See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/320445434

Performance analysis of LoRa radio for an indoor IoT applications

Conference Paper · July 2017

DOI: 10.1109/IoTGC.2017.8008973

CITATIONS READS
9 3,151

1 author:

Eyuel Debebe Ayele

University of Twente
13 PUBLICATIONS   68 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Smart Park View project

All content following this page was uploaded by Eyuel Debebe Ayele on 17 October 2017.

The user has requested enhancement of the downloaded file.

 Performance Analysis of LoRa Radio for an Indoor
IoT Applications
Eyuel D. Ayele, Chiel Hakkenberg∗ , Jan Pieter Meijers, Kyle Zhang, Nirvana Meratnia, Paul J.M. Havinga

Pervasive Systems Research Group, University of Twente, Enschede, the Netherlands

{e.d.ayele, j.p.meijers, k.zhang, n.meratnia, p.j.m.havinga}@utwente.nl

∗
[email protected]

Abstract—LoRa is an emerging wireless standard specifically spread-spectrum technique operational in the sub-GHz fre-
designed for Low Power Wide Area Networks (LPWANs). quency spectrum such as 433MHz, 868MHz, and 915MHz,
It provides long range, low data rate, and energy efficient which have less interference with other frequency bands such
wireless communication and is believed to have high potential
for realization of a large number of Internet of Things (IoT) as 2.4GHz used by WiFi and Bluetooth [3]. In principle in
applications. Various research papers have already reported on these sub-GHz bands, signals are more robust against noise
the performance analysis of LoRaWAN protocol in terms of radio while drawing relatively low energy. This makes them ideal
communication range, and reliability for outdoor environments, for many IoT applications.
while performance analysis for indoor environments have not yet
received enough attention. In this paper, we provide an in-depth Since the release of the first version of the LoRaWAN
performance evaluation of LoRa for indoor IoT applications. specification by the LoRa Alliance [3], some research papers
Index Terms—LPWAN, LoRa, LoRaWAN, Internet of Things, reported on the performance analysis of LoRaWAN protocol
RSSI, performance analysis in terms of radio communication range, and reliability. For
instance, Wendt et. al [5] presented their analysis of LoRa
I. I NTRODUCTION radio’s signal wall penetration in an indoor environment by
It is expected that by 2020 tens of billions of battery- observing the RSSI values. The capacity and scalability of
powered data generating end-devices will be connected to LoRaWAN were studied by [6, 7]. In [8], the authors have
the Internet [1]. The limited energy resource of these end- evaluated LoRa’s communication range and developed an
devices necessitates having low-power communication proto- outdoor propagation model. They stated to have reached a
cols in place. To this end, the low-power wide-area networks communication range of 15 km in air. They also developed
(LPWANs) have been designed to ensure a very long battery a channel attenuation model for estimating the communica-
lifetime, specifically for applications requiring small amount tion distance between LoRa end-devices and LoRa gateways.
of data to be transmitted over very long geographical distances. Brecht et. al [9] compared LoRaWAN with other long-range
There exist various LPWAN-based technologies and solutions, unlicensed technologies in terms of communication range and
such as UNB (Ultra Narrow Band) from Sigfox and WAVIoT coexistence capabilities. They also analyzed the long-range
Narrowband M2M protocol, and LoRaWAN [2]. Among all of performance in terms of packet error rate and throughput
them, LoRa (Long-Range) is believed to have high potential without explicitly analyzing the propagation characteristics.
for realization of a large number of Internet of things (IoT) Complementary to research of LoRa performance evaluation
applications. While LoRa defines the physical layer of the for outdoor IoT applications, in this paper we focus on indoor
communication link, LoRaWAN [1, 3] defines its open stan- IoT applications. To this end, the contributions of this paper
dard communication protocol. It enables a much larger range include: (1) testing the decoding performance of LoRa radio
of machine-to-machine (M2M) and Internet of Things (IoT) receivers and comparing it with the specification defined by
applications as they are expected to tackle both the energy the LoRa Alliance [3]; (2) performance analysis of LoRa in
issue and infrastructure costs. By doing so, interconnecting terms of link quality between the LoRa end-devices and the
a very large number of end-devices to each other and to the gateway, looking specifically into packet loss and received
Internet, as envisioned by Internet of Things applications, seem signal strength (RSS); (3) proposing a suitable LoRa radio
to be feasible more than ever. configuration including the LoRa radio spreading factor to
As part of LoRaWAN connectivity efforts, Semtech [4] improve performance for indoor environments.
released the LoRa RF platform (SX127X) that complements The rest of this paper is organized as follows. LoRa protocol
M2M cellular infrastructure and provides a cost-efficient way specification and its network architecture are introduced in
to connect battery operated mobile devices to the network Section II. The indoor experimental set-up and discussion of
infrastructure. LoRa physical layer protocol implements a test results are presented in Section IV, followed by concluding
978-1-5386-2064-9/17/$31.00
c 2017 IEEE remarks in V.
 II. S PECIFICATION OF L O R AWAN PROTOCOL AND The LoRaWAN specification [3] describes three classes
NETWORK ARCHITECTURE of end-devices (i.e. Class A, Class B, and Class C). The
difference between these classes relates to the timing of the so-
called receive-window. Class A is the default LoRa end-device
Monitoring Application operation as shown in Figure 2. It allows a bidirectional link
LoRaWAN Network Server with the gateway, whereby each end-device’s upstream trans-
Secure IP Connection
mission is followed by two time-slots for a short downstream
reception. Class B end-devices allow a bi-directional link with
LoRaWAN Gatways scheduled listening time-slots, whereas Class C end-devices
allow continuous listening time-slot but the listening slot will
LoRa RF Link be closed during transmission.
LoRa End-Nodes
Depending on the data rate used, it takes a certain amount of
time to complete an upstream data transmission. This time is
called T ime−On−Air (T oA) (see Figure 2). After T oA has
passed and due to the ETSI imposed restrictive of 1% trans-
mission duty-cycle per sub-band (DutyCyclesubBand ) [10],
the end-device is not allowed to access the channel for at
least Tof f subBand = T oA × (1/(1 − DutyCyclesubBand ))
Fig. 1: Generic LoRaWAN Star Network Topology Architecture seconds.
The T oA is expressed by Equation 1, where npreamble is the
LoRa radio link is based on a proprietary chirp spread spec- number of preamble symbols npreamble =8, P L is the number
trum modulation scheme. LoRaWAN is the medium access of P HY payload bytes, and CRC and H specify the presence
control protocol of the LoRa standard. Relative to the OSI of CRC and P HY header, respectively.
reference model, LoRa represents the physical layer (layer
1) while LoRaWAN represents layer 2 and layer 3. Figure 1
shows the network architecture for setting-up a LoRaWAN  !
infrastructure. The LoRaWAN communication is achieved T oA(PL) = (Ts ) × (Tpre ) + max dΓe × (CR + 4), 0
through a simple star topology. The network architecture
includes three components, i.e. end-devices, gateways, and a (1)
network server. LoRa gateways forward the control signals and where the symbol period Ts = (2SF /BW ), the
the generated data messages of the end-devices to a central preamble length Tpre = (npreamble + 12.25), Γ =

network server and are able to decode multiple signals at the 8P L−4SF +28+16CRC−20H
4×(SF −2DE) .
same time. The LoRaWAN specification [3] defines that the
LoRaWAN specification [3] prescribes CRC = 1, H = 0
LoRa gateways are connected to the network server with an IP
for uplink, CRC = 0 and H = 0 for downlink. The Low
connectionn and the end-devices communicate to the gateway
Data Rate Optimization feature (denoted by DE) is a feature
through a single hop wireless communication. Communication
that aims to improve the robustness of the transmission to
between LoRa end-devices and LoRa gateways is bidirectional
frequency variations during the transmission of the packet on
to support services such as software upgrade, over-the-air
high spreading factors (SF s) which is obligatory for low bit
activation, and multi-casting. The network server is used as
rate settings. CR indicates coding rate ranging between 1 and
the sole manager of the network. It can manage, among
4 where CR = 1 corresponds to a 4/5 coding rate and CR =
other things, the communication data rate settings for each
4 corresponds to the maximum coding rate of 4/8.
end-device separately through the adaptive data rate (ADR)
scheme. As shown in Figure 2, for every successful transmission,
an end-device will have two receive windows, during which
it can receive downstream messages. The first receive window
ToffSubBand ToffSubBand uses the same frequency channel as the preceding upstream
(ToA)1 (ToA)n
RX1 RX2
(ToA)2 RX1 RX2 message and a data rate that is a function of the data rate
used for the preceding upstream message. By default, the data
...
rate of the first receive window is identical to the data rate
RECEIVE_DELAY1
Check Free
Channels
of the last upstream. RX1 opens RECEIV E DELAY 1
ToffSubBand = ToA(1/DutyCycleSubBand-1)
RECEIVE_DELAY2 seconds after the end of the upstream modulation. The
Upstream
Downstream second receive window uses a configurable, non-adaptive,
channel and data rate and opens RECEIV E DELAY 2
Fig. 2: LoRa channel access for Class-A end-devices, where T oA is (=RECEIV E DELAY 1 +1s) seconds after the end of the
the data frame Time-On-Air and Tof f subBand is the off channel time upstream modulation. The frequency and bit rate used can be
in seconds. modified by the LoRaWAN MAC commands. The LoRaWAN
channel access scheme is similar to the Aloha MAC [3], in Transmission Power (TXP): A high TXP will result in a
which LoRa end-devices could access the channel randomly. higher RSSI, increasing the range of reception while
allowing a lower PER.
III. L O R A PHYSICAL LAYER PARAMETERS AND THEIR Band Width (BW): A smaller BW increases the receiver sen-
EXPECTED IMPACT sitivity while lowering the noise floor, allowing a lower
When testing the performance of LoRa, it is critical to PER.
measure a variety of physical layer parameters that impact We perform measurements on all spreading factors sup-
the performance of the radio link. The three physical layer pa- ported by the LoRa end-device at 868MHz frequency band
rameters, namely, Spreading factor (SF), Bandwidth (BW),and and 125KHz bandwidth, as presented in Table II for LoRa
Coding Rate (CR), influence the effective bit rate of the radio parameter settings. The transmit power (TXP) is fixed at
modulation, its resistance to noise and interference, and its the maximum value of 14dBm for LoRaWAN end-devices for
ease of decoding. the EU868MHz band, unless explicitly defined differently. For
the bandwidth (BW) we use the default LoRaWAN value for
SF the EU868MHz band, which is 125KHz. In theory, the useful
7 8 9 10 11 12
bit rate is Rb = SF × (BW/2SF ) × CR, which shows that a
BW
higher bit rate can be achieved with a broader bandwidth. This
125kHz -123 -126 -129 -132 -133 -136
250kHz -120 -123 -125 -128 -130 -133 is however not possible on the default EU868Mhz LoRaWAN
500kHz -116 -119 -122 -125 -128 -130 channels.
TABLE I: Semtech LoRa receiver sensitivity specification in dBm for
various BW and SF settings (data sheet [3]). SF = Spreading Factor,
IV. P ERFORMANCE EVALUATION OF L O R A R ADIO IN AN
BW = Band Width. INDOOR ENVIRONMENT

Table I shows a summary of the expected receiver sensitivity

with respect to BW and SF, using LoRa radios sd reported
Hallway
in the SX1276 data sheet [3]. We aim to compare our test LoRaWAN Gateway

results with this specified sensitivity values to evaluate how the

practical implementation deviates from the expected values.
LoRa radio channel bandwidth determines the noise floor
and thereby the receiver sensitivity. It is expected that an
increase in channel bandwidth results in decrease of the G
receiver sensitivity, and the higher spreading factor the higher L4

the receiver’s sensitivity. Furthermore, an increase in coding

L1 L2 L3
rate improves the packet delivery ratio. Staircase

Office 2
ToA for 1 Byte data frame

12
827.4
LoRa End-devices

413.7
G Gateway
Spreading Factor (SF)

11

206.8
10

103.4
Fig. 4: LoRa End-nodes deployment across building hallway floor at
9
locations L1 to L4. The four locations are chosen in an increasing
51.71
8 order of transmission range from the gate way, which is located at
7
25.86
the left corner of the building.

0 100 200 300 400 500 600 700 800 900

Time-On-Air (ToA) (ms) To evaluate performance of the LoRaWAN protocol in an
Fig. 3: Time-on-air (ToA) (frame duration) for various spreading indoor environment, we built a LoRa radio enabled end-device
factor (SF) settings. For higher SF results in higher ToA using RN2483 and Arduino Uno modules and mounted the
LoRa radio module on top of a custom made Arduino shield
Theoritically, the expected impact of the four LoRa physical equipped with an antenna. The prototyped end-device is shown
layer parameters can be summarized as follows: in Figure 4.
Spreading Factor (SF): A high SF should be easily decod- The receiving node of the network (the LoRa gateway)
able, resulting in a high T ime−on−Air (Figure 3), lower is a MultiConnect Conduit MTCDT-H5-210L gateway, as
PER and lower minimum RSSI. A lower SF therefore shown in Figure 4, manufactured by MultiTech [11]. Since
should result in a higher PER and a higher minimum the gateway is designed to listen for incoming messages on
RSSI. the 868MHz band, all end-devices support the three default
Coding Rate (CR): A higher CR transmits more redundant channels (0, 1 and 2, i.e. 868.1, 868.3, 868.5 MHz) that must
data bits, consequently producing a lower overall PER. be implemented in every EU868MHz end-device to comply
with the LoRaWAN specifications. We set up a local Node.js up to SF12 perform worse than expected. It is not possible
web application as the server-side of the LoRa network server. to prove this by only looking at the RSSI. Therefore we
It runs a Mariadb database to collect and store the transmitted also measure the packet error rate and analyze it in the
packets from the LoRa end-devices. The database provides next section. A possible alternative reason for the unexpected
various information about the received packets such as RSSI high minimum RSSI values for higher SFs is the way the
values and sequence number. RSSI is calculated. For higher SFs, a packet has a longer
T ime − on − Air (Equation 1), giving a longer time over
A. Deplyment which the signal strength can be integrated. This depends on
Figure 4 shows the LoRa gateway deployment at one of the method implemented inside the proprietary (”black box”)
the buildings of the University of Twente campus, where LoRa radio module.
our experiments took place. The indoor experiments were For locations closer to the gateway, the link fluctuation is
performed with end-devices and a gateway placed on one floor high for high SFs, with maximum standard deviation of σ =
of our office building as shown in Figure 4, the floor is roughly 4.18dBm. The link is observed to be more stable at the farthest
103m by 20m. To test LoRa for different communication location at high SFs (SF11, SF12), with maximum change of
scenarios, the transmitting end-devices were placed at four σ = 3.6dBm only.
locations (denoted by L1 - L4 in Figure 4) on the floor
while the receiving gateway was placed at the corner of the RSSI LoRa
building. Experiments were performed during normal office -140 L1 L2 L3 L4 Expected RSSI

working hours to include the influence of people movements -130

and environmental dynamics on LoRa network performance. -120

-110
As mentioned in Section II, due to the channel access

RSSI
-100
restrictions imposed on EU863-870MHz LoRa operating fre- -90
quency, data packets will defer their next data packet transmis- -80

sion for a time period of at-least TOf f SubBand . Changing the -70

spreading factor (SF) results in a change in TOf f SubBand and -60

time on air T oA, which will subsequently result in a change 7 8 9

SFLoRa
10 11 12

in the link budget, i.e. battery lifetime versus range trade-off.

For example from Equation 1, it will take at least 3.8 hours Fig. 5: Minimum RSSI (in dBm) for SF=7 to 12 at locations (L1, L2,
L3 and L4).
for SF = 12 and 21 minutes for SF = 7 to finish sending a
50 Byte long data frame. This is in accordance with the 1%
duty-cycle regulation, calculated with respect to the payload
size for every spreading factor settings. In addition all the C. Impact of Packet Error Rate (PER)
measurements are performed using unconfirmed mode (unac- Figure 6 illustrates the packet error rate (PER) versus LoRa
knowledged) data frame types to prevent the acknowledgement end-node locations (L1-L4) for spreading factors SF7 to SF12.
from clogging the channel for upstream packets. The transmitted LoRa packet is set to unconfirmed with no
B. Impact of Received Signal Strength Indication (RSSI) retransmission. End-devices transmit 50 data packets each with
50 bytes of payload for every test location. One can see that
To test the overall capability of LoRa radio receiver to in general the packet loss has an increasing trend when the
demodulate data from a received signal, we analyze the re- transmitter end-device is farther from the gateway location.
ceived packets in terms of their RSSI values. These values are From the theoretical analysis presented in Section III, we
compared to the Semtech SX1276 LoRa receiver theoretical expect that a higher SF should result in a lower PER. This
sensitivity which is specified for various BWs and SFs [4] is, however, not observed at any one of the four locations.
settings (Table I). In our experiment the end-device transmits
50 data packets, each with 50 bytes of payload, at every test
PER
location, and for every SF. All packets are unconfirmed with 25
L1 L2 L3 L4

no retransmissions, as presented in Table II. 20

Figure 5 shows the minimum RSSI measured at locations
L1, L2, L3, and L4. As the number of obstructions (non- 15
PER[%]

line-of-sight) and distance to the gateway increases, the RSSI

10
values decreases. This trend is observed for all SF values.
Theoretically as explained in III, one would expect the min- 5

imum RSSI for received packets decrease as the SF value

increase. In our results this is only true for SF7, SF10, SF11, 0
7 8 9 10 11 12
SFLoRa
and SF12. In case of SF8 and SF9, SF9 performs better
than expected, and SF8 performs even much better. In other Fig. 6: Packet Error Rate (PER) for SF=7 to 12 at Locations (L1,
words, SF7 and SF8 are performing as expected, while SF9 L2, L3 and L4).
 LoRa settings
Parameter Value Details
Center Freq. EU 868MHz Class A (Either of the Default three chan-
nels in case of 868MHz band)
Band Width (BW) 125Hz Default for LoRaWAN configuration
Spreading Factor (SF) SF7-SF12 SF7-SF12 are to be set (as per [3])
Tx-Power 14dBm The maximum default tx power for 868
MHz
Tx Payload Size (byte) 50 (used with all Spreading Factors (SF)
Data Frame Mode Unconfirmed Unacknowledged data frame
Radio Antenna Properties
End-Mote Frequency range: 868MHz, Gain: 2.1dBi RN2483 MICROCHIP LoRa Module,
VSWR: < 1.5
Gateway Female SMA, 2dBi Detachable Omni-directional antennas
MUTITECH-Gateway

TABLE II: LoRa Radio parameter settings used in the performance test.

Since the LoRa gateway is placed in an indoor environment, D. Impact of Coding Rate (CR) and Transmission Power
it is more susceptible to indoor signal shadowing and multi- The idea behind this set of experiments is to evaluate the
path fading. Using high spreading factors, such as SF=12, impact of the Coding Rate and the transmission power on
increases the T ime − on − Air as expressed by Equation 1, LoRa performance by fixing the Data Rate. To do so, instead
resulting in a 50 byte payload LoRaWAN packet to have a of transmitting at different spreading factors, all packets are
T oA = 2794s. In an indoor environment with people moving sent at a fixed SF.
around signal paths can change very fast. In the first experiment, we transmit 15 packets on each
coding rate, transmission power and three data rates, i.e., DR
The reason for higher SFs to perform worse than lower 5, DR 3 and DR 1 at each transmission round. After every
SFs is partly related to the longer T oA of higher SFs. When 15 packets, the coding rate is increased until the maximum
a packet is received by the gateway, the preamble of the is reached. Then the coding rate is set back to the minimum
packet is used by the receiver to lock onto the transmitter. and the transmission power is decreased by one. The end-
This ”locking on to” means both synchronizing in time and devices were programmed to start at DR5, coding rate 4/5,
adjusting the gain of the receiver’s preamplifier. Preamplifiers and transmission power of 14 dBm. After each 15 packets,
can have a relatively small dynamic range compared to the the coding rate was increased until the maximum of 4/8 was
variations in the signal path at an indoor location. Our theory reached. Then the transmission power was lowered one step
is that a packet with a long T oA will be received at a specific and the same process was repeated. The packet loss from the
preamplifier setting, but shortly after this the signal strength end-device at location L3 transmitting on DR5 (i.e. SF=7) is
will change rapidly because of multiple signal paths as well as shown in Figure 7. It can be seen that generally transmitting
quick fading due to movement of people in the building. As packets with a higher transmission power decreases the packet
soon as the received signal strength falls outside the dynamic loss.
range of the receiver’s preamplifier, the rest of, or part of
the packet is not received. For longer packets, the fraction 90
of the packet that is not received becomes more significant, 14dBm
11dBm
8dBm
80
causing the forward error correction to fail. For lower SFs, 5dBm
2dBm
Combined

the T oA is much shorter. This lowers down the chance of 70

falling outside the dynamic range of the receiver. This will 60

Packet Loss (%)

lower down the fraction of the packet that is not received and 50

will consequently increase the probability of the forward error 40

correction to succeed. 30

20
If our assumption about the preamplifier in the ”black box”
10
proprietary LoRa receiver is correct, then the results illustrated
in Figure 6 can be described as follows: closer to the gateway, 0
1 2 3 4
Coding Rate (CR)
packets with shorter T oAs are more likely to be received. For
locations further from the gateway, lower spreading factors are Fig. 7: The packet loss from the end-device at location L3, transmit-
more likely to fail. Location L1 is close to the gateway and ting on DR=5; for varying transmission power and coding rate.
the effect of packet ToA can be clearly seen. The further we
go from the gateway, the effect of SF is more visible as PER We repeat the same experiment with DR3 and DR1. Fig-
is rapidly increased for lower SFs and slightly increased for ures 8 and 9 show the results. One can see that When the data
higher SFs. rate is decreased (resulting in an increase of the spreading
factor), the packet loss decreases as well. This behavior is performance From our investigation of LoRa radio RSSI
expected from the spread spectrum concept as signals with values, we observed that due to the broadband chirp pulses
higher spreading factors are more likely to reach a gateway. and higher sensitivity of the LoRa modulation, LoRa offers
There are, however, also some notable differences in packet immunity against multi-path and signal fading especially at
loss between the end-device at different CRs. high spreading factor. At closer distances to the gateway, the
RSS is high for low spreading factor scheme. In addition
14
to that, when the spreading factor is increased the packet
14dBm
11dBm
8dBm
loss decreases at the expense of decreased effective bit rate,
12 5dBm
2dBm
Combined
which is not suitable for high throughput IoT applications.
10
And at farthest locations from the gateway interferences are
significantly high, therefore, end-devices should communicate
Packet Loss (%)

8 at high spreading factors.

6
ACKNOWLEDGMENT
4 This research was supported by the Smart Parks Project,
funded by the Netherlands Organization for Scientific Re-
2
search (NWO).
0
1 2
Coding Rate (CR)
3 4
R EFERENCES
[1] A. Mohammed Al-Fuqaha, A. Guizani and A. Moussa.
Fig. 8: The packet loss from the end-device at location L3, transmit-
ting on DR=3; for varying transmission power and coding rate. Internet of things: A survey on enabling technologies,
protocols, and applications. IEEE Communications Sur-
veys Tutorials, 17(4):2347–2376, 2015.
[2] M. Centenaro, L. Vangelista, A. Zanella, and M. Zorzi.
14
14dBm Long-range communications in unlicensed bands: The
11dBm

12
8dBm
5dBm rising stars in the iot and smart city scenarios. IEEE
2dBm
Combined
Wireless Communications, 23(5):60–67, 2016.
10
[3] Lora alliance–wide area networks for iot. [Online].
Packet Loss (%)

8
Available: https://www. lora-alliance. org/[Accessed: 06
February 2016], 2016.
6
[4] Sx1276, August 2016. [online] http://www.semtech.com/
4
wireless-rf/rf-transceivers/sx1276/.
[5] T. Wendt, F. Volk, and E. Mackensen. A bench-
2
mark survey of long range (loratm) spread-spectrum-
0
communication at 2.45 ghz for safety applications. In
1 2 3 4
Coding Rate (CR) WAMICON, 2015 IEEE 16th Annual, pages 1–4. IEEE,
2015.
Fig. 9: The packet loss from the end-device at location L3, transmit-
[6] K. Mikhaylov, J. Petäjäjärvi, and T. Haenninen. Analysis
ting on DR=1; for varying transmission power and coding rate.
of capacity and scalability of the lpwan technology. In
22th EWC; Proceedings of, pages 1–6, 2016.
Some end-devices show a slightly higher packet loss at
[7] A. Augustin, J. Yi, T. Clausen, and W. Townsley. A study
higher coding rates or transmission powers, which contradicts
of lora: Long range & low power networks for the iot.
the expected lower packet loss at higher coding rate and
MDPI Sensors, 16(9):1466, 2016.
transmission powers. The numerical differences between the
[8] J. Petajajarvi, K. Mikhaylov, A. Roivainen, T. Hanninen,
lost packets on different transmission settings are, however,
and M. Pettissalo. On the coverage of lpwans: range
small and might have been caused by environmental changes
evaluation and channel attenuation model for lora tech-
of office environment due to factors such as people moving
nology. In ITS, 2015 14th Conf. on, pages 55–59. IEEE,
around and doors being opened and closed.
2015.
V. C ONCLUSION [9] B. Reynders, W. Meert, and S. Pollin. Range and coexis-
In this paper we studied LoRa wireless technology, a tence analysis of long range unlicensed communication.
new LPWAN protocol for IoT applications, and conducted In ICT, 2016 23rd International Conference on, pages
its network performance analysis. A general overview of 1–6. IEEE, 2016.
LoRa modulation and network architecture is introduced. The [10] TR ETSI. Tr 102-313 v1. 1.1,”. ERM, pp8.
associated LoRa physical layer parameters such as spreading [11] Multitech, Aug. 2016. [online] http://www.multitech.
factor, bit rate and coding rate are discussed. We built and com/.
prototyped LoRa radio enabled network, to test the network

View publication stats