Convergence Analysis of Massive MIMO Antenna Arrays Using a Geometry-Based Stochastic Channel Model
No Thumbnail Available
Date
2019-10
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
KNUST
Abstract
In order to reap the full benefits of massive MIMO, the Base Station (BS) antennas must be significantly large to converge to favorable propagation condition or
attain convergence (the state where the effect of noise and fast fading vanishes).
However, increasing BS antennas results in closely spaced antenna elements which
inadvertently leads to the detrimental effects of Spatial Correlation (SC) and that
can affect the convergence of Massive MIMO System. So far, research works on
Convergence have been investigated using Correlation-based Stochastic (CBSCM)
channel models, which does not reflect accurate massive MIMO channel. This
channel model does not take into consideration channel parameters such as the
Pathloss, power delay profile as well as the characteristics of the antenna array needed for the practical massive MIMO system implementation. Therefore
analysis of massive MIMO convergence regarding the Geometry-Based Stochastic
Channel model (GBSCM), which reflects real practical massive MIMO channel
and the effects of SC is needed. In this thesis, the convergence of massive MIMO
based on GBSCM is studied. The effect of 3D SC of uniform rectangular and
cylindrical array (URA and CA) based on the Maximum Power of Arrival (MPA)
is incorporated. In the analysis, the Diagonal Dominance and Mean Absolute
Deviation convergence metrics, were considered to study the effects of the SC
on the asymptotic behavior of the channel matrix. The results support existing development that, increase in azimuth and elevation spreads of the angular
distributions of arrival and antenna element spacing reduces the SC and thereby
increases convergence rate of massive MIMO. Results also show poor performance
for the Laplacian and Von Misses distributions in GBSCM, even for higher AS
and ES, and angular spreads. Further it was realized that, in GBSCM, the convergence of massive MIMO was sensitive to antenna array topology regarding
Gaussian and Student’s t-distributions. This contradicts previously established
results that convergence is insensitive to antenna topology in CBSCM.
Description
A Thesis submitted to the Department of Telecommunications Engineering College of Engineering
in partial fulfillment of the requirements for the degree of MPHIL. TELECOMMUNICATIONS ENGINEERING