Research On Key-based Physical Layer Security Technology In 5G Massive Mimo System

In the era of continuous development of wireless communication technology,with the commercialization of 5 G technology,information interaction is becoming more and more frequent.Meanwhile,how to ensure the security of wireless communication received widespread attention.At present,with the enhancement of computer computing capability,traditional encryption technology will face challenges.Physical Layer Security(PLS)technology is not affected by the computing power of eavesdroppers and has become a research hotspot for its low computational overhead and the potential of theoretical security.Most of the research on physical layer key security technology is based on singleantenna or small multi-antenna paired communication scenarios.Otherwise,the fusion of millimeter-wave and Massive MIMO has become the key technology of 5G.At the same time,massive MIMO systems can greatly enrich the randomness of physical layer keys.Therefore,this thesis focuses on the key-based PLS technology under massive antenna systems.Firstly,to avoid a mass of extra signaling overhead resulting from high-dimensional channel estimation in massive MIMO systems,this thesis proposed a scheme using beam direction information as a random source.This scheme designs the relevant signaling process based on the existing beam management mechanism of 5G NR and can generate keys synchronously during the two-stage beam search process between the base station and the user,which requires less additional signaling overhead.In addition,given the problem of poor randomness of the generated key caused by uneven beam distribution,Huffman coding is used for quantization in the feature quantization stage,thereby improving the randomness.And the proposed scheme can reduce the key disagreement rate(KDR)to less than 1%when the signal-to-noise(SNR)is-10dB.Secondly,in purpose to gain a settlement between the key generation rate(KGR)and KDR,this thesis proposed an adaptive quantization algorithm based on K-means clustering.To avoid high-dimensional channel estimation,the channel feature quantized by this algorithm is the equivalent baseband channel information estimated after beam search,and there is no need to take the entire Channel State Information(CSI).In the key capacity analysis part of the proposed algorithm,the key capacity of the key generated based on the equivalent CSI is deduced,to determine the maximum number of bits that can be quantified by the extracted channel characteristics.In the part of adaptively selecting the quantization level K algorithm design,the relationship curves between SNR and KDR under different K values were obtained in advance.And the upper limit of the value of the quantization level K is determined according to key capacity.The value of K becomes smaller than smaller until the KDR corresponding to the selected K value is lower than the preset threshold value,to achieve a higher KGR under the condition of ensuring a lower disagreement rate.Finally,under the multi-user hybrid beamforming fixed sub-array architecture,a multi-user key generation mechanism based on phase compensation beam searching is proposed.Firstly,a beam searching algorithm based on phase compensation is proposed to reduce the complexity of multi-user equivalent baseband channel acquisition,the best beams for other sub-arrays are determined by phase compensation between sub-arrays based on beam searching results of the first sub-array.Based on this,the corresponding multi-user key generation scheme is further designed.The proposed scheme can generate an initial key with a lower disagreement rate by K-means clustering quantization at a lower implementation complexity.