure transmission of information,the maximum security rate of the signal transmission should approach as closely as possible to the secrecy capacity of the communication system.The existing effective method is to add the artificial noise to the transmission.However,the allocation coefficient selection of the artificial noise power for massive MIMO system is still imperfect.In view of the above problems,this thesis studies the methods of spoofing attack detection and secrecy capacity estimation based on random matrix theory(RMT).The creative works of this paper are as follows:(1)A modified Flexible Detection Criterion(FDC)-based spoofing attack detection algorithm is proposed for the Massive MIMO system,which is applicable for the situation that the number of base station antennas and the length of pilotsequence are in the same magnitude.The RMT-FDC criterion is used to decide whether or not the eavesdroppers exist.Based on the FDC criterion,the algorithm uses RMT to modify the eigenvalues of the sample covariance matrix of the received signals by the base station,which makes the sample covariance matrix replace the statistical covariance matrix to improve the detection performance.The algorithm does not need prior information such as known noise variance,channel state information and so on.(2)Two eigenvalue-based spoofing attack detection algorithms are proposed for the Massive MIMO system,including the eigenvalue-trace ratio(ETR)and the standard condition number(SCN)algorithms based on RMT.By analyzing the distribution characteristics of the eigenvalues of the sample covariance matrix,the decision threshold is calculated by combing the M-P law,so that the detection of the spoofing attack is realized.The two algorithms reduce the computational complexity compared with the modified FDC criterion algorithm.(3)A random matrix theory-based secrecy capacity estimation algorithm is proposed for the massive MIMO system.In order to ensure the security of information transmission in Massive MIMO system,the artificial noise is added to restrain the eavesdropper's SNR and improve the secrecy capacity of the system.Also,based on RMT,the problem of the power allocation coefficient of the artificial noise is analyzed under different base station transmitting powers.The results show that the full power should be used in the transmission of information when the transmission power is low,and the optimal power allocation coefficient is selected to produce artificial noise and obtain the maximum safety capacity when the power is large.Further,under optimal power allocation coefficient,the safety capacity can be increased by increasing the transmit power to realize the secrecy of information transmission.