Research On Channel Estimation Of Mm Wave Massive MIMO Systems Via Tensor Theory

In the era of the Internet of Everything,the demand for wireless communication services is growing exponentially,and wireless communication technology needs to be further developed in high speed,large connection,low latency,and low cost to cope with diverse communication scenarios.Millimeter-wave(mmWave)massive MIMO technology is regarded as the most potential technology for future wireless communication because of its abundant spectrum resources,considerable antenna gain,and higher throughput.The development and transformation of the communication system is a process of coexistence of opportunities and challenges.The changes in channel characteristics,the increase in complexity,and the upgrade of the system architecture brought about by technological improvements have made the acquisition of Channel State Information more difficult.This paper takes the channel estimation problem of mm Wave systems as the core,and summarizes the main challenges through a large number of classic literature,and focuses on the analysis of the propagation characteristics of mm Wave channels.Then,focusing on the three requirements of the effectiveness,accuracy,and low overhead of the channel estimation algorithm,combined with tensor theory,the channel estimation problem of 3D mm Wave channels and multi-user time-varying mm Wave channels in practical application scenarios is studied.First of all,this paper introduces the development of mm Wave massive MIMO systems and discusses new challenges encountered in channel estimation under the background of mmWave massive MIMO technology.Further,this paper introduces the research status of channel estimation algorithms in detail,and analyzes the existing problems like the mm Wave channel sparsity mining is not thorough,the spatial characteristics of uniform arrays are ignored,the algorithm is insufficient stability and high pilot overhead.Then,the geometric parameter model is introduced in combination with the propagation characteristics of mmWave,and the manifestations under different conditions are analyzed.At the same time,the mathematical concepts related to tensor theory used in this paper are expounded.Secondly,starting from the 3D mmWave system under the digital architecture,this paper proposes a multi-dimensional Vandermonde constrained 3D mmWave uplink channel estimation algorithm.In this method,we fully exploit the mmWave channel sparsity and uniform array response information,and use the space-time diagonal code as the pilot to model the received signal at the base station as a tensor which admits CANDECOMP/PARAFAC Decomposition(CPD),and correlate the spatial response of the array as a vandermonde structure of the factor matrices.Based on this structure,a stable tensor decomposition algorithm is derived and the channel parameters are solved from the factor matrices.Theoretical analysis and numerical simulation show that the proposed method has lower computational complexity compared with the traditional compressive sensing(CS)based algorithm and the improved 2DESPRIT algorithm,and has significantly improved in stability and accuracy.Finally,for high-speed mobile scenarios such as UAV swarms and Internet of Vehicles in reality,this paper studies the multi-user uplink time-varying mm Wave channel estimation problem under the hybrid beamforming architecture.In order to better separate the dynamic changes of the channel in the time dimension,a new pilot structure extended in the time dimension is designed.Based on this structure,the pilot signals at the receiving end are combined into a third-order low-rank tensor,so as to realize joint estimation of multiple mobiles’channel under pilot synchronization.In order to further reduce the pilot overhead,the relationship between the pilot allocation and the number of channel propagation paths is analyzed under the premise that the essential uniqueness condition is satisfied,and the relationship between the number of pilots and the system performance is verified in the numerical simulation.Therefore,a pilot allocation scheme that takes into account both performance and overhead can be obtained.Simulation results show that,compared with other schemes based on CS and its improved adaptation algorithm,the proposed scheme can improve performance while reducing the pilot overhead.