5G Millimeter-wave Massive Array Channel Parameter Estimation And Characteristic Analysis

With the increasing ammount of wireless devices and users' requirements for high data rate and high communication quality,the development of the fifth generation(5G)wireless communication system is extremely necessary.In order to solve the bottlenecks and challenges of existing communication systems,operators need to design the wirelss network architectures properly and employ advanced techniques to fulfill 5G visions and requirements.Millimeter wave(mmWave)and massive multiple-input multiple-output(MIMO)are two key 5G techniques.The deployment of mm Wave and massive MIMO can provide abundants spectrum resources and improve channel capacity,spectrum efficiency,and energy efficiency.High attenuation of mm Wave communication can be compensated by massive MIMO,while array size of massive MIMO channel can be reduced because of the small wavelength.The combination of mm Wave massive MIMO is very meaningful for the development of 5G systems.However,the deployment of key techniques will introduce new channel propagation characteristics.In order to have a better understanding of mm Wave massive MIMO channels and further benefit to the design and optimization of wireless communication systems,extensive channel measurements at different environments,array architectures,and operated frequency bands are needed.In addition,with the increase of array size and delay resolution,high resolution algorithms with low complexity are necessary for measurement data processing.Thus,this work aims to give an indepth and extensive analysis of mm Wave massive MIMO channel characteristics,as well as a thorough study of various channel parameter estimation algorithms.To process measurement data and further to analyze mm Wave massive MIMO channel characteristics,we introduce some array signal processing methods to wireless channel parameter estimation.The basic principles behind every method,their pros and cons,and applications to real channel measurement data processing are analyzed and summarized.Therefore,several algorithms are selected and extended to the uniform rectangular array(URA)scenario.Those algorithms can be classied into two groups,one can be used to the estimation of azimuth angle of arrival(AAoA)and elevation AoA(EAoA),the other one is applicable to the jointly estimation of AAoA,EAoA,and delay.Through simulations we can see that the extended MUSIC performs better than LS-ESPRIT and Unitary ESPRIT algorithms especially with certain array sizes and channel conditions.But the complexity is very high due to the angle searching procedure.The extended Unitary ESPRIT algorithm shows similar performance with LS-ESPRIT algorithm but with reduced complexity.The second group of algorithms can resolve more multipaths due to the inclusion of delay information.Besides the extended JADE MUSIC,JADE ESPRIT,SI-JADE,and SAGE algorithms,we propose the novel improved SI-JADE and 3D FD-SAGE algorithms to further reduce the complexity without sacrificing the accuracy.The improved SI-JADE algorithm performs better than other JADE-based algorithms in terms of accuracy and robustness.SAGE and 3D FD-SAGE algorithms can provide the most accurate estimation results.However,3D FD-SAGE algorithm can resolve more closed multipaths than SAGE algorithm.Applying the above mentioned two groups of algorithms into 60 GHz indoor NLoS scenario channel measurement data processing,we can compare the differences of estimated multipath parameters.The correctness can be verified referring to the measurement environment and setups.It turns out that MUSIC algorithm can be used to determine the 3D angle of several strong clusters,while LS-ESPRIT and Unitary ESPRIT algorithms have dissatisfied performance.In 3D angle and delay joint estimation,estimated multipaths using SAGE,FD-SAGE and the improved SI-JADE algorithms are mostly overlapped.However,it should be noticed that with the increase of array size and frequency points,the complexity of algorithms except for SAGE and FD-SAGE algorithms will increase rapidly.For mm Wave massive MIMO channel measurements,the environment,array structure,and hardware device setups are classified.Three measurement methods are introduced,including their applicative measurement scenarios,pros and cons.Through summarizing the existing mm Wave,massive MIMO,and two techniques combined channel measurement campaigns and relative characteristic analyses,it can be seen that the mm Wave massive channel measurements and analysis are insufficient.So,we carry out indoor mm Wave channel measurements at different frequency bands using massive array.In this paper,two measurement campaigns carried out at 60 GHz and at 16,28,and 38 GHz frequency bands are introduced.By using 3-dimensional(3D)FD-SAGE algorithm to extract multipath parameters from 60 GHz channel measurement data,it is found that multipaths are shown in clusters.Furthermore,with measurement data obtained at three different frequency bands,we focus on analyzing mm Wave massive MIMO channel characteristics,such as high delay resolution,spherical wavefront,spatial and frequency domain non-stationarities,birth-death process,etc.The variances of spatial cross-correlation function(SCCFs)along different axes of antenna array are compared at the same frequency band.The differences of SCCFs at different frequency bands along the same antenna array are also compared.It is shown that,in mm Wave massive MIMO channels,SCCFs can not only be affected by the measurement environment,but also can be influenced by the relative location of Tx and Rx array,as well as the operated center frequency.This phenomenan should be considered in future channel modeling and even array architecture design of wireless network.