Research On Physical Layer Security Mechanisms For Massive MIMO Systems

The requirements for higher capacity,throughput and energy efficiency of the mobile networks increase with the rapid development of mobile Internet and Internet of Things applications.In the research and development of the Fifth Generation Mobile Communication Networks(5G),massive MIMO technology has been recognized as one of the important enabling technologies for the future 5G networks.In this work,we focus on the potential physical layer security risks of massive MIMO,and propose two active eavesdropping detection mechanisms that are based on random symbols and signal subspace techniques.The first active eavesdropping detection mechanism exploits the statistical characteristics of dedicated random symbol to form hypothesis testing.The distribution of the detection statistics in the absence of active eavesdropping is theoretically derived,forming a "secure region".With this basis,the hypothesis testing can be performed to check whether there exists active eavesdropping.The detection performance is further improved by combining the detection results of multiple independent tests.An energy detector based mechanism is also employed to prevent the active eavesdropper from evading the detection mechanism.Both theoretical analysis and numerical simulations prove that the proposed detection mechanism can effectively detect the active eavesdropping.Its detection performance is significantly improved with the increase of the number of base station antennas,the signal to noise ratio of base stations,the power of active detection and the length of random symbol sequence.The algorithm is shown to be better than other existing detection algorithms.Finally,the detect algorithm can improve the secure capacity of the system.The second active eavesdropping detection mechanism applies the signal subspace technique.With the increase of the received signal dimension and the number of samples,empirical spectral distribution of the received signal converge to its limit distribution.So the limit distribution can be used as the theoretical basis to determine whether the empirical spectral distribution is normal or not.Firstly,this algorithm solves the limit distribution of legitimate user's signal components by large dimensional random matrix theory,and calculates the theoretical value of the test statistics.Secondly,the random matrix theory is used to eliminate the noise effect from the empirical spectral distribution.The resulting noise-free empirical spectral distribution is consequently employed to perform the hypothesis test of active eavesdropping.The simulation results show that the detection mechanism can effectively detect the active eavesdropping.The detection performance is significantly enhanced with the increase of the ratio of the number of base station to that of legitimate users,the number of sampling points,the signal to noise ratio of the receiving signal in base station and the signal power of active detection.The performance of the proposed algorithm is better than the comparable algorithm.