Research Of Hybrid Precoding Technology For Mm Wave Massive MIMO Systems

With the significant data volume demand of broadband mobile wireless devices,higher speeds,larger system capacity,and more considerable bandwidth requirements than 4G systems cannot be met at frequencies below 5G.To solve the shortage of spectrum resources,millimeter-wave communication technology has become a hot spot in academic research.However,in the millimeter-wave band,the signal has a large path loss,and the channel has spatial selectivity.To overcome the shortcomings of the millimeter wave technology,researchers have combined the millimeter-wave communication system and the Massive MIMO system in recent years,which can be effectively realized.The complementary advantages of the two technologies,improving system performance,and increasing system capacity are an important direction for the continuous advancement of 5G research.Millimeter-wave massive MIMO communication is realized by a hybrid transceiver that combines a high-dimensional analog signal processing unit with a low-dimensional digital signal processing unit.Hybrid transceiver-based hybrid precoding technology is resistant to multipath fading.Anti-interference,high spectral efficiency,high energy efficiency,and other characteristics,and can reduce the hardware cost and power consumption of the system,becoming an important signal processing technology for millimeter-wave massive MIMO systems.In this paper,the hybrid precoding technology for millimeter-wave massive MIMOsystem is researched,and hybrid precoding schemes based on different millimeter-wave massive MIMO transmitter architecturesare proposed.The existing precoding method has high hardware costs and high energy consumption.The system has high complexity and low energy efficiency.The main work of this paper is summarized as follows:1.A hybrid precoding scheme based on improved Generalized Low-Rank Approximations of Matrices(GLRAM)is proposed.Firstly,the hybrid precoding algorithm is transformed into analog precoding and digital precoding,respectively.The analog precoding algorithm uses the improved GLRAM algorithm to obtain the analog precoding matrix/combination matrix to get the High array gain.of the large scale MIMO system.Then,in the design stage of the baseband precoder/combiner,the system's RF equivalent channel is used to consider the feature space of the target user's received signal,and the target user's signal power is enhanced while eliminating interference.The block diagonalization(BD)algorithm obtains the digital precoding matrix and attains the hybrid precoding scheme with the best performance.Finally,the simulation results show that the improved GLRAM based hybrid precoding algorithm for the overlapped phaseshifter network can reduce the number of iterative calculations,and has higher spectral efficiency and lower performance than the existing hybrid precoding algorithm based on overlapped subarray.2.For the existing lens array-based millimeter-wave massive MIMO system,only the beam selection algorithm is considered,the beam selection algorithm is not combined with the digital precoding algorithm,and the resolution of the phase shifter in the actual deployment is limited.Inspired by the overlapped antenna arrays in radar,a hybrid precoding architecture for low-precision overlapping phase shifter networks with lens arrays is proposed.Based on this architecture,a two-stage hybrid precoding scheme based on quantization beam alignment and Wiener filtering is proposed.The hybrid precoding algorithm,in particular,the analog precoding algorithm,uses a quantized beam alignment method to obtain analog precoding with a higher array gain.The digital domain employs a Wiener filter precoding algorithm based on a minimum mean square error criterion to obtain a multiplexed gain.The simulation results show that the proposed hybrid precoding scheme based on low-precision overlapping phase shifter beam selection network can obtain not only satisfactory and rate performance effectively,but also is superior to a traditional scheme in energy efficiency.3.A hybrid precoding based on machine learning for mmWave Beamspace MIMO is proposed.Aiming at the problem that the existing millimeter-wave massive MIMO system has large hardware power consumption,high complexity,and high energy efficiency,a millimeter-wave beamspace MIMO hybrid precoding method based on machine learning is proposed.Firstly,ammWave Beamspace MIMO transmitter architecture is proposed,which is a lens array and a sub-connected switch selection network.The analog part of the beam space MIMO system is realized by a partially connected switch selection network instead of a sub-connection.The phase shifter selects the network to implement.Secondly,Based on the proposed architecture and inspired by the cross-entropy(CE)optimization developed in machine learning,an optimal hybrid cross-entropy(HCE)-based hybridprecoding scheme is designed to maximize the achievable sum rate,where the probability distribution ofthe hybrid precoder is updated by minimizing CE with unadjusted probabilities and smoothing constant.Simulation results show that the proposed HCE-based hybrid precoding can not only effectively achieve the satisfied sum-rate,but also outperform the PSs schemes concerning energy efficiency.