Research On Reciprocity Extraction And Key Generation Of Massive MIMO FDD Wireless Channel

The openness of wireless channels makes wireless communication systems more vulnerable to security threats such as channel eavesdropping.The traditional security mechanism based on cryptography needs a lot of computing overhead and complex key management,which is difficult to adapt to the requirements of future mobile communications.The wireless physical layer security mechanism uses the characteristics of the wireless channel itself to realize key generation and distribution,providing a technical premise for secure communication in multiple application scenarios in the future mobile communication network.The mobile communication network mainly has two communication modes: Time Division Duplexing(TDD)and Frequency Division Duplexing(FDD),and the two modes are of equal importance.In TDD mode,due to the natural good reciprocity of the channel state information of the uplink and downlink channels,it is easy to generate high consistency wireless keys.At present,there are many research results.In the FDD mode communication system,the channel state information of the uplink and downlink channels does not have obvious reciprocity.It is necessary to study how to extract the reciprocity information from the vector signal domain,which is difficult to study.Therefore,there are few existing research achievements in this field.To solve the above problems,based on the channel modeling of the massive MIMO FDD system,this paper studies the wireless channel reciprocity and the generation technology of wireless channel keys.Through the analysis of channel covariance matrix,virtual angle domain transformation,compressed sensing,and super resolution,this paper proposes a variety of channel reciprocity extraction methods and their corresponding key generation methods,It is an important supplement to the existing TDD wireless channel reciprocity extraction methods.The main contributions and innovative research achievements of this paper include:1.The channel of massive MIMO is modeled,the influence of massive unique MIMO channel characteristics on channel reciprocity extraction is analyzed,and the channel hardening and favorable propagation characteristics are quantitatively analyzed through simulation,which provides theoretical basis and foundation for system modeling in subsequent chapters.Channel modeling is very important for the research of channel reciprocity extraction methods.This paper first summarizes the classification of massive MIMO channel models,and describes the characteristics,applicable scenarios and limitations of various channel models.On this basis,the corresponding classification of specific system channel models in each subsequent chapter is given.Then,the unique characteristics of massive MIMO are described in detail.The channel hardening and favorable propagation characteristics are quantitatively analyzed through simulation,and the influence of spherical wavefront assumption and spatial non-stationary characteristics on channel reciprocity extraction is discussed.2.A channel reciprocity extraction method based on channel covariance matrix eigenvalue is proposed.The channel state information of the uplink and downlink channels in the FDD system does not have explicit reciprocity,but some statistical channel state information in the FDD system,that is,the covariance matrix of the uplink and downlink channels,has reciprocity.This paper proposes a projection method of the uplink and downlink covariance matrices,establishes the reciprocity relationship of the uplink and downlink covariance matrices,and obtains the nonzero eigenvalues of the reciprocity uplink and downlink channel covariance matrices through the eigenvalue decomposition of the uplink and downlink channel covariance matrices.3.A channel reciprocity extraction method based on virtual angle domain transformation is proposed.MIMO channels are sparse in the virtual angle domain,so the channel dimension can be reduced by equivalent transformation in the virtual angle domain.Because of the ”angle reciprocity” between the uplink and downlink channels in the FDD system,it is found through theoretical analysis that the spatial feature set of the uplink and downlink channels in the virtual angle domain has a simple linear mapping relationship,so it can be used to extract the reciprocity characteristics of the uplink and downlink channels in the FDD system.On this basis,a method based on discrete Fourier transform sparse basis and angle rotation spatial feature set is proposed,which effectively improves the reciprocity of uplink and downlink channels in FDD systems.Simulation results show that this method has similar performance with the joint space division multiplexing method,while its computational complexity ratio is greatly reduced.4.A channel reciprocity extraction method based on compressed sensing is proposed.There is a problem of energy leakage when using sparse basis(such as DFT basis)for channel sparse approximation.Large scale MIMO channels are very suitable for the application of compressed sensing technology due to the sparsity of local scattering distribution environment.Based on this,this paper proposes an improved compressed sensing greedy algorithm,which makes full use of the sparsity of large scale MIMO channels.Aiming at the limitation that the original NOMP algorithm can only estimate one of the angle or delay parameters,and the computational complexity is high,which is not suitable for resource constrained communication systems,an improved joint NOMP algorithm that can simultaneously obtain the angle and delay parameters is proposed to extract the frequency independent eigenvalues in the uplink and downlink channels,which can effectively construct the reciprocity between the uplink and downlink channels of FDD.The computational complexity of this method is low,and it has practical significance.5.A channel reciprocity extraction method based on super resolution is proposed.When the actual angle information is not located on the grid,the channel reciprocity method based on compressed sensing will have a grid deviation effect.Compared with the compressed sensing technology,the super resolution algorithm not only uses the sparsity of the channel,but also uses the structure of the antenna array.Based on this,according to the characteristics that super resolution can estimate signals in one band from signals in another band,a super resolution algorithm based on semi definite programming is proposed to extract channel parameters with frequency independence of uplink and downlink channels as a random source for generating symmetric keys,which can effectively construct the reciprocity between uplink and downlink channels of FDD.On this basis,a key generation method based on super resolution is proposed.Finally,the performance of the key method is simulated and analyzed.6.A key generation method based on channel reciprocity is proposed.In massive MIMO FDD systems,the channel covariance eigenvalue matrix,virtual angle domain transformation,compressed sensing algorithm and super-resolution algorithm can be used to extract the frequency independent eigenvalue between uplink and downlink channels,so as to construct or improve the reciprocity of uplink and downlink channels.On this basis,a specific wireless key generation method based on channel reciprocity is proposed.For each reciprocity extraction method,select an appropriate quantization method to obtain the initial key sequence,and then obtain the final key with high consistency through information reconciliation and privacy amplification.Furthermore,the performance of the key method is verified by simulation.