| Massive MIMO has become a hot technology in 5G Communication System because of the advantages in improving system performance.In Massive MIMO systems,base stations equipped with a very large number of antennas can serve multiple users within the same time-frequency resources,which can not only improve the capacity and reliability of wireless systems significantly,but also effectively reduce the transmitting power of the base station.However,all these advantages depend on the accurate channel state information(CSI).In FDD mode,the CSI can be obtained through UE(limited)feedback.While the training burden for forward-link pilots in a FDD system is proportional to the number of antennas.Using FDD would forever impose a severe limit on the number of base station antennas.Thus,massive MIMO technology is usually used in TDD(time-division duplex)mode.An efficient way to obtain downlink CSI in TDD mode is using uplink pilots and reciprocity of the downlink channel.Due to the limitation of pilot length,users in different cells reuse a set of pilot sequence.Inevitably,non-orthogonal pilots from different cells contaminate each other.The performance of Massive MIMO systems is severely impacted by the pilot contamination.How to suppress pilot contamination becomes an urgent problem in Massive MIMO systems.This paper studied the pilot contamination in the linear cellular network firstly.Channel estimation error and spectral efficiency are analyzed in detail,as well as the influence factors of the pilot contamination,including the number of antennas,the number of cells,pilot length,cross gain and so on.Extend to the plane cellular network,the conclusions of linear cellular network still work.The pilot contamination in plane cellular network is more serious with the increase of cell number.This paper solves this problem in the following two aspects:reducing interference and enhancing useful signals.In order to reduce interference,time-shifted pilot scheme and pilot power control scheme are proposed.Both methods need to partition cells into different groups.Cells in different groups transmit pilot in different time slots,which can reduce the cross gains(in the same frequency/time)relative to the direct gains and suppress the inter-cell pilot contamination effectively.However,in time-shifted pilot scheme,useful pilot signals will suffer uplink and downlink data interference.Besides,it is inflexible and impractical to modify the frame structure frequently.In contrast,pilot power control scheme is more simple and practical.It has no introduce extra interference and only needs to shorten the pilot length.In terms of enhancing the useful signal,a low-complexity hybrid analog and digital beamforming structure is deployed at the base station.In the hybrid beamforming structure,the base station maintains a set of beams in analog domain and each user needs to select the best beam based on the measurement of beam reference signals.It is obvious that beam selection and beam design will affect the accuracy of the channel estimation and the quality of received pilots.So,analog beam generation,beam distribution model and beam selection are detailed designed in this paper.What's more,two analog beam interpolation methods,angle and amplitude interpolation,are further proposed to increase the accuracy of downlink beamforming and achieve higher spectral efficiency.Interpolation methods are based on the spatial correlation.In interpolation methods,base stations can successively approximate the best beam through interpolate the candidate beams.Simulation results show that interpolation methods can enhance the quality of received useful pilots and improve the system performance significantly. |
PDF Full Text Request