Research On User Selection Algorithm In Multi-user MIMO Downlink System

Multiple-Input Multiple-Output (MIMO) is a multi-antenna technology. Byconfiguring multiple transmitting antennas and receiving antennas, MIMO can getspatial diversity and multiplexing gain. It can be proved that MIMO can multiplyincrease capacity without providing additional transmission power and bandwidth.So it gets extensive attention in recent decades. Compared with Single-user MIMO,Multi-user MIMO allows the base station (BS) communicate with multiple users atthe same time, which can further improve capacity. However, in the Multi-userMIMO system, multi-user interference has been an urgent problem.Precoding at the transmitter can solve multi-user interference effectively,however, faulty user selection will cause serious decrease of capacity when thenumber of user is more than the number of antennas of the BS. This article mainlystudies the user selection algorithm in the multi-user MIMO downlink system. Wehope to get better system performance with appropriate user selection algorithm.Firstly, the article introduces the theoretical knowledge of single-user andmulti-user MIMO, derives the capacity, presents the existing precoding technologyand user selection algorithm, and analyzes the performance of this user selectionalgorithm briefly. Secondly, assuming the BS knows perfect channel stateinformation(CSI) of each user, this article gets the conclusion that the lower limit ofsum capacity is proportional to the product of effective channel gain when the BSutilizes Zero-Forcing Beamforming, and proposes a maximum product of effectivechannel gain(MPECG) algorithm, then compares MPECG with all space searchalgorithm, semi-orthogonal user selection(SUS) algorithm, and minimum of theFrobenius norm of the pseudo-inverse(MFNPI) algorithm by simulation. Thesimulation results show that all user selection algorithms can improve sum capacitywhen increasing the number of antennas of the BS; compared with SUS, MPECGcan achieve better the system sum capacity at high signal-noise-ratio(SNR) or thenumber of users is large; compared with MFNPI, MPECG slightly increases thesystem sum capacity and lower computational complexity. Finally, this articleresearches the user selection scheme in the limited feedback system, proposes userselection schemes based on sum capacity and proportional fairness scheduling(PFS),and analyzes the system sum capacity and the user fairness with different feedbackbits and the number of users. The simulation results show that, increase in thenumber of users and feedback bits can improve the system sum capacity. However,with their linear growth, the system sum capacity tends to be saturated. That is, number of users and feedback bits cannot improve the capacity significantly. Inaddition, compared with the user selection scheme based on sum capacity, PFSscheme can get the tradeoff between the system sum capacity and the user fairness.It can be also seen that, when the number of user is small, we can use few bits toachieve the user fairness and save the system bandwidth, and increasing the numberof feedback bits can improve the user fairness when the number of users is large.