Comparison and Fine Tuning Empirical Pathloss Models at 1800MHZ and 2100MHZ Bands for Addis Ababa, Ethiopia
No Thumbnail Available
Date
2018-10
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
AAU
Abstract
Pathloss models play a very important role in wireless communications in coverage planning,
interference estimations, frequency assignments, Location Based Services (LBS), etc. They are
used to estimate the average pathloss a signal experience at a particular distance from a
transmitter. Inaccurate propagation models may result in poor coverage, poor quality of service
or high investment cost. Both second generation (2G) and third generation (3G) networks in the
city of Addis Ababa (AA), Ethiopia, have problems like poor coverage, low data throughput, call
drops and others. One of the root causes of these problems is the use of untuned pathloss
model during network planning. So, it is mandatory for the operators to select the best fit
pathloss model and tune it according to the specific situation the pathloss model is used. This
thesis compares three pathloss models; namely, COST231, ECC-33 and SUI and tunes the one
that performs best in the specific area type.
At 1800MHZ band, COST231 was best in estimating the measured path loss in urban areas with
a Root Mean Squared Error (RMSE) of 3.27dB before tuning and the RMSE could improve to
3.25dB after tuning. COST231 was also best in suburban areas with an RMSE of 5.27dB. Tuning
the model could improve the RMSE to 4.18dB. SUI was best in open areas. It has an RMSE value
of 6.0dB before tuning. Tuning has improved the RMSE value to 4.91dB.
At 2100MHZ, 25 sites are used to collect path loss data. Similar analysis was done in the three
path loss models. Based on the analysis, SUI is found to be best in predicting the path loss for all
the three morphology types. Although ECC-33 was equally competent for urban area sites, SUI
could predict the path loss better for the overall all average measured path loss with an RMSE
of 4.27dB. Tuning the model could improve the RMSE to 2.23dB. The measured path loss for
suburban areas could also be better predicted by SUI with an RMSE value of 5.75dB before
tuning and 2.57dB after tuning. Path loss in open areas can also be better predicted by SUI. It
has an RMSE value of 6.53dB before tuning. An improvement in RMSE to 3.38dB could be
achieved after tuning.
Description
Keywords
Path loss, tuning, prediction, modelling, error, urban, suburban