Spatial and Temporal Traffic Distribution Modeling Using Statistical Techniques: The Case of UMTS Network in Addis Ababa, Ethiopia
No Thumbnail Available
Date
2019-12
Authors
Woinshet, Taddele
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
In cellular mobile networks due to the emerge of different application services, the data
traffic generated by these application services is increasing. Moreover, the traffic consumption
pattern
is
changing
in
time
and
space.
Understanding
and
characterizing
the
traffic
demands
and
dynamics
associated
with
different
mobile
services
(e.g.,
knowing
when
and
where
users
use
dominant
application
services)
is
instrumental
for
operators.
By
doing
so,
operators
can
improve
application
services
usability,
optimizing
network
service
quality
and
use
it
as
an
input
for
technical
and
business
strategies.
In
the
case
of
ethio
telecom,
the
sole
telecom
service
provider
in
Ethiopia,
knowledge
of
the
application
services
distribution
is
not a
known
practice.
Therefore,
the main purpose of this thesis work is modeling the spatial and temporal
traffic distribution of observed application services for Addis Ababa Universal Mobile
Telecommunication System (UMTS) network. In this regard, to study a temporal and
spatial distribution analysis of the traffic density (the traffic load per unit area) for selected
mobile
application
services,
data
traffic
is
collected
from 739
bases
Stations
(BSs).
To model the temporal distribution of the selected application services, which are Streaming,
Social
Networks
and
Web
browsing,
four candidate models: Normal, Lognormal,
Weibull and Stable are used. Based on maximum log-likelihood and probability plot evaluation
criteria, out of four models, a Stable distribution modeling best fit for this
particular application services. Similarly, to model spatial distribution, out of the candidate
four
models,
Weibull
distribution
model
best
fits
for
these
particular
services.
Description
Keywords
UMTS, Modeling, Application Services, Temporal, Spatial, Statistical Modelling, Weibull, Normal, Lognormal, Stable distribution