ACCEPTED FROM OPEN CALL

Exploring Multi-Hop LoRa for Green Smart Cities

Muhammad Shehryar Aslam, Alishba Khan, Abeera Atif, Syed Ali Hassan, Aamir Mahmood, Hassaan Khaliq Qureshi, and Mikael Gidlund

Abstract To realize such systems, it is essential to have

a wireless network with wide-area coverage and
With the growing popularity of IoT-based low-power consumption in order to connect bat-
smart city applications, various long-range and tery-operated devices deployed across a city.
low-power wireless connectivity solutions are Although fifth generation (5G) systems make use
under rigorous research. LoRa is one such of the licensed bands for most of their applica-
solution that works in the sub-GHz unlicensed tions, many new applications such as millimeter
spectrum and promises to provide long-range wave (mmWave) communications and IoT-based
communication with minimal energy consump- technologies make use of the unlicensed bands.
tion. However, conventional LoRa networks are A major reason for this adoption is the spec-
single-hop, with the end devices connected to a trum scarcity in the licensed bands that renders
central gateway through a direct link, which may the new applications to work in other portions
be subject to large path loss and hence render of the spectrum. In this perspective, low power
low connectivity and coverage. This article moti- wide area networks (LPWANs) provide a feasible
vates the use of multi-hop LoRa topologies to solution not only in terms of their operation in
enable energy-efficient connectivity in smart city the unlicensed bands but also their performance
applications. We present a case study that exper- envisions wide-area coverage range and energy
imentally evaluates and compares single-hop efficiency. LPWAN technology is designed for
and multi-hop LoRa topologies in terms of range networks with massive battery-operated devic-
extension and energy efficiency by evaluating es which are typically required for a smart city
packet reception ratio (PRR) for various source and machine-to-machine (M2M)-based applica-
to destination distances, spreading factors (SFs), tions. At present, the prominent competing unli-
and transmission powers. The results highlight censed LPWAN technologies are SigFox, LoRa,
that a multi-hop LoRa network configuration can and Weightless, whereas the licensed technolo-
save significant energy and enhance coverage. gies include Long Term Evolution for M2M (LTE-
For instance, it is shown that to achieve a 90 per- M), and narrowband IoT (NB-IoT) [2]. However,
cent PRR, a two-hop network provides 50 per- Long Range (LoRa) stands out among the others
cent energy savings as compared to a single-hop because of its spread spectrum technology [3],
network while increasing 35 percent coverage whereas it operates in the frequency bands of
at a particular SF. In the end, we discuss open 433 MHz, 868 MHz and 2.4 GHz unlicensed
challenges in multi-hop LoRa deployment and radio spectrum. The technology, as the name
optimization. indicates, promises long-range communication
based on excellent receiver sensitivity and chirp
Introduction spread spectrum (CSS) modulation. The range
In the upcoming era of Internet-of-Things (IoT), goes up to as high as 13 km in a rural setting.
smart systems for applications such as predictive However, with the increased range, the data rates
maintenance and traffic flows are expected to are compromised, with a minimum of 0.3 kb/s
be widely deployed across the globe. The smart and a maximum of 50 kb/s. Consequently, the
device market is growing considerably, while 20 power consumption for each device is minimal,
billion IoT devices are estimated to be deployed which makes the technology ideal for a green
by 2020 [1]. With such a massive number of smart city implementation where a slower data
devices, the concept of smart cities is now real- rate is acceptable as long as the battery life can
izable. A smart city is an integrated urban system be increased.
involving the use of infrastructures such as smart In a smart city, a large number of battery oper-
surveillance and control systems. Hundreds of ated devices need to uplink information with
thousands of end devices are utilized to gather minimal energy consumption. It has been fore-
real-time data from city systems. This data is then casted that approximately 60 thousand devices
used to analyze trends and make decisions to per square kilometers will be installed in a smart
not only maximize efficiency but also automate city [4]. Such a large number of devices requires
most major city operations. Areas of application a large amount of power. Therefore, to envi-
include smart homes, smart traffic control, smart sion a green smart city, a robust system model
metering, agriculture, health care, manufactur- is required which may use the multi-hop topolo-
ing, and urban infrastructure management, to gies to conserve energy. Multi-hop networks are
name a few. considered a maturing area of research and an

Muhammad Shehryar Aslam, Alishba Khan, Abeera Atif, Syed Ali Hassan (corresponding author) and Hassaan Khaliq Qureshi are with the Digital Object Identifier:
National University of Sciences and Technology (NUST); Aamir Mahmood and Mikael Gidlund are with Mid Sweden University. 10.1109/MNET.001.1900269

IEEE Network • March/April 2020 0890-8044/20/$25.00 © 2020 IEEE 225

Authorized licensed use limited to: AMITY University. Downloaded on June 30,2020 at 18:20:36 UTC from IEEE Xplore. Restrictions apply.
 Random Phase Multiple Access (RPMA):
A LoRa network is comprised of two main solutions: LoRa, a proprietary PHY layer modulation scheme RPMA envisions a wide area network that is
developed by Semtech, and LoRaWAN, developed by the LoRa Alliance, defining the protocol stack, capable of providing services deep inside build-
ings and underground. Although the data rates
network architecture, device classes, and regional regulations. of RPMA are low (31 kb/s download and 15.6
kb/s upload), the inherent security features of
the RPMA make it vulnerable to threats. The key
important mechanism for efficient energy usage application areas for RPMA include smart build-
and range extension [5]. The communication ings, agricultural monitoring, personal tracking,
devices in wireless sensor networks and other IoT smart metering, asset tracking, oil and pipeline
applications have limited battery life. Under the monitoring, and so on.
right conditions, multi-hop networks, in addition Weightless: Weightless is a low-cost tech-
to enhancing throughput due to shorter hops, nology specifically designed for M2M commu-
can also extend battery life due to lower required nications. A salient feature of Weightless is the
transmission power. It has been shown that to handling of a large number of devices efficiently.
achieve the same quality-of-service (QoS), lower There are various sub-categories of Weightless
transmission power at each device is required for such as Weightless-P, Weightless-N, and Weight-
a multi-hop network as compared to single-hop less-W that differ in specifications such as range of
networks where the end devices communicate operation, data rate, and battery life.
directly with the base station [6]. The LoRa Technology: LoRa enables long-range
The main contributions of this article are as operation with its special modulation technique,
follows: which is discussed in detail in the later sections
• We present an empirical proof of range of this article. From a research perspective, many
extension and increased energy efficiency of aspects of LoRaWAN technology such as energy
the network using multi-hop LoRa configu- efficiency, range and coverage have been investi-
rations. The single-hop serves as a reference gated recently in a smart city scenario [8].
topology, against which the operations of
the multi-hop have been compared. The Conventional LoRa Network
• We test various topologies such as two-hop, A LoRa network is comprised of two main solu-
three-hop, and star-of-stars to provide a com- tions: LoRa, a proprietary PHY layer modulation
prehensive comparison under various system scheme developed by Semtech (https://www.
parameters such as spreading factors (SFs), semtech.com/), and LoRaWAN, developed by
transmission powers and distances. the LoRa Alliance (https://www.lora-alliance.
• To further improve the energy and range effi- org/about-lorawan), defining the protocol
ciency of the network, optimal relay node stack, network architecture, device classes, and
placement in two or three hops is consid- regional regulations. In this section, we briefly
ered which elevates the system performance describe these solutions before discussing our
in terms of QoS. multi-hop extension of the conventional LoRa
• The work is finally concluded with futuristic network.
recommendations and open research chal-
lenges. LoRa PHY Layer
With an extensive analysis provided in this arti- LoRa PHY uses a proprietary derivative of
cle, it can be concluded that a multi-hop LoRa the chirp spread spectrum (CSS) modulation
configuration can be used as an enabler wire- scheme. In CSS techniques, the data symbols are
less communication technique for IoT devices modulated by chirp pulses, which are frequen-
deployed in large numbers for smart city appli- cy varying sinusoidal pulses of fixed bandwidth
cations. B and time interval. One way to overcome the
scarcity in the unlicensed bands is to vary the
Smart City Technologies chirp duration so that quasi-orthogonal signals
can be created and serve as virtual channels.
Using Unlicensed Bands The chirp duration, however, leads to a trade-
Unlicensed bands provide an extra space for tech- off between the throughput and the robustness
nologies to operate which otherwise would be dif- against noise and interference.
ficult. Although many IoT-based applications would For a fixed chirp duration, data symbols are
still be operable in licensed bands with 2G/3G/4G coded by unique instantaneous frequency tra-
compatibility, examples include LTE-M, NB-IoT, jectory, obtained cyclically shifting a reference
EC-GSM and so on, but because of rising costs chirp. Chirp wrapping is discretized, that is, only
of licensed bands and bandwidth limitations, the 2SF possible edges in the instantaneous frequency
unlicensed technologies are gaining momentum exist, each one representing SF bits where SF is
for smart city applications [7]. Herein, we briefly referred to as the spreading factor. The network
list the unlicensed IoT technologies, which are controller can adapt the data rate by changing
potential candidates for future smart cities. the bandwidth B ∈ {125, 250} kHz and SF ∈ {7,
SigFox: SigFox operates in the sub-GHz unli- …, 12}, which together relate to chirp duration
censed spectrum with a physical (PHY) layer as T c = 2SF/B. Note that the chirp rate remains
consisting of ultra-narrow band technology which the same, and is equal to B, while the chirp dura-
enables a long-range operation at the cost of data tion (consequently time-on-air) increases drasti-
rates. The data rates for a SigFox transceiver do cally with the SF. On the positive side, a higher
not exceed 100 bps. At the network layer, a sim- SF yields higher processing gain and thus reduces
ple star topology enables a robust network oper- the target signal-to-noise ratio for correct recep-
ation. tion at the receiver [9].

226 IEEE Network • March/April 2020

Authorized licensed use limited to: AMITY University. Downloaded on June 30,2020 at 18:20:36 UTC from IEEE Xplore. Restrictions apply.
FIGURE 1. LoRaWAN architecture and its extension for multi-hop communication.

LoRaWAN different topologies forming a multi-hop LoRa net-

In terms of network coverage and architecture, work, as illustrated in Fig. 1 (right).
the LoRaWAN connectivity solution is similar The first topology introduces a relay node
to cellular systems. However, in LoRaWAN, the between an end device and the gateway. The
primary focus is on energy efficient communi- relay node employs decode-and-forward scheme
cation for smart city IoT applications based on based on LoRa modulation. The second topology
battery-operated devices. A typical LoRa net- extends the LoRaWAN’s star-of-star architecture
work consists of a star-of-stars topology, where by allowing a LoRa gateway to connect to multi-
the end devices can communicate to a single ple LoRa gateways over LoRa PHY. As wide-area
or several gateways using LoRa PHY. The gate- networks in urban settings employ massive devic-
ways, or base stations as some may refer, are es, it becomes difficult for a central gateway to
connected to a common network server via gather/process information from all the devices
the standard IP protocol. The network server is simultaneously.
connected to an application server, as shown In a star-of-stars topology, the devices are cat-
in Fig. 1 (left). egorized into several clusters. Clustering is consid-
Role of Each Network Entity: The end devic- ered a powerful tool to streamline the operations
es can transmit(receive) to(from) the gateway. of the network to maximize energy efficiency
However, the emphasis is on the event-triggered and consequently prolonging network lifetime
uplink transmissions. The gateway functions trans- [10]. Each cluster contains multiple end devices
parently as a relay between the end devices and and a gateway of its own. Each end device com-
the network server. The network server manag- municates with the gateway of the cluster. Thus,
es the overall network. It allocates resources, for each gateway has to deal with a smaller number
example, the spreading factor or the bandwidth of devices as compared to the scenario when a
for an efficient data rate and also authenticates single central gateway receives data from every
the end devices. The application server handles device. Cluster gateways then transmit the data
data encryption and decryption and the admission to a central gateway where it can be processed
of the end devices to the network. and relayed to the network server. This formation
Medium Access: LoRa operates on unlicensed realizes a two-hop LoRa network, which is tested
sub-GHz frequency bands. The ISM band is sub- in this article for range extension and power effi-
ject to regulation on either the listen before talk, ciency for different SFs and the permitted range
medium access duty cycling or effective radiated of transmission powers.
power (ERP). A simple duty-cycled ALOHA pro-
tocol, regulated by the network server, serves as Performance Evaluation
the most common approach for accessing the This section outlines the experimental setup for
wireless medium. the proposed case study and presents the results
obtained.
Multi-Hop LoRa Network: A Case Study
As the devices in IoT wireless networks are mostly Hardware Prototype
battery-operated, they have limited energy. To The experimental setup for the tests is shown in
increase their lifetime, energy utilization must be Fig. 2. The tests were carried out using LG01-S,
managed efficiently. In addition, wide-area cov- which is an open source single channel LoRa
erage is an intrinsic demand of smart city appli- gateway. It enables users to bridge a LoRa wire-
cations. To address these issues, we propose two less network to an IP network via WiFi or Eth-

IEEE Network •March/April 2020 227

Authorized licensed use limited to: AMITY University. Downloaded on June 30,2020 at 18:20:36 UTC from IEEE Xplore. Restrictions apply.
 ernet. Dragino LoRa shields were used as end used as end devices have a maximum transmit
devices. The communication between the gate- power of 20 dBm. We evaluate the packet recep-
way and the end device took place in the 433 tion ratio (PRR) based on the data transmission
MHz frequency band. As this band is not used period of two minutes. This implies that nearly 10
for any local transmission, negligible interference thousand packets are transmitted in each test.
was present during the experiments. LoRa shields
Single-Hop vs. Two-Hop LoRa Network
To investigate the range extension with a two-
hop network, where a relay node assisted the end
device to send its data to the gateway, the fol-
lowing setup was established. The gateway was
placed in an indoor environment; a lab, situat-
ed at the premises of SEECS, NUST, Islamabad,
whereas, a relay node was placed outdoor at a
point where all of the messages were conveyed
from the relay to the gateway. Similarly, an end
device was placed further outdoor at different
locations, such that the distance between the end
device and the gateway varied with each location.
A comparison between the PRR of a sin-
gle-hop and a two-hop network for two SFs is
shown in Fig. 3. It can clearly be seen that the
maximum range is significantly extended by intro-
ducing a relay node in the network. For example,
if the QoS, which is the PRR in this case, is set at
80 percent, a single-hop network with SF7 can
provide communication up to 180 m, whereas
with the same settings, the range of a two-hop
network goes up to 260 m, that is, an increase of
80 m. One would expect the range to be almost
doubled, but the communication link between the
end device and the relay node is not as strong as
the link between the end device and the gateway.
Therefore, the communication range between
end device and relay is limited.
Note that for different SFs, the range of com-
munication between the end device and the gate-
way is different [9]. It is also evident from Fig. 3
that for larger spreading factors, the range exten-
sion, because of the introduction of the relay
node, is greater as compared to the single-hop.
For example, for 90 percent QoS, extension in
range for SF7 is about 60 m which goes up to
100 m for SF9.
FIGURE 2. Setup for star-of-stars topology experiments. As compared to a single-hop network, a two-hop
strategy results in the reduction of transmit power
100 to obtain a specific PRR. Even though the configu-
SF7 Single-hop ration demands two devices to transmit individually,
SF7 Multi-hop as compared to one device in a single-hop network,
90 SF9 Single-hop the total power by both devices is collectively less
SF9 Multi-hop
than that of a single-hop device to cover the same
Packet reception ratio (%)

range of operation with the same QoS.

80
Relay Node Placement
The position of the relay node in the network
70 has a direct impact on both range extension and
power reduction. The network is energy efficient
only if the relay is placed optimally. Figure 4
60 shows the trends of required transmit power at
various positions of the relay node to obtain a
certain PRR. The abscissa refers to the distance of
50 the relay from the gateway, whereas the total dis-
tance between the end device and the gateway
is kept constant at 150 m. It can be seen that the
40 least amount of power is required when the relay
50 100 150 200 250 300 350 400 450 node is placed farther from the gateway, implying
Distance (m) that the distance between the end device and
relay node is small. For example, for a QoS of 90
FIGURE 3. Packet reception ratio (PRR) for single-hop and two-hop networks at two percent, the system requires 22 mW if the relay
different SFs. The transmit power of end device in both topologies is 20 mW. node is placed at 50 meters from the gateway.

228 IEEE Network • March/April 2020

Authorized licensed use limited to: AMITY University. Downloaded on June 30,2020 at 18:20:36 UTC from IEEE Xplore. Restrictions apply.
However, if the relay is placed at 100 m from
100
the gateway, only 16 mW of transmit power is 0 2
required to maintain the same QoS. This trend Required transmit power (mW)
changes if the relay node is moved further away 18

26

24

20
95
from this point. At 120 m, the required transmit
power is about 18 mW.

Packet reception ratio (%)

Star-of-Stars LoRa Network 90

22
To evaluate a star-of-star network, a central gate-
way was placed in the campus lobby. Two clus-

16
85
ters were formed, as shown in Fig. 2, where one

18
20
cluster had four end devices including the gate- 16

24
way while the other had three end devices. Each
80
end device transmitted data to the cluster gate-

14
way, and the cluster gateway concatenated data
from all the devices in an array and transmitted it

22
75
to the central gateway.

14
The transmit power required to maintain a PRR 18

12
16
of 90 percent for different SFs under all network

20
70
topologies is compared in Fig. 5. It is evident that 30 40 50 60 70 80 90 100 110 120
multi-hop topologies require less transmit power Distance of relay node from the gateway (m)
than a single-hop configuration to cover the
same distance. For instance, a two-hop network FIGURE 4. Required transmit power as a function of PRR and location of relay
required 18 mW transmit power to achieve a 90 node for SF = 7.
percent PRR at SF7, while for the same settings
a single-hop required 24 mW power, thus giv- 25
ing 25 percent power saving. A three-hop setup
makes the network even more energy efficient, Single-hop
giving 37.5 percent energy saving compared to Two-hop
the single-hop network. Similarly, the star-of-stars 20
Three-hop
is much more energy efficient than the single or Star-of-stars
multi-hop network as it requires only 12 mW of
Transmit power (mW)

transmit power. The power shown in the graph

for the star-of-stars topology is the average power 15
required by one device to transmit data to the
central gateway via the cluster gateway. With
the increase of SF, how required transmit power
declines even further for multi-hop and star-of-star 10
topologies can also be observed from Fig. 5.
The required transmit power with respect to
the PRR target is compared for single-hop, two-
hop, and star-of-stars networks in Fig. 6. To obtain 5
these results, the distance between the end device
and gateway is kept constant at 150 m in all three
cases. It is observed that both the two-hop and star-
0
of-stars networks require less transmit power than SF7 SF9 SF12
a single-hop network. In fact, the average power
used by one device in a star-of-stars topology is FIGURE 5. Comparison of transmit power required for single hop, two-hop, three-
almost half of that required by an end device in a hop networks and star-of-stars networks for different spreading factors.
single-hop configuration. In addition, as the PRR
target is increased (i.e., higher required QoS), the
gain in power saving for a multi-hop and star-of- to capitalize on such gains, there are plenty of
stars networka is also increased. challenges and research opportunities to the LoRa
To quantify the energy efficiency of the clus- system in general and the multi-hop LoRa system
tering technique in comparison with a single-hop in particular, which we highlight below.
topology, we define the metric, fraction of energy
saved (FES), as Intelligent LoRa Networks
⎛ Power required in star-of-stars ⎞ LoRa/LoRaWAN provides many tunable trans-
FES= ⎜ 1− . (1) mission parameters such as transmission power,
⎝ Power required in single-hop ⎟⎠
coding rate, spreading factor, bandwidth and so
We observed that the FES for a star-of-stars on, resulting in hundreds of possible combinations.
topology is 28 percent, 41 percent, 48 percent, Increasing the spreading factor nearly halves the
51 percent and 54 percent for 1, 2, 3, 4, and 5 data rate and doubles the energy consumption
devices in the cluster, respectively. and airtime, thereby improving the link reliability,
as we have depicted in some of our results. Similar-
Challenges and Future Directions ly, the increased bandwidth doubles the data rate
Our results have shown that multi-hop and star- and halves the energy consumption and airtime,
of-stars topologies are potential candidates to reducing link reliability (due to additional unwant-
extend the coverage and improve the energy effi- ed noise). These configuration parameters must
cient of LoRa networks simultaneously. However, be optimally tuned while taking into account the

IEEE Network •March/April 2020 229

Authorized licensed use limited to: AMITY University. Downloaded on June 30,2020 at 18:20:36 UTC from IEEE Xplore. Restrictions apply.
 of duty cycling, that is, the IoT devices cannot
26
occupy the band for an infinite amount of time
Single-hop [13]. This critical requirement becomes prom-
24 Two-hop inent in case of a multi-hop scenario since the
Star-of-stars entire multi-hop link should be in the ON state for
22 data delivery to the gateway. However, making
an entire multi-hop link to work in the ON state
Transmit power (mW)

20 would affect the transmission duty cycle (TDC)

of other links, which causes an imbalance in legal
18 regulations of the ISM band, which permits a spe-
cific amount of transmission time per hour to be
16 occupied by various devices.

14 Synchronization and Queuing in

Multi -Hop LoRa Networks
12 Cooperative multi-hop networks, although promis-
ing, require tight synchronization among different
10 hops for their efficient operation [14]. Similarly,
if various hops are subject to different spreading
8 factors, which affects the data rates on each link,
60 65 70 75 80 85 90 95 100 the problem of queuing appears as the intermedi-
Packet reception ratio (%) ate nodes have to store the information according
to link adaptation. This may create unavoidable
FIGURE 6. Comparison of transmit power required for single-hop, multi-hop, and delays in the network that require the use of
star-of-stars networks for various SFs. sophisticated techniques to alleviate the problem.

Conclusion
local electromagnetic environment, constraints, LoRaWAN has been designed to meet the require-
and objectives, that is, the performance metrics. ments of modern age massive IoT and can be con-
Recent research on LoRa/LoRaWAN has main- sidered a potential candidate for realizing a smart
ly focused on LoRa performance evaluation in city network. This article briefly encompassed the
terms of coverage, capacity, scalability and lifetime. various unlicensed IoT techniques and focused on
Recent work has also proposed adaptive approach- the architecture and working of LoRa systems in
es to allocate optimal transmission parameters [11]. particular for their operation. An experimental case
However, most of these methods are based on study on multi-hop LoRa systems has been pro-
state-of-the-art mathematical/statistical models and vided that ought to make the system more ener-
suffer from limited modeling assumptions, limited gy efficient and long range. Two topologies were
learning, an inability to deal with non-linear com- presented; the first one was based on introducing
plex behaviors, poor scalability, and no time series/ a forwarding relay node, whereas the second one
temporal data exploitation. Machine learning and was oriented toward clustering the devices and
deep learning could, therefore, be a potential area forming a star-of-stars topology. The results indi-
of research in forming intelligent LoRa networks cated that the two-hop networks can significantly
complying with intelligent radio resource manage- extend the range as compared to a single-hop net-
ment and diverse communication requirements in work. In terms of energy efficiency, both the two-
the massive IoT perspective. hop topologies required less power to achieve the
same QoS as compared to a single-hop network.
Multi-Hop Deployment Optimization As the star-of-stars network saved more energy for
Although, the initial experimental results stated a higher number of end devices in a cluster, it can
in this article motivate the use of multi-hop trans- be concluded that LoRa networks can greatly ben-
missions in LoRa networks, the scalability issues efit from the proposed scheme.
in multi-hop communications poses an import- Although there are many research challeng-
ant challenge to overcome [12]. The number of es with multi-hop LoRa networks, in summary, as
devices per cluster and the number of clusters, the smart city concept would require hundreds of
the number of hops to the gateway, and the thousands of end devices to be deployed across a
power consumption per node/cluster remain an city, multi-hop assisted star-of-stars topologies may
optimization issue that should be looked upon be considered a viable option for clustering many
from the perspective of the sensor density in a end devices together and making the network
city. For instance, the requirements of a well-con- more energy efficient.
nected urban city differ greatly from a rural area.
How and where multi-hop transmission would References
[1] “Gartner Internet of Things predictions,” available: https://
complement the performance remains an open www.gartner.com/newsroom/id/3598917; accessed:
challenge. Another challenge specific to multi- 27.04.2019.
hop deployments is denial of service attacks (e.g., [2] K. Mikhaylov, J. Petjjrvi, and T. Hnninen, “Analysis of Capac-
ity and Scalability of the LoRa Low Power Wide Area Net-
black-hole and grey hole attacks) that would work Technology,” Proc. 22nd European Wireless, Oulu,
require intrusion detection techniques. Finland, 2016, pp. 1–6.
[3] F. Adelantado et al., “Understanding the Limits of
Multi-hop Duty Cycling LoRaWAN,” IEEE Commun. Mag., vol. 55, no. 9, 2017, pp.
34–40.
Because of the unlicensed usage of spectrum in [4] O. Georgiou and U. Raza, “Low Power Wide Area Network
LoRaWANs, the occupied transmission time of Analysis: Can LoRa Scale?” IEEE Wireless Commun. Lett., vol.
a LoRa link is subject to a stringent requirement 6, no. 2, 2017, pp. 162–65.

230 IEEE Network • March/April 2020

Authorized licensed use limited to: AMITY University. Downloaded on June 30,2020 at 18:20:36 UTC from IEEE Xplore. Restrictions apply.
[5] M. Ahsen, S. A. Hassan, and D. N. K. Jayakody, “Propagation Abeera Atif received her B.S. degree in electrical engineering
Modeling in Large-Scale Cooperative Multi-Hop Ad Hoc from National University of Sciences and Technology (NUST),
Networks,” IEEE Access, vol. 4, 2017, pp. 8925–37. Pakistan. Her research interests include signal processing and
[6] M. S. Omar et al., “An Experimental Evaluation of a Coopera- wireless circuit designs.
tive Communication-Based Smart Metering Data Acquisition
System,” IEEE Trans. Ind. Informat., vol. 13, no. 1, 2017, pp. Syed Ali Hassan received the Ph.D. in electrical engineering
399–408. from the Georgia Institute of Technology (Georgia Tech), Atlan-
[7] U. Raza, P. Kulkarni, and M. Sooriyabandara, “Low Power ta, USA, in 2011, an M.S. in mathematics from Georgia Tech in
Wide Area Networks: An Overview,” IEEE Commun. Surveys 2011, and an M.S. in electrical engineering from the University
Tuts., vol. 19, no. 2, Second Qtr. 2017, pp. 855–73. of Stuttgart, Germany, in 2007. Currently, he is working as an
[8] D. Magrin, M. Centemnaro, and L. Vangelista, “Performance associate professor at NUST, Pakistan, where he is the direc-
Evaluation of LoRa Networks in a Smart City Scenario, ” tor of the Information Processing and Transmission Lab, which
Proc. IEEE ICC, Paris, France, May 2017, pp. 1–7. focuses on various aspects of theoretical communications. He
[9] A. Mahmood et al., ”Scalability Analysis of a LoRa Network has authored more than 150 conference and journal articles,
Under Imperfect Orthogonality,” IEEE Trans. Ind. Informat. mainly in the area of wireless communications.
vol. 15, no. 3, Mar. 2019, pp. 1425–36.
[10] P. Nayak and B. Vathasavai, “Energy Efficient Clustering Aamir Mahmood received the B.E. in electrical engineering from
Algorithm for Multi-Hop Wireless Sensor Network Using National University of Sciences and Technology, Islamabad,
Type-2 Fuzzy Logic,” IEEE Sensors J., vol. 17, no. 14, 2017, Pakistan, in 2002, and the M.Sc. and D.Sc. degrees in commu-
pp. 4492–99. nications engineering from Aalto University School of Electrical
[11] Z. Qin and J. A. McCann, ”Resource Efficiency in Low-Pow- Engineering, Aalto, Finland, in 2008 and 2014, respective-
er Wide-Area Networks for IoT Applications,” Proc. IEEE ly. Since 2016, he has been a post-doctoral researcher with
Globecom, Dec. 2017. the Department of Information Systems and Technology, Mid
[12] C. Liao et al., “Multi-Hop LoRa Networks Enabled by Sweden University, Sundsvall, Sweden. His research interests
Concurrent Transmission,” IEEE Access, vol. 5, 2017, pp. include energy-delay efficient radio resource allocation and
21430–46. analysis and optimization of low-power wide-area network
[13] R. M. Sandoval et al., “Optimal Policy Derivation for Trans- technologies.
mission Duty Cycle Constrained LPWAN,” IEEE Internet
Things J., vol. 5, no. 4, Aug. 2018, pp. 3114–25. Hassaan Khaliq Qureshi received the M.Sc. degree in electri-
[14] M. Hussain and S. A. Hassan, “Performance of Multi-Hop cal engineering with first class honors from the Blekinge Insti-
Cooperative Networks Subject to Timing Synchronization tute of Technology, Blekinge, Sweden, in 2006, and the Ph.D.
Errors,” IEEE Trans. Commun., vol. 63, no. 3, Mar. 2015, pp. degree in electrical engineering from City University, London,
655–66. U.K., in 2011. He is a recipient of the EU Erasmus Mundus
staff research mobility and Post Doctoral fellowship under
Additional Reading the STRONG TIES and INTACT program, respectively. He is
currently working as an associate professor with the School
[1] G. S. Aujla et al., ”Data Offloading in 5G-Enabled
of Electrical Engineering and Computer Science, National Uni-
Software-Defined Vehicular Networks: A Stackel-
versity of Sciences and Technology, Islamabad, Pakistan. His
berg-Game-based approach,” IEEE Commun. Mag., vol. 55,
main research interests include wireless networks and network
no. 8, Aug. 2017, pp. 100–08.
optimization techniques.
Biographies M ikael G idlund received the M.Sc. and Ph.D. degrees in
Muhammad Shehryar Aslam received his B.S. degree in elec- electrical engineering from Mid Sweden University, Sundsvall,
trical engineering from National University of Sciences and Sweden, in 2000 and 2005, respectively. Currently he is a full
Technology (NUST), Pakistan. His research interests include professor of computer engineering with Mid Sweden Univer-
cooperative communication, millimeter wave technology and sity. He has authored or co-authored more than 100 scientific
device-to-device communication. publications in refereed journals. He holds more than 20 pat-
ents (granted and pending applications) in the area of wireless
Alishba Khan received her B.S. degree in electrical engineering communications. His current research interests include wireless
from National University of Sciences and Technology (NUST), communication and networks, wireless sensor networks, access
Pakistan. Her research interest lies in routing for wireless networks. protocols, and security.

IEEE Network •March/April 2020 231

Authorized licensed use limited to: AMITY University. Downloaded on June 30,2020 at 18:20:36 UTC from IEEE Xplore. Restrictions apply.