Research On Physical Layer Key Generation Technology Based On Millimeter Wave Massive MIMO Channel

In the next generation of 5G & Beyond communication technology,millimeter-wave(mm Wave)and massive MIMO technologies play a vital role.In today’s digital age,communication security has become increasingly important.Physical layer key generation(PLKG)technology is a method that extracts randomness from wireless channels to generate keys.Its security mechanism is based on the inherent properties of wireless channels,offering high security and lightweight advantages.It is considered a valuable supplement to enhance the wireless communication security of 5G and Beyond.This paper considers the new characteristics of mm Wave massive MIMO channels,which are not covered by traditional PLKG schemes and are no longer suitable for this communication scenario.Therefore,this paper focuses on the research of PLKG in the mm Wave massive MIMO scenario,aiming to design a more suitable and effective key generation scheme for this communication scenario.In addition,since the information reconciliation and privacy amplification stages in PLKG are universal and independent of the channel itself,this paper focuses on the preprocessing and quantization stages of PLKG.The main research content and results of this paper are outlined as follows:(1)Research on the characteristics of mm Wave massive MIMO channel,including spatial and temporal characteristics.The study of channel characteristics is the foundation for designing PLKG schemes.Firstly,the propagation characteristics of mm Wave massive MIMO channels are introduced,and the applicable clustered channel model is presented.Based on this,an initial analysis is conducted on the spatial characteristics of mm Wave massive MIMO channels,including spatial correlation and channel sparsity analysis.By exploring the spatial correlation matrix and the beamspace representation,it is verified that mm Wave massive MIMO channels exhibit high spatial correlation and high angular domain sparsity.Furthermore,the temporal variations of mm Wave channel is explored through Doppler power spectrum(DPS)and Doppler spread analysis,and the slow variations of mm Wave channel caused by channel sparsity is demonstrated and verified in the numerical analysis based on NYUSIM.(2)Research on the preprocessing scheme of PLKG in mm Wave massive MIMO scenario.Due to the spatial and temporal correlation redundancy in mm Wave massive MIMO channels,the channel samples obtained from channel probing need to be preprocessed.In this paper,two types of channel probing modes are constructed to study two types of redundancy problems of channel samples.One is that channel samples only have spatial redundancy,and the other is that channel samples have both spatial and temporal redundancy.Aiming at the problem of spatial redundancy in channel samples,a light-weight and high noise robustness beamspace refinement preprocessing scheme is proposed,which is based on the angular domain sparsity of the mm Wave channels.Aiming at the problem of spatial-temporal redundancy in channel samples,a preprocessing scheme combining beamspace refinement and principal component analysis(PCA)is proposed,which uses the beamspace refinement scheme to remove the redundancy in the spatial domain,and uses PCA to remove the redundancy in the temporal domain,also has the advantages of light-weight and nice noise robustness.(3)Research on the improvement of key generation rate for PLKG in mm Wave massive MIMO scenario.To address the limitations of channel randomness caused by channel sparsity and the resulting low key generation rate of beamspace refinement schemes,this paper proposes a key generation scheme based on multi-dimensional random sources.This scheme improves upon the proposed beamspace refinement scheme by jointly using peak value and peak direction information(i.e.,virtual AOA/AOD)as multi-dimensional features to enhance the rate of key generation.Additionally,in the quantization stage,this paper designs an optimized quantization method for the beamspace refinement scheme called flexible multi-bit quantization.This method determines the number of bits used for multi-bit quantization based on the signal-to-noise ratio of peaks,further enhancing the key generation rate.Experimental results demonstrate that the combination of the multi-dimensional random sources scheme and the optimized flexible multi-bit quantization method can improve the key generation rate effectively while ensuring high key agreement and randomness,thus compensating for the defects of the beamspace refinement scheme.