Académique Documents
Professionnel Documents
Culture Documents
Abstract—Considering the rising popularity and rapid from well known outdoor propagation models is also
implementation of Smart City environments, evaluation and discussed.
testing of wireless technologies performances in those This paper is organized as follows: After the
environments become reasonably important. The introductory section, a variety of propagation models are
importance increases with the variety of existing, as well as discussed as well as the related work in propagation model
emerging wireless technologies. Besides well known and application, tuning and comparison. In the initial phase,
widely used technologies such as: Wi-Fi, ZigBee, 3G/4G/5G; three propagation models are indentified as very suitable
in recent years new types of technologies have been for further research. The further research has the goal to
targeting the market. Those technologies are LoRa, find a mathematical model for as accurate prediction of
LoRaWAN and SigFox, also known as Low Power Wide
wireless smart city technologies performance as possible
Area Network (LPWAN). In this paper an approach in
in the sense of accurate signal strength prediction in the
exploring performances of various newer and older
first place. Those models should be applicable for
technologies is presented. In this particular case ZigBee and
LoRa communication modules are tested. The approach is
frequencies of 868MHz (LoRa) and 2.4GHz (ZigBee).
based on the low-cost and open-source hardware and uses
Those three models are Okumura-Hata, Stanford
all its advantages. Besides equipment, research methodology
University Interim (SUI) model and Ericsson 9999. In the
and field measurements are also presented. The possibility following section the measurement devices built around
of comparing field measurement results with the calculated open-source hardware platforms are presented, as well as
values obtained from well known outdoor propagation the method of their utilization and obtained results. The
models is discussed. The accuracy and applicability of next section gives the results analyses, together with the
prediction models are also discussed together with the comparison of the calculated values with the usage of
possibility of tuning those models using field measurements. presented models.
Attenuation [dB]
140
Seaside and lakeside areas etc. 130
0 5 10 15 20 25 30
Distance d [km]
The parameters are as follows:
fc - frequency in MHz in range between 150 MHz Fig. 1. Comparison of propagation loss calculation for 868MHz for five
and1,500MHz, different environments
Difference between signal strength in large cities and free space - 868 MHz
hb - effective transmitter antena height in meters, in 50
42
km, in range between 1km to 20km 40
a(hm) - correction factor for effective antenna heightas a 38
function of dimension of covered area. 36
IV. RESULTS
Presentation of the measurement results is given in
three tables. Table III shows basic data about start and
end time of measurements, measurement duration in
seconds and minutes.
TABLE III. MEASURMENTS TIME AND MEASURING DURATION DATA FOR
Fig. 6. Transmitter ZigBee station with the addition of GPS
LORA MEASURMENTS
module for location detection
Start Time End Time Duration (s) Time
Loc_01 16:22:56 16:36:36 820 0:13:40
Six measuring locations are selected with the ZigBee
Loc_03 17:14:02 17:27:09 788 0:11:13 modules: four locations in the wider range of city center
Loc_05 18:08:30 18:22:05 814 0:12:58 and two locations within the city center. Unfortunately, at
Loc_12 13:59:24 13:44:32 892 0:14:52 the time of the experiment only omni-directional 2 dBi
Loc_13 14:43:02 14:28:40 862 0:14:22 and 5 dBi antennas were used on the transmitter and the
Loc_14 15:06:10 14:51:45 865 0:14:25
receiver side. The platform is optimized for giving GPS
Table IV shows the basic data about the number of location of the measurement, and measuring PER values,
packets sent and the number of packets received. From but not for retrieving RSSI values as well (like LoRa
these two values the PER (Packet Error Rate) is platform). The two locations in the city center are shown
calculated. The fourth column represents the height of in Fig. 7.
transmitting station which was mounted during the Since this platform in this phase of research is not
measurement on given height in centimeters. adjusted for retrieving RSSI values, the measurements
cannot be used for comparing these values with the
TABLE IV. NUMBER OF PACKETS, PER AND TRANCIEVER ANTENNA
HEIGHT DATA FOR LORA MEASURMENTS values calculated using the prediction models. Still,
during the measurement, the platform proved to be
Pkts. Sent PktsRcv. PER (%) Height (cm)
efficient and very usable for measuring, and further
Loc_01 530 528 99.44 175
Loc_03 510 510 100.00 175 development directed to enabling RSSI value logging will
Loc_05 527 455 86.34 175 be continued. So, for further research, after adjustment of
Loc_12 577 530 91.85 160
the platform, common propagation loss model supporting -50
2.4 GHz frequencies will be important as well. -60
medium/small cities
large cities
The results of the measurement with ZigBee based sub-urban
-70 rural
platform are published in the [13], with Packet Error measured values
-90
-100
-110
-120
-130
-140
0 1 2 3 4 5 6
Fig. 7.Measuring locations in the Zrenjanin city center for B. Ericsson 9999 model comparison
ZigBee modules The second comparison of measured results is made
with Ericsson 9999 model for 868MHz. The prediction is
V. RESULTS AND DISCUSSION
made for urban, suburban and rural areas. The best fitting
Calculations needed for propagation loss are made is with the Ericsson 9999 model for urban areas.
according to the formula for received power calculation. Ericsson 9999 model is an extension of Hata model and
Received power calculation or link budget may be done it is applicable for 1900MHz, and therefore it should not
with (15) and it is based on the authors’ experience in [9]. be suitable for predicting ZigBee network propagation
The link budget is a tabulation of all gains and losses for (2,400 MHz). Nevertheless, after retrieving RSSI values
the link added in order to arrive at the mean signal level at for ZigBee network with the previously presented
the receiver. [12] platform, Ericsson 9999 model should not be left out and
Prx=Ptx+Gtx-Ltx-Lpl-Lm+Grx-Lrx (15) should be compared if not in its original form, than in the
form tuned with the retrieved empirical measurements.
Where:
Ericsson 9999 Model -Urban
Prx is received power (dBm) -20
urban
Ptx is transmitter output power (dBm) -40 sub-urban
rural
Gtx is transmitter antenna gain (dBi) -60
measured values
-80
Ltx is transmitter losses (coax, connectors...) (dB)
Signal Level [dB]
-100
Lpl is propagation loss or path loss (dB)
-120
Lm is miscellaneous losses (fading margin, body loss,
-140
polarization mismatch, other losses...) (dB)
-160
Grx is receiver antenna gain (dBi)
-180
Lrx is receiver losses (coax, connectors...) (dB)
-200
are very low and therefore disbanded. Fig.9. Ericsson 9999 model prediction for 868MHz compared with
LoRa measured results
In this section, the Okumura-Hata model accuracy in
predicting LoRa measurement is analyzed first. C. Stanford University Interim (SUI) model comparison
A. Okumura-Hata model comparison The third comparison is made with Stanford University
Interim (SUI) model. It is shown in Fig. 10.Three terrain
Comparison of Okumura-Hata for large and type prediction calculations are used for the comparison,
small/medium cities, as well as suburban and rural areas is for terrain types A, B and C.
presented in Fig. 8. These results are presented together
with the results obtained from field measurements with In all three cases prediction calculations based on the
LoRa stations. For Fig. 8 it should be noted that large and model are not fitted well with measured LoRa values. So,
small/medium cities curves are almost identical. The best in this case, SUI model showed itself inapplicable. SUI
fitting using Okumura-Hata model is made with formula model should not be excluded from future work, due to
for large and small/medium cities; with exception of the fact that it is designed to be used up to and above 2.4
measurements for distances of around 5 km between the GHz as well. Its future usage should include model
transmitter and the receiver which better fit suburban adjustment as well.
areas.
Stanford University Interim (SUI) Model - Terrein A, B, C of10th Advanced International Conference on
0
Terrein A
Telecommunications AICT2014, July 20 - 24, Paris, France, 2014.
-20
Terrein B [6] J. Milanovic, S. Rimac-Drlje and K. Bejuk, “Comparison of
Terrein C
Measured values
Propagation Model Accuracy for WiMAX on 3.5GHz”, 14th IEEE
-40 International Conference on Electronic Circuits and Systems,
Morocco, (2007), pp. 111-114. 2007.
Signal level [dB]