Research On Using The AODV Protocol For A LoRa Mesh Network

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Research on Using the AODV Protocol

for a LoRa Mesh Network

Van Dai Pham1 , Duc Tran Le2(B) , Ruslan Kirichek1,3 ,


and Alexander Shestakov1
1
Bonch-Bruevich Saint-Petersburg State University of Telecommunications,
Saint Petersburg 193232, Russia
fam.vd@spbgut.ru, kirichek@sut.ru, alexandr.shestakov01@yandex.ru
2
The University of Danang - University of Science and Technology,
Danang, Vietnam
letranduc@dut.udn.vn
3
V.A. Trapeznikov Institute of Control Sciences of Russian Academy of Sciences,
Moscow 117997, Russia

Abstract. In this paper, we consider the LoRa technology to expand


sensor network coverage in smart sustainable cities. A model of a LoRa
mesh network is proposed using the AODV protocol in packet routing.
With a simulation model developed based on OMNET++, a series of
computer experiments was carried out with changing various param-
eters. In the experiments results, the end-to-end delay and packet loss
ratio were analyzed in the dependence on the number of nodes and packet
size in the network. The simulation results show that the latency is rela-
tively high in the LoRa mesh network, but it might be accepted for some
applications.

Keywords: IoT · LoRa · Mesh network · AODV · Delay · Packet loss

1 Introduction
Over the past decade, the Internet of Things (IoT) has received significant atten-
tion in the scientific and industrial fields. People can control, monitor, and do
a lot more from a remote distance. It is done by connecting various objects
reducing physical distance. The IoT movement creates the need for new wire-
less technologies, capable of supporting the large numbers of devices in the IoT
space. These systems require a technology that consumes less power and also
covers long distances. However, many technologies such as ZigBee, WiFi, Blue-
tooth popularly used at present consumes high power and is not suitable for
battery-operated systems. To fulfill the communication requirements of IoT, we
need new technology. Low-Power Wide Area Network (LPWAN) offers radio
coverage over a large area by way of base stations and adapting transmission
rates, transmission power, modulation, duty cycles, where end-devices incur a
very low energy consumption due to their being connected.
c Springer Nature Switzerland AG 2020
V. M. Vishnevskiy et al. (Eds.): DCCN 2020, LNCS 12563, pp. 149–160, 2020.
https://doi.org/10.1007/978-3-030-66471-8_13
150 V. D. Pham et al.

According to the ITU-T Y.4903/L.1603 recommendation for the development


of Smart Sustainable Cities, it is necessary to provide full Internet access covered
in the city. One of the well-known technologies included in the LPWAN network
group is LoRa (Long-Range) technology developed by Semtech. The long-range
and low-power nature of LoRa makes it a promising candidate for smart sensing
technology in civil infrastructures (such as health monitoring, smart metering,
environment monitoring, etc.), as well as in industrial applications.
Many technologies are proposed to use in IoT applications. Every technology
has its features, advantages, and disadvantages. However, no single technology
can serve all IoT applications because different applications will have different
requirements. Based on the requirement, we can only choose a technology that
is best suited for the specific application from the existing technologies.
WiFi is widely used in many IoT applications, the most common being the
link from the gateway to the router that connects to the Internet. However, it is
also used for primary wireless links requiring high speed and medium distance.
The WiFi is usually confined within a small area such as a building, home
office. It can span over a limited range that is 300 m in radius. However, Lora
technology provides long-range communication. A single LoRa gateway can span
an area of 100 km2 . In addition, the LoRa is based on the chirp spread spectrum
(CSS), which is highly resistant to multipath and fading.
ZigBee is an IEEE 802.15.4-based specification for a suite of high-level com-
munication protocols used to create personal area networks. It consists of small,
low-power digital radios. These are best suited for home automation, medical
device data collection, small scale projects that need to transfer data over small
distances. Its range is 10–100 m. Zigbee also consumes power more than LoRa
and is not suitable for outdoor applications with low power requirements. It also
does not have the long-range capability of LoRa technology.
Bluetooth is a wireless transmission technology that theoretically allows
short-distance connections between devices up to 100 m away but is only about
10 m in actual use. It is used mainly for speakers, health monitors, and other
short-range applications. That is the reason why in the context of long-range
transmission, LoRa proves superior to Bluetooth.
In summary, LoRa technology will revolutionize IoT by enabling data com-
munication over a long-range distance while using very little power. LoRa fills
the technology gap of Cellular and WiFi/BLE based networks that require either
high bandwidth or high power or have a limited range or inability to penetrate
deep indoor environments. In effect, LoRa Technology is flexible for rural or
indoor use cases in smart cities, smart homes and buildings, smart agriculture,
smart metering, and smart supply chain and logistics.

2 LoRa Overview

LoRa commonly refers to two distinct layers: LoRa physical layer, which is based
on the Chirp Spread Spectrum (CSS) [2] radio modulation technique, provides
long-range, low-power, and low-throughput communications. LoRaWAN (Long
Research on Using the AODV Protocol for a LoRa Mesh Network 151

Range Wide Area Network), a MAC layer protocol, provides a medium access
control mechanism for many end-devices to communicate with a gateway using
the LoRa modulation. The LoRa physical layer is based on the Chirp Spread
Spectrum (CSS) with integrated Forward Error Correction (FEC) that provides
low sensitivity for long communication ranges. CSS has many advantages:

– Great link budget, low-power transmission.


– Resistant to multipath and other interference.
– Orthogonality of spreading factors.
– Simplified electronic for receiver synchronization.
– Robust against Doppler shift (mobile application).

With this design, devices using different data rates do not interfere with each
other, and communication between devices and gateways can be operated over
multiple frequency channels, increasing the network capacity.
LoRa networks operate in unlicensed ISM frequency bands, which for Europe
is the frequency band 867–869 MHz with a central frequency of 868 MHz. For
this band, the LoRa specifications define 10 channels of 125/250 kHz uplink and
125 kHz downlink. On the other hand, for North America, the frequency band
is 902–928 MHz with a central frequency of 915 MHz, and it defines 64 channels
of 125/500 kHz uplink and 500 kHz downlink.
Besides the frequency range specifying possibility, several parameters are
available for the customization of the LoRa technology [17]:

• Bandwidth: The Bandwidth (BW) is the distance between the lowest and
highest frequency in each chirp. A higher BW will increase the data rate and
decrease the transmission time on-air for a packet. It will also decrease the
decoding sensitivity since the radio signal is exposed more to noise. Using a
low BW for the same size packet means longer transmission time and a higher
risk that the receiver will fall out of synchronization due to imperfect receiver
clock drift.
• Spreading Factor (SF): The spreading Factor (SF) is the ratio between the
chip rate and the underlying symbol rate. If we increase the SF (i.e. more
bits per symbol), the Signal to Noise Ratio (SNR) will be increased, and
consequently, it will increase the range and the sensitivity, but also the time
on-air to send a symbol.
• Coding Rate (CR) LoRa uses Forward Error Correction (FEC) for the pay-
load. The level of FEC is set by the Coding Rate (CR) parameter. CR
increases robustness against interference but increases the time on-air when
more redundant bits are used for corrections.

By varying these parameters, it is possible to determine the optimal mode of


the operation for different data transmission conditions between the transmit-
ter and the receiver. In LoRa, the chirp rate depends only on the bandwidth:
the chirp rate is equal to the bandwidth (one chirp per second per Hertz of
bandwidth). The symbol rate and the bit rate at a given spreading factor are
proportional to the frequency bandwidth, so a doubling of the bandwidth will
152 V. D. Pham et al.

effectively double the transmission rate. The duration of a symbol (Ts) to the
bandwidth and the spreading factor is shown in formula (1).

2SF
TS = (1)
BW
Moreover, LoRa includes a forward error correction code. The code rate (CR)
4
equals 4+n , with n ∈ {1, 2, 3, 4}.
In the formula (2), the relationship between parameters BW, SF, and CR is
determined:
BW
DR = SF × SF × CR (2)
2
These parameters also influence decoder sensitivity. Therefore, increasing
BW will decrease receiver sensitivity, and increasing the spreading factor will
raise the receiver sensitivity.

3 LoRa Mesh Networks


In this work, we consider the use of LoRa to expand the coverage of the sensor
network. The proposed network model is a mesh network. Why the topology is
mesh?
In the IoT networks, we have two popular types of topology: star and mesh.
Star topology is used in the IoT networks. The advantage of star topology is
that all the network complexity is driven to a central node, so all the other nodes
only need to communicate in their time or frequency slot. How they communicate
depends on whether wireless multiplexing is done through frequency-division
multiple access, time-division multiple access, or code-division multiple access.
The primary disadvantage of star topology is that the radio link between the
gateway and the end node or terminal can be very long, which means the further
a node is away from the gateway, the more energy it has to expend relaying a
message. Also, LoRa devices may still be unable to communicate wirelessly with
a nearby gateway due to physical obstacles that can attenuate wireless signal
strength and result in data losses and communication errors [9].
That is the reason why most of the existing technologies, nowadays, are
based on mesh networks [5]. Mesh networking is a solution for increasing the
communication range and packet delivery ratio without installing additional
LoRa gateways. In the mesh network, the infrastructure node can relay a packet
and cooperate with other nodes to efficiently route a packet to the gateways. Each
node acts as a repeater. Mesh networks dynamically connect end-nodes and self-
configure the routing paths [1]. The primary advantage of mesh topology is that it
has low transmitted power and shorter links, which allows for a fairly long battery
life and enables us to move much data around the network. Moreover, routing
and configuration functionalities are dedicated to wireless mesh networks, and
the load on mesh clients is decreased considerably due to this dedication. This
makes LoRaWAN mesh one of the best ways to collect data from many remote
sensors at once.
Research on Using the AODV Protocol for a LoRa Mesh Network 153

4 Routing in LoRa Mesh Networks


Designing and developing routing protocols in wireless mesh networks is a chal-
lenging issue that should cover multiple performance metrics such as minimum
hop count, preventing disruption of the service methods according to robustness
concepts. Using the mesh infrastructure to perform routing processes as effi-
ciently as possible, and increasing the scalability of routing protocols to install
or maintain routing paths in a mesh network with large capacity [6]. To imple-
ment a LoRa mesh network model with the LoRa gateways, we need to choose
the appropriate routing protocol.
Typically, ad-hoc routing protocols are divided into three categories based on
the network topology information used for route discovery: proactive, reactive,
or hybrid.
• Proactive Routing Protocols: This kind of protocol periodically exchanges
the topology information between all the network nodes. Therefore, proactive
routing protocol has no route discovery since the destination route is saved
and maintained within a table. The tables usually must be updated. These
protocols are used where the route requirements are frequent. However, the
drawback of this protocol is that it gives low idleness to constant application
[12]. DSDV (Destination Sequenced Distance Vector), OLSR (Optimized Link
State Routing) are examples.
• Reactive Routing Protocols: These routing protocols choose routes to other
nodes only when they are needed. A route discovery process is launched when
a node wants to communicate with another station for which it does not
possess any route table access. AODV (Ad-hoc On-Demand Distance Vector
routing protocol), DSR (Dynamic Source Routing) are examples.
• Hybrid Routing Protocols: Hybrid Routing joining nearby proactive routing
protocols and global reactive routing protocols to reduce routing overhead and
delay due to route disclosure process. The advantages of these protocols are
higher efficiency and scalability. However, the disadvantage is high latency for
locating new routes [4]. Examples of these protocols are ZRP (Zone Routing
Protocol).
There are many protocols within each of these three categories, and a com-
prehensive review cannot be provided here. We will instead examine and evaluate
a representative selection of protocols commonly used and referenced in the lit-
erature.
There are many kinds of research focused on the comparison of the proto-
cols mentioned above. In [7,8], the authors found that AODV has shown better
performance than other routing protocols (DSR, DSDV) with the increment
of nodes number in all performance metrics (end-to-end delay, routing load,
received packets, packet delivery ratio, dropped packets). In [19], Singh et al.
showed that although DSR performs well in quick transmission, but it has high
packet loss. Jogendra Kumar in [11] proved that AODV has the best execution as
far as normal jitter and end-to-end delay in comparison with DYMO (Dynamic
MANET On-Demand Routing Protocol), DSR, OSLR, ZRP.
154 V. D. Pham et al.

In [14], Makodia et al. even pointed out that AODV is advised for secured
communication. In addition, AODV also gives a good value of average through-
put [21].
In addition to those evaluations, we can see that, due to its characteristics,
OSLR is well-suited to large and dense networks with random and sporadic
traffic. However, we need a few gateways to cover even a large area in the LoRa
network. As such, the added overhead of choosing relays and updating topology
information is unnecessary in our case [13].
Although DSDV has lower control overhead than OLSR [16], continuous
updates are nonetheless unnecessary for networks with static nodes, as in a
typical LoRa deployment scenario.
Since reactive protocols require sharing of topology information only when
routes fail or a new route needs to be established, they allow for a reduced control
overhead, and thus energy cost, in comparison to proactive protocols [15].
DSR protocol is designed for a network with potentially high mobility. There-
fore it does not suit out case [13].
In summary, although the comparisons are made mainly in MANET and
VANET networks, it also partly points out the advantages and disadvantages of
the protocols. From those evaluations, perhaps, the suitable routing protocol for
our LoRa mesh network is AODV. This is the reason why we choose to use the
AODV protocol in the LoRa mesh network.

5 Network and Simulation Model

5.1 Network Model

Currently, mesh networks are popularly used in various IoT applications. In


many cases, sensor networks are deployed far away from the access point to the
Internet. In these cases, a model is proposed based on mesh networks in two
segments. As shown in Fig. 1, sensor networks communicate with the gateways
to connect to the server cloud. The gateways are able to communicate with each
other in the mesh network while some gateways have access to the Internet.
Moreover, using long-range communication as LoRa for the gateway network is
the main idea to expand the coverage.
In particular, the mesh network can provide connectivity for IoT devices in
smart sustainable cities applications. Using the LoRa technology as a communi-
cation method at the physical layer, devices may communicate with the others
over a hundred meters. As shown in Fig. 2, a network of devices is deployed
in an area of the city Saint-Petersburg. Each node is equipped with the LoRa
interface configured with the same parameters. We can consider these devices
used to collect data from the short-range communication sensors, then collected
data are transmitted to the sink node connected to the external network to the
remote server. The AODV protocol is proposed to establish routing paths to the
sink node.
Research on Using the AODV Protocol for a LoRa Mesh Network 155

Internet

Server Cloud
Internet

Gateway Mesh
Network

User

Sensor Mesh
Networks

Fig. 1. Network model

Sink

Fig. 2. Deployment area

5.2 Simulation Methodology and Parameter

The frameworks OMNET++ and inet are used to carry out the network sim-
ulation. They are known well to be used in numerous domains for simulating
wired and wireless networks. Based on these frameworks and the work in [20],
we have developed a module of the LoRa node. Since the OMNET ++ library
and frameworks are designed based on modular and component-oriented princi-
ples, the LoRa node can be integrated with the build-in modules from the inet
framework.
156 V. D. Pham et al.

The LoRa node consists of modules modeling radio and upper-layer protocols.
In the LoRa radio module, we can set radio parameters that correspond to our
hardware model. The other parameters, such as spreading factor and coding rate
configured for the LoRa node in addition to the usual configurations. According
to the datasheet of SX127x LoRa chipset [18], the bandwidth and spreading
factor influence the reception sensitivity. The receiver has high sensitivity when
the low bandwidth and the high spreading factor are used in the network. The
data rate is also decreased while the communication range increases. Analyzing
from work in [10], we have chosen to use 125 kHz and 8 configured for the
bandwidth and the spreading factor, respectively.
Moreover, in the LoRa medium module, the path loss propagation model is
configured in considering the wireless signal propagation in the urban environ-
ment. In this work, the propagation model parameters have been received from
a series of measurements performed in [3]. In particular, the measurements in
[3] correspond to the build-up urban environment, where devices are partially
deployed indoors. The simulation parameters and values used in this work are
presented in Table 1.
A series of experiments are carried out with considering two cases:
– Case 1: the number of nodes in the network is changed, while the data packets
are generated with random length. The payload size is in the interval from
20 bytes to 150 bytes.
– Case 2: while the number of nodes is constant in the network, the payload
size is set in {20, 40, 60, 150} bytes corresponding for each experiment.

The interval between messages is generated randomly according to the expo-


nential distribution with a mean equal to 120 s. In a field with a size of 2000 ×
2000 m2 , the coordinates of the node A and the sink node are fixed. The node
A is located at the position (400, 400), and the sink node is located at (1500,
1500). Hence, it is required to have relay nodes to communicate between them
in the considered propagation environment.
We have analyzed the end-to-end delay and packet loss ratio from the node
A to the sink node in each experiment. The obtained results are shown in the
next section.

6 Result Analysis

6.1 Analysis by Varying the Number of Nodes

The experiments were performed with a changing number of nodes in the net-
work. The coordinates of nodes were generated randomly according to the uni-
form distribution in the interval (0, 2000). Figure 3 shows a histogram of the
delay required to deliver the data packet from the node A to the sink node.
According to Fig. 3a, the delay does not change much when increasing the num-
ber of nodes in the network. Based on the probability density histogram, we
can see that the delay varying in the interval to 2 s has a probability equal to
Research on Using the AODV Protocol for a LoRa Mesh Network 157

Table 1. Simulation parameters and values

Module Parameter Value


LoRa radio Frequency, MHz 868
Bandwidth (BW), kHz 125
Spreading factor (SF) 8
Coding rate (CR) 4/5
Transmission power, dBm 20
Antenna, dBi 5
LoRa medium Log-distance path loss model PL(d0) = 127.41 dB
d0 = 40 m
gamma = 2.08
sigma = 3.57
Background noise, dBm −96.616
Application Payload length, byte Case 1: randint(20, 150)
Case 2: {20, 40, 60, 150}
Send interval, s Exponential(120s)
AODV maxJitter, s 5
helloInterval, s 60
activeRouteTimeout, s 200
rreqRetries 4
timeoutBuffer, s 10
Others Number of nodes {15, 20, 25, 30}
Field simulation size, m2 2000 × 2000
Simulation time, s 50000

(a) Latency Distribution (b) Packet Loss Ratio

Fig. 3. Analysis by varying number of nodes

2 · 0.45 = 0.9. However, the packet loss ratio increases when increasing the num-
ber of nodes in the network, as shown in Fig. 3b. Adding the number of nodes
significantly affects the change of the packet loss ratio.
158 V. D. Pham et al.

6.2 Analysis by Varying Payload Length

In the second case, with the fixed number of nodes in the network, a series of
experiments was performed with changing payload length generated randomly
in the first case. In this case, the delay distribution and packet loss ratio are
presented in Fig. 4a and b, respectively. The delivery delay can reach up to
several seconds via relay nodes to the destination. Besides, the packet loss ratio
increases when increasing the payload length. However, the data packet size
does not significantly affect the change of the packet loss ratio. Comparing the
sending packets of size 20 and 150 bytes shows that the packet size does not
greatly affect the probability of loss in this considered network.

(a) Latency Distribution (b) Packet Loss Ratio

Fig. 4. Analysis by varying payload length

7 Conclusion

This paper presented the results of studying a LoRa mesh network using the
AODV protocol to find a route from a source node to a sink node. Mostly LoRa-
based networks are deployed in the star topology. However, taking the advantages
of long-range communication and low power consumption, LoRa technology was
considered to expand sensor network coverage. A LoRa mesh network was pro-
posed to transmit sensor data from different clusters to the sink node connected
to the Internet. The study was conducted in a simulation model developed based
on the frameworks omnet++ and inet. The results of the experiments showed
that the end-to-end delay is relatively high in the LoRa mesh network. More-
over, changing the number of nodes in the network and payload size affected the
packet loss ratio.
Based on the study results, the AODV protocol might be used for the LoRa
mesh network. However, the delay needs to be considered in such a network.
Looking into the future, we intend to consider in developing the other protocol
that has compatibility with both the LoRa mesh network and LoRaWAN.
Research on Using the AODV Protocol for a LoRa Mesh Network 159

Acknowledgment. The publication has been prepared with the support of the grant
from the President of the Russian Federation for state support of leading scientific
schools of the Russian Federation according to the research project SS-2604.2020.9.

References
1. Alotaibi, E., Mukherjee, B.: A survey on routing algorithms for wireless ad-hoc
and mesh networks. Comput. Netw. 56(2), 940–965 (2012)
2. Berni, A., Gregg, W.: On the utility of chirp modulation for digital signaling. IEEE
Trans. Commun. 21(6), 748–751 (1973)
3. Bor, M.C., Roedig, U., Voigt, T., Alonso, J.M.: Do LoRa low-power wide-area
networks scale? In: Proceedings of the 19th ACM International Conference on
Modeling, Analysis and Simulation of Wireless and Mobile Systems, pp. 59–67
(2016)
4. Chandra, A., Thakur, S.: Qualitative analysis of hybrid routing protocols against
network layer attacks in MANET. Apoorva Chandra et al., Int. J. Comput. Sci.
Mob. Comput. 4(6), pp. 538–543 (2015)
5. Devalal, S., Karthikeyan, A.: LoRa technology-an overview. In: 2018 Second Inter-
national Conference on Electronics, Communication and Aerospace Technology
(ICECA), pp. 284–290. IEEE (2018)
6. Eslami, M., Karimi, O., Khodadadi, T.: A survey on wireless mesh networks: Archi-
tecture, specifications and challenges. In: 2014 IEEE 5th Control and System Grad-
uate Research Colloquium, pp. 219–222. IEEE (2014)
7. Ferreiro-Lage, J.A., Gestoso, C.P., Rubiños, O., Agelet, F.A.: Analysis of unicast
routing protocols for VANETs. In: 2009 Fifth International Conference on Net-
working and Services, pp. 518–521. IEEE (2009)
8. Haerri, J., Filali, F., Bonnet, C.: Performance comparison of AODV and OLSR in
VANETs urban environments under realistic mobility patterns. In: Proceedings of
the 5th IFIP mediterranean ad-hoc networking workshop, pp. 14–17 (2006)
9. Ke, K.H., Liang, Q.W., Zeng, G.J., Lin, J.H., Lee, H.C.: A LORA wireless mesh
networking module for campus-scale monitoring: demo abstract. In: Proceedings
of the 16th ACM/IEEE International Conference on Information Processing in
Sensor Networks, pp. 259–260 (2017)
10. Kirichek, R., Vishnevsky, V., Pham, V.D., Koucheryavy, A.: Analytic model of a
mesh topology based on LoRa technology. In: 2020 22nd International Conference
on Advanced Communication Technology (ICACT), pp. 251–255. IEEE (2020)
11. Kumar, J.: Comparative performance analysis of AODV, DSR, DYMO, OLSR
and ZRP routing protocols in manet using varying pause time. Int. J. Comput.
Commun. Netw. (IJCCN) 3(1), 43–51 (2012)
12. Kumar, R., Dave, M.: A comparative study of various routing protocols in VANET.
arXiv preprint arXiv:1108.2094 (2011)
13. Lundell, D., Hedberg, A., Nyberg, C., Fitzgerald, E.: A routing protocol for LoRa
mesh networks. In: 2018 IEEE 19th International Symposium on A World of Wire-
less, Mobile and Multimedia Networks (WoWMoM), pp. 14–19. IEEE (2018)
14. Makodia, B., Patel, T., Parmar, K., Hadia, S., Shah, A.: Implementing and analyz-
ing routing protocols for self-organized vehicular adhoc network. In: 2013 Nirma
University International Conference on Engineering (NUiCONE), pp. 1–6. IEEE
(2013)
15. Pandey, K., Swaroop, A.: A comprehensive performance analysis of proactive, reac-
tive and hybrid MANETs routing protocols. arXiv preprint arXiv:1112.5703 (2011)
160 V. D. Pham et al.

16. Perkins, C.E., Bhagwat, P.: Highly dynamic destination-sequenced distance-vector


routing (DSDV) for mobile computers. ACM SIGCOMM Comput. Commun. Rev.
24(4), 234–244 (1994)
17. Pham, V.D., Dinh, T.D., Kirichek, R., et al.: Method for organizing mesh topology
based on LoRa technology. In: 2018 10th International Congress on Ultra Mod-
ern Telecommunications and Control Systems and Workshops (ICUMT), pp. 1–6.
IEEE (2018)
18. Semtech Corporation: Datasheet SX1276/77/78/79 LoRa Transciever, p. 132
(2019)
19. Singh, P.K., Lego, K., Tuithung, T.: Simulation based analysis of adhoc routing
protocol in urban and highway scenario of VANET. Int. J. Comput. Appl. 12(10),
42–49 (2011)
20. Slabicki, M., Premsankar, G., Di Francesco, M.: Adaptive configuration of LoRa
networks for dense IoT deployments. In: NOMS 2018–2018 IEEE/IFIP Network
Operations and Management Symposium, pp. 1–9. IEEE (2018)
21. Talooki, V.N., Ziarati, K.: Performance comparison of routing protocols for mobile
ad hoc networks. In: 2006 Asia-Pacific Conference on Communications, pp. 1–5.
IEEE (2006)

You might also like