Font Size: a A A

Downlink Channel Reconstruction Method For FDD Massive MIMO System

Posted on:2020-06-14Degree:MasterType:Thesis
Country:ChinaCandidate:S B XieFull Text:PDF
GTID:2428330572487261Subject:Information and Communication Engineering
Abstract/Summary:PDF Full Text Request
In a massive MIMO cellular network with frequency division duplex(FDD)mode,channel state information(CSI)of downlink is needed for beamforming.However,since the uplink and downlink are deployed in different frequency bands and don't have reciprocity,additional time-frequency resources proportional to the number of base station antennas need to be consumed for downlink CSI estimation.Therefore,reducing the acquisition cost of downlink CSI has become a key problem in system design.In this paper,we concentrate on reconstructing downlink CSI directly through uplink CSI without any feedback.Firstly,the correlation between the uplink and downlink channels of FDD system is discussed.At the same time,the performance of the partial parameter reconstruction scheme based on spatial angle and channel covariance matrix is analyzed.Due to the poor performance of existing scheme,we propose two reconstruction schemes for the two types of channels:line-of-sight(LOS)and non-line-of-sight(NLOS).In LOS scenario,the spatial reciprocity of the uplink and downlink channels is strong and channel paths are sparse.A reconstruction scheme based on the coherent subchannel model is proposed,in which the mapping relation between uplink and downlink is generated from traditional geometric channel modeling method.The scheme has the advantages of high real-time and excellent performance and simple deployment.In NLOS scenario,the spatial complexity of wireless channels increases greatly,and the performance of schemes dependent on channel path parameter estimation decreases sharply.Each element in the CSI matrix is continuous,which is very similar to the nature of the pixels in the image.Thus,we convert CSI matrix into images,CSI reconstruction can be solved through deep learning algorithms.Better reconstruction performance can be obtained especially in NLOS channel.The disadvantage is that the algorithm implementation complexity is higher than the traditional scheme.All in all,this paper analyzed the correlation between the uplink and downlink in FDD massive MIMO system.Furthermore,two downlink CSI reconstruction schemes are proposed for LOS and NLOS channel,respectively.It provides possible ideas and references for FDD massive MIMO channel modeling and system design.
Keywords/Search Tags:Massive MIMO, Frequency division duplex, Channel reconstruction, Channel modeling, Deep learning
PDF Full Text Request
Related items