Professional Documents
Culture Documents
net/publication/317845033
CITATIONS READS
12 3,257
6 authors, including:
Some of the authors of this publication are also working on these related projects:
Wireless BANK SURVEILLANCE SYSTEM WITH MONEY TRACKER (WIBANK) View project
All content following this page was uploaded by Olawoyin Abiodun on 11 October 2017.
1 Introduction
For an efficient and effective radio access network planning, a knowledge of the
characteristics of radio wave propagation in a built-up environment is required [1, 2].
Due to the medium of propagating Electromagnetic (EM) waves and the antecedent
effect such as Reflections, diffractions, and scatterings which are dominant during
propagation in urban areas as a results of physical obstructions (clutter) in the propa-
gation environment. The interaction of transmitted EM waves with building walls,
surfaces of bill boards and other artificial structures, and the bodies of moving objects
usually results in reception of multiple copies of the transmitted signals at the receivers
which can be termed as multipath [3]. The multiple copies reach the mobile station
from different directions with different time delays, leading to signal fading.
Consequently, the complex nature of urban environments is responsible for the random
and linear time-varying characteristics of the radio channel [4].
Signal fading in wireless communication may be at small scale or large scale [3].
Small-scale signal fading occurs as a result of multipath with an effect of rapid variation
in the received signal strength with time [5]. On the other hand, large-scale signal
fading is the attenuation of the mean signal power by virtue of the position of the
receiver relative to the transmitter. This is also known as path loss [6]. Since radio
engineers do not have control over the man-made environment, it is important to
properly quantify the resulting power loss for an efficient radio network.
Path loss prediction models are mathematical formulation of the propagation
channel as a function of the separation distance between the transmitter and the
receiver, the frequency of transmission, the antenna heights of both transmitter and
receiver, and other environmental factors. Radio network planners rely on path loss
prediction models to ensure an acceptable Quality of Service (QoS) and satisfactory
customer experience. In order to avoid incessant call drops and other network con-
nectivity issues, the received signal level at the receiver must be greater than the
minimum sensitivity of the receiver. The reception sensitivity varies with different
types of receiver. Estimation of signal path loss is one of the main components of the
link budget of a wireless communication system [3].
Radio propagation models have been broadly categorized into two: deterministic
and empirical models. Deterministic models [7–10] have proven to be more accurate
but they require a detailed information about the propagation environment while
empirical models [11–13] are simple and requires less computation effort. They are
formulated based on extensive measurement campaigns which makes them to be highly
environment-dependent. Although several empirical path loss models have been pro-
posed in the literature [13–22], it is still very difficult to synthesize a global model
which is suitable for every built-up area. The use of an empirical path loss model for
wireless network design in an environment other than the one it was intended for will
produce significant prediction errors. Large prediction errors will consequently lead to
poor network coverage as RF engineers are much likely to situate the base stations at
inappropriate locations.
In this paper, we seek to investigate the prediction accuracy of widely available
empirical path loss prediction models in built-up environment of Lagos, Nigeria.
Measurement campaigns were conducted to collect received signal strength data over
commercial base stations operating at 1800 MHz. In addition, we proposed an
adaptable empirical model for path loss predictions by tuning the parameters of SPM to
give a better representation of the local environment.
The remaining part of the paper is organized as follows: Sect. 2 reviews the
available empirical path loss models suitable for 1800 MHz frequency band; Sect. 3
describes the materials and the methodology used in this study; Sect. 4 gives the results
and discusses the findings while the paper is concluded in Sect. 5.
SPM Tuning for Path Loss Predictions in Built-Up Environments 365
The correction factors for mobile antenna height in a built-up environment is given by
Eqs. (2) and (3)
where
fc = Frequency (in MHz) from 150 MHz to 1500 MHz
ht = Effective transmitter antenna height in meters
hr = Effective receiver antenna height in meters
d ¼ Tx Rx separation distance in km
The path loss for a suburban area is given by Eq. (4)
fc 2
PLsuburban ¼ PLurban ðdBÞ 2½logð Þ 5:4 ð4Þ
28
For an open rural area, Eq. (5) is used
PLðdBÞ ¼ 46:3 þ 33:9 logðfc Þ 13:82 logðht Þ aðhr Þ þ ½44:9 6:55 logðht Þ
logðdÞ þ Cm
ð6Þ
366 S.I. Popoola et al.
where
Afs = free space attenuation
Abm = basic median path loss
Gt = transmitter antenna height gain factor
Gr = receiver antenna height gain factor
A3 = −13.82
B1 = 44.90
B2 = −6.55
B3 = 0
Therefore, the path loss model for GSM technologies that operate in the 900 MHz
band becomes Eq. (19)
PLðdBÞ ¼ 69:55 þ 26:16 logð f Þ 13:82 logðht Þ þ ½44:9 6:55 logðht Þ½logðd Þ
aðhr Þ Cclutter ð19Þ
On the other hand, the path loss predictions for the DCS 1800 counterpart is given
by Eq. (20) [9]:
PLðdBÞ ¼ 46:3 þ 33:9 logð f Þ 13:82 logðht Þ þ ½44:9 6:55 logðht Þ½logðd Þ aðhr Þ
Cclutter
ð20Þ
SPM ignored the effects of diffraction, clutter, and terrain to produce Eq. (21). It
assumed that appropriate settings of A1 and K1, which account for only one clutter
class, will cater for the influence of these external factors on signal propagation. The
correction function for the mobile receiver antenna height was also ignored for
hr 1:5 m since it has negligible values for an average mobile antenna height. The
resulting path loss model is given in Eq. (5):
SPM Tuning for Path Loss Predictions in Built-Up Environments 369
Presenting the reduced Hata equation as a model which accepts distance input in m as
in SPM, we have Eq. (7):
A maximum vehicle speed of 40 km/h was ensured during the survey to eliminate
Doppler effects. The measurements were taken under normal climatic conditions to
ensure high-quality data collection. Also, a good radio frequency clearance was
maintained.
where
HORx = the ground elevation above sea level at the receiver (m)
K = the ground slope calculated over a minimum distance from the receiver
The diffraction loss over the transmitter-receiver profile was calculated using the
Millington method. The average clutter height specified for each clutter class was used
for the transmitter-receiver profile in the calculation of the diffraction edge. The losses
due to clutter was estimated over a maximum distance from the receiver using Eq. (26):
Xn
f ðclutterÞ ¼ i¼1
Li wi ð26Þ
SPM Tuning for Path Loss Predictions in Built-Up Environments 371
where,
L = loss due to clutter (in dB).
w = weight determined through the weighting function.
n = number of points taken into account over the profile.
Weighting function is the mathematical formula used to calculate the weight of the
clutter loss on each pixel from the pixel with the receiver in the direction of the
transmitter, up to the defined maximum distance. Logarithmic weighting function
employed is given by Eq. (27):
logðdi þ 1Þ
wi ¼ Pn D dj ð27Þ
j¼1 logðD þ 1Þ
where, di is the distance between the receiver and the ith point and D is the maximum
distance defined. The maximum distance indicates the distance from the receiver for
which clutter losses will be considered via a weighting function, with an effect on the
influence of clutter on total losses which diminishes with distance from the receiver.
This value was set to 150 m for the urban environment.
The model tuning was carried out using ATOLL planning tool with the aim of
minimizing the mean error and standard deviation of measured values versus calculated
values. A mathematical solution was obtained for both Line of Sight (LOS) and
Non-Line of Sight (NLOS) conditions.
1 Xn p
MAE ¼ ðPLm i PLi Þ ð28Þ
n i¼1
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 Xn p 2
RMSE ¼ ðPLm i PLi Þ ð29Þ
n i¼1
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1 Xn p 2
SD ¼ ðjPLm i PLi j lÞ ð30Þ
n i¼1
Figures 1 and 2 provide the comparison of path loss prediction at Locations A and B
respectively. In both Figures, Egli model under-estimates the path loss and ECC-33
over-estimates the path loss. The Okumura-Hata and COST 231 models are the base
line SPM, these models provides good fitness at distances above 400 m, below which
they underestimate the path loss. However, the tuned SPM provides better predictions
along all the routes, across all distances. Table 1 provides the tuned parameters for the
tuned model. The SPM ignored the effects of diffraction, clutter, and terrain. This
assumes single clutter class, while, the turned model, considers the diffraction losses.
This increases the clutter losses from 0 dB in the case of conventional SPM to 50 dB as
in the turned model. The tuned SPM is expressed mathematically in Eq. (31) and (32).
Equation (31) is applicable to LOS scenarios while Eq. (32) can be used for NLOS
propagation scenarios in built-up areas.
Tables 2 and 3 provide the statistical evaluation of path loss at Locations A and B. For
Location A, RMSE values of 13.75 dB, 12.48 dB, 14.12 dB and 38.21 dB respectively
for Hata, COST 231, ECC-33 and Egli models. These are higher than the acceptable
6 dB threshold values defined in [7]. While, for Location B, Hata, COST 231, ECC-33
and Egli modes obtained RMSE values of 15.60 dB, 14.42 dB, 14.45 dB and
37.18 dB respectively, wit COST 231 model providing the last error amongst the four
contending models, across the locations. It worth noting that despite the high prediction
error of the ECC-3 model across the locations, the model has the least standard
deviation error. However, the tuned SPM achieves RMSE of 5.92 dB and 5.72 dB for
the two locations. These values are 47.4% and 39.6% decrease in prediction error when
compared with COST 231 model along the locations.
5 Conclusion
This paper presents measurement campaigns of EM waves in the GSM 1800 MHz
band within built-up areas. A simple optimization procedure using ATOLL planning
tool for standard propagation model is provided. The optimization process involves the
use of DTM, clutter classes, clutter heights, vector maps, scanned images, and WMS.
A Logarithmic weighting function was used to calculate the weight of the clutter loss
on each pixel from the pixel with the receiver in the direction of the transmitter, up to
the defined maximum distance. The approach has proven promising by achieving high
accuracy and minimizing the prediction errors by 47.4%.
Acknowledgment. The authors wish to appreciate the Center for Research, Innovation, and
Discovery (CU-CRID) of Covenant University, Ota, Nigeria for partly funding of this research.
References
1. Lempiäinen, J., Manninen, M.: Radio Interface System Planning for GSM/GPRS/UMTS.
Kluwer Academic Publishers, Dordrecht (2001)
2. Laiho, J., Wacker, A., Novosad, T.: Radio Network Planning and Optimisation for UMTS.
Wiley, New York (2002)
3. Rappaport, T.S.: Wireless Communications: Principles and Practice. Prentice Hall PTR,
Upper Saddle River (2002)
4. Parsons, J.D.: The Mobile Radio Propagation Channel, 2nd edn. Wiley, New York (2000)
5. Clark, R.H.: A Statistical Description of Mobile Radio Reception. BSTJ 47, 957–1000
(1968)
6. Mishra, A.J.: Advanced Cellular Networks Planning and Optimization 2G/2.5G/3G…
Evolution to 4G, pp. 1–12. Wiley, New York (2007). ISBN 13 978-0-470-01471-4
7. Luebbers, R.J.: Propagation prediction for hilly terrain using GTD wedge diffraction. IEEE
Trans. Antennas Propag. 32(9), 951–955 (1984)
8. Mohtashami, V., Shishegar, A.A.: Modified wavefront decomposition method for fast and
accurate ray-tracing simulation. Microw. Antennas Propag. IET 6(3), 295–304 (2012)
9. Hufford, G.A.: An integral equation approach to the problem of wave propagation over an
irregular surface. Q. J. Mech. Appl. Math. 9(4), 391–404 (1952)
10. Zelley, C.A., Constantinou, C.C.: A three-dimensional parabolic equation applied to
VHF/UHF propagation over irregular terrain. IEEE Trans. Antennas Propag. 47(10), 1586–
1596 (1999)
SPM Tuning for Path Loss Predictions in Built-Up Environments 375
11. Masahara, H.: Empirical formula for propagation loss in land-mobile radio services. IEEE
Trans. Veh. Technol. 29(3), 317–325 (1980)
12. COST 231 Project: Urban Transmission Loss Models for Mobile Radio in the 900 and
1800 MHz band, COST 231 TD (90) 119 Rev. 2, The Hague, Netherlands (1991)
13. Popoola, S.I., Oseni, O.F.: Performance evaluation of radio propagation models on GSM
network in urban area of Lagos, Nigeria. Int. J. Sci. Eng. Res. 5(6), 1212–1217 (2014)
14. Popoola, S.I., Oseni, O.F.: Empirical path loss models for GSM network deployment in
Makurdi, Nigeria. Int. Refereed J. Eng. Sci. 3(6), 85–94 (2014)
15. Oseni, O.F., Popoola, S.I., Abolade, R.O., Adegbola, O.A.: Comparative analysis of
received signal strength prediction models for radio network planning of GSM 900 MHz in
Ilorin, Nigeria. Int. J. Innov. Technol. Explor. Eng. (IJITEE) 4(3), 45–50 (2014)
16. Faruk, N., Ayeni, A.A., Adediran, Y.A.: Error bounds of empirical path loss models at
VHF/UHF bands in Kwara State, Nigeria. In: Proceedings of IEEE EUROCON Conference,
Croatia, pp. 602–607 (2013)
17. Faruk, N., Adediran, Y.A., Ayeni, A.A.: On the study of empirical path loss models for
accurate prediction of TV signal for secondary users. Prog. Electromagn. Res. (PIER) B
USA 49, 155–176 (2013)
18. Rath, H.K., Verma, S., Simha, A., Karandikar, A.: Path loss model for Indian terrain -
empirical approach. In: 2016 Twenty Second National Conference on Communication
(NCC), Guwahati, pp. 1–6 (2016)
19. Al Salameh, M.S.H., Al-Zu’bi, M.M.: Prediction of radio wave propagation for wireless
cellular networks in Jordan. In: 2015 7th International Conference on Knowledge and Smart
Technology (KST), Chonburi, pp. 149–154 (2015)
20. Nimavat, V.D., Kulkarni, G.R.: Simulation and performance evaluation of GSM propagation
channel under the urban, suburban and rural environments. In: 2012 International
Conference on Communication, Information and Computing Technology (ICCICT),
Mumbai, pp. 1–5 (2012)
21. Ibhaze, A.E., Ajose, S.O., Atayero, A.A.A., Idachaba, F.E.: Developing smart cities through
optimal wireless mobile network. In: 2016 IEEE International Conference on Emerging
Technologies and Innovative Business Practices for the Transformation of Societies
(EmergiTech), Balaclava, pp. 118–123 (2016)
22. Ibhaze, A.E., Imoize, A.L., Ajose, S.O., John, S.N., Ndujiuba, C.U., Idachaba, F.E.: An
empirical propagation model for path loss prediction at 2100 MHz in a dense urban
environment. Indian J. Sci. Technol. 8(1) (2017)
23. Abhayawardhana, V.S., Wassell, I.J., Crosbsy, D., Sellars, M.P., Brown, M.G.: Comparison
of empirical propagation path loss models for fixed wireless access systems. In: IEEE
Vehicular Technology Conference, Spring, vol. 1, pp. 73–77 (2005)
24. TEMS: Testing, Monitoring, and Analytics Software, Ashburn, VA, USA. www.tems.com
25. Pitney Bowes: MapInfo Pro. http://www.pitneybowes.com/us/location-intelligence/
geographic-information-systems/mapinfo-pro.html
26. Forsk: ATOLL 3.2.0 Radio Planning and Optimization Software, France. www.forsk.com