BDMA Transmission For Massive MIMO Systems

With the rapid development of intelligent terminals and the boom in the requirement of data services,the performance requirement of communication systems can hardly be met by traditional wireless physical layer transmissions.As the radio spectrum resources is becoming increasingly tense,the massive multiple-input multiple-output(MIMO)technology has become one of the key technologies in future wireless communication systems for its potential in utilizing spatial resources and increasing spectral efficiency.Due to the difficulty in obtaining the instantaneous channel state information(CSI)of all users,in this thesis,the beam division multiple access(BDMA)transmission for massive MIMO system under Rician channel and universal mobile telecommunications system(UMTS)is investigated respectively,with only statistical CSI of all users known at the base station(BS).This thesis starts with the channel model and spatial characteristics of massive MIMO system,and an introduction to BDMA transmission.Each link between a user equipment(UE)and a BS forms a jointly correlated Rician channel,on which some properties of the channel covariance matrices in the massive MIMO scenario are presented.The eigenvector matrix of the transmit correlation matrix can be constructed by a discrete Fourier transform(DFT)matrix as base stations are equipped with large scale uniform linear array(ULA).DFT is the way to perform beam domain transform,after which channels are uncorrelated.It is also proved that the statistical CSI in the beam domain is independent of sub-carriers.To cope with the instantaneous CSI acquisition,a closed-form upper bound of the ergodic achievable rate is utilized as the objective function,and the asymptotically necessary and sufficient conditions for input covariance matrices are then obtained,which identifies beam domain transmission is optimal,and maximal upper bound can be obtained by transmitting signals to different users with different beams.Numerical results validated the higher spatial resolution of the large scale antenna array and the character that channel gains are concentrated in a small amount of beams,which paves the way for beam division multiple access transmission for massive MIMO communications.Subsequently,based on the aforementioned theoretical results,the power allocation problem for downlink multicell massive MIMO communications over Rician fading channels is studied.With beam domain transmission where optimal power allocation requires the transmitted beams of different UEs to be nonoverlapping,the policy for allocating power among these beams is investigated.The objective function of the power allocation problem turns out to be a difference of concave functions(d.c.),thus the optimal solutions can be searched iteratively with concave-convex procedure(CCCP).With the assumption of perfect CSI at UEs and statistical CSI at BSs,the deterministic equivalent is utilized to approximate the sum-rate and the closed-form power allocation policy is then derived.Furthermore,the uplink statistical CSI acquisition method is provided,from which downlink statistical CSI can be acquired due to reciprocity of uplink and downlink for both frequency division duplexing(FDD)and time division duplexing(TDD)systems.The performance of the proposed algorithm,compared with that of the power allocation scheme maximizing the upper bound,the deterministic equivalent-based algorithm for non line of sight(NLOS)conditions and the beam scheduling algorithm,was demonstrated in the numerical results.Finally,the application of massive MIMO BDMA transmission to wideband code division multiple access(WCDMA),under statistical CSI at the BS,is investigated.For the downlink system where orthogonal variable spreading factor(OVSF)codes carve up the signaling dimensions orthogonally,there is a hard limit on the number of orthogonal channels that can be acquired.In order to mitigate the capacity restriction caused by such code shortage,the scheme to assign the same channelization code to more than one user is considered.Starting with investigation of the physically motivated beam domain channel model,it is first shown that transmitted beams of UTs sharing the same code channel should be non-overlapping under the criterion of sum-rate maximization in single-path propagation.With BDMA transmission,the achievable ergodic rate of each user totally depends on the beams allocated to them and an approximation which only relies on the statistical CSI is derived under the assumption of only statistical CSI at the BS.Based on this approximation,the beam selection problem with equal power allocation is analyzed,and it is proved that assigning spatially separated users with the same code channel is optimal.Motivated by this condition,spatially-overlapping users can be grouped into the same cluster via hierarchical clustering and a low complexity user scheduling algorithm which assigns users in the same cluster to different code channels is proposed.Based on the statistical reciprocity of uplink and downlink,a method to acquire channel coupling matrices(CCMs)which are required for clustering and scheduling via the uplink control channel is also provided.Simulation results verify the effectiveness of WCDMA with massive MIMO BDMA technology,and the proposed user clustering scheme provides significant performance gains over the random method in terms of achievable sum-rate for both ULA and uniform planar array(UPA)scenarios.