Flying ad-hoc networks (FANETs): a survey

Full text

Survey Paper
Flying Ad-Hoc Networks (FANETs): A survey
_
Ilker Bekmezci
a,
⇑
, Ozgur Koray Sahingoz
a
,Sßamil Temel
b
a
Computer Engineering Department, Turkish Air Force Academy, 34149 Yesilyurt, Istanbul, Turkey
b
Turkish Air Force Aeronautics and Space Technologies Institute (ASTIN), 34149 Yesilyurt, Istanbul, Turkey
article info
Article history:
Received 11 December 2012
Accepted 13 December 2012
Available online 3 January 2013
Keywords:
Ad-hoc networks
MANET
VANET
Multi-UAV
abstract
One of the most important design problems for multi-UAV (Unmanned Air Vehicle)
systems is the communication which is crucial for cooperation and collaboration between
the UAVs. If all UAVs are directly connected to an infrastructure, such as a ground base or a
satellite, the communication between UAVs can be realized through the in-frastructure.
However, this infrastructure based communication architecture restricts the capabilities
of the multi-UAV systems. Ad-hoc networking between UAVs can solve the problems
arising from a fully infrastructure based UAV networks. In this paper, Flying Ad-Hoc
Networks (FANETs) are surveyed which is an ad hoc network connecting the UAVs. The
differences between FANETs, MANETs (Mobile Ad-hoc Networks) and VANETs (Vehicle
Ad-Hoc Networks) are clarified first, and then the main FANET design challenges are
introduced. Along with the existing FANET protocols, open research issues are also
discussed.
Ó2013 Elsevier B.V. All rights reserved.
1. Introduction
As a result of the rapid technological advances on elec-
tronic, sensor and communication technologies, it has been
possible to produce unmanned aerial vehicle (UAV) sys-
tems, which can fly autonomously or can be operated re-
motely without carrying any human personnel. Because
of their versatility, flexibility, easy installation and rela-
tively small operating expenses, the usage of UAVs prom-
ises new ways for both military and civilian applications,
such as search and destroy operations [1], border surveil-
lance [2], managing wildfire [3], relay for ad hoc networks
[4,5], wind estimation [6], disaster monitoring [7], remote
sensing [8] and traffic monitoring [9]. Although single-UAV
systems have been in use for decades, instead of develop-
ing and operating one large UAV, using a group of small
UAVs has many advantages. However, multi-UAV systems
have also unique challenges and one of the most
prominent design problems is communication. In this
paper, Flying Ad-Hoc Network (FANET), which is basically
ad hoc network between UAVs, is surveyed as a new
network family. The differences between Mobile Ad-hoc
Network (MANET), Vehicular Ad-hoc Network (VANET)
and FANET are outlined, and the most important FANET
design challenges are introduced. In addition to the exist-
ing solutions, the open research issues are also discussed.
Along with the progress of embedded systems and the
miniaturization tendency of microelectromechanical sys-
tems, it has been possible to produce small or mini UAVs
at a low cost. However, the capability of a single small
UAV is limited. Coordination and collaboration of multiple
UAVs can create a system that is beyond the capability of
only one UAV. The advantages of the multi-UAV systems
can be summarized as follows:
Cost: The acquisition and maintenance cost of small
UAVs is much lower than the cost of a large UAV [10].
Scalability: The usage of large UAV enables only limited
amount of coverage increases [11]. However, multi-
UAV systems can extend the scalability of the operation
easily.
1570-8705/$ - see front matter Ó2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.adhoc.2012.12.004
⇑
Corresponding author.
E-mail addresses: i.bekmezci@hho.edu.tr (_
I. Bekmezci), sahingoz@
hho.edu.tr (O.K. Sahingoz), s.temel@hho.edu.tr (Sß. Temel).
Ad Hoc Networks 11 (2013) 1254–1270
Contents lists available at SciVerse ScienceDirect
Ad Hoc Networks
journal homepage: www.elsevier.com/locate/adhoc

Survivability: If the UAV fails in a mission which is
operated by one UAV, the mission cannot proceed.
However, if a UAV goes off in a multi-UAV system, the
operation can survive with the other UAVs.
Speed-up: It is shown that the missions can be com-
pleted faster with a higher number of UAVs [12].
Small radar cross-section: Instead of one large radar
cross-section, multi-UAV systems produce very small
radar cross-sections, which is crucial for military appli-
cations [13].
Although there are several advantages of multi-UAV
systems, when compared to single-UAV systems, it has
also unique challenges, such as communication. In a sin-
gle-UAV system, a ground base or a satellite is used for
communication. It is also possible to establish a communi-
cation link between the UAV and an airborne control sys-
tem. In all cases, single-UAV communication is
established between the UAV and the infrastructure. While
the number of UAVs increases in unmanned aerial systems,
designing efficient network architectures emerges as a vi-
tal issue to solve.
As in a single UAV system, UAVs can also be linked to a
ground base or to a satellite in a multi-UAV system. There
may be variants of this star topology based solution [14].
While some UAVs communicate with a ground base, the
others can communicate with satellite/s. In this approach,
UAV-to-UAV communication is also realized through the
infrastructure. There are several design problems with this
infrastructure based approach. First of all, each UAV must
be equipped with an expensive and complicated hardware
to communicate with a ground base or a satellite. Another
handicap about this network structure is the reliability of
the communication. Because of the dynamic environmen-
tal conditions, node movements and terrain structures,
UAVs may not maintain its communication link. Another
problem is the range restriction between the UAVs and
the ground base. If a UAV is outside the coverage of the
ground base, it becomes disconnected. An alternative com-
munication solution for multi-UAV systems is to establish
an ad hoc network between UAVs, which is called FANET.
While only a subset of UAVs can communicate with the
ground base or satellite, all UAVs constitute an ad hoc net-
work. In this way, the UAVs can communicate with each
other and the ground base.
FANET can be viewed as a special form of MANET and
VANET. However, there are also certain differences be-
tween FANET and the existing ad hoc networks:
Mobility degree of FANET nodes is much higher than
the mobility degree of MANET or VANET nodes. While
typical MANET and VANET nodes are walking men
and cars respectively, FANET nodes fly in the sky.
Depending on the high mobility of FANET nodes, the
topology changes more frequently than the network
topology of a typical MANET or even VANET.
The existing ad hoc networks aim to establish peer-to-
peer connections. FANET also needs peer-to-peer con-
nections for coordination and collaboration of UAVs.
Besides, most of the time, it also collects data from
the environment and relays to the command control
center, as in wireless sensor networks [15]. Conse-
quently, FANET must support peer-to-peer communica-
tion and converge cast traffic at the same time.
Typical distances between FANET nodes are much
longer than in the MANETs and VANETs [16]. In order
to establish communication links between UAVs, the
communication range must also be longer than in the
MANETs and VANETs. This phenomenon accordingly
affects the radio links, hardware circuits and physical
layer behavior.
Multi-UAV systems may include different types of sen-
sors, and each sensor may require different data deliv-
ery strategies.
The main motivation of this paper is to define FANET as
a separate ad hoc network family and to introduce unique
challenges and design constraints. Although, there exists a
few studies that address some specific issues of networked
UAVs [17,18,14], to the best of our knowledge, this is the
first comprehensive survey about FANETs.
The paper is organized as follows. In Section 2, we pres-
ent several FANET application scenarios and introduce FA-
NET design characteristics in Section 3. We provide an
extensive review of the existing communication protocols
and the open research issues in Section 4. We also present
the existing multi-UAV test beds and simulation environ-
ments in Section 5. We conclude the paper in Section 6.
2. FANET application scenarios
In this section, different FANET application scenarios
are discussed.
2.1. Extending the scalability of multi-UAV operations
If a multi-UAV communication network is established
fully based on an infrastructure, such as a satellite or a
ground base, the operation area is limited to the communi-
cation coverage of the infrastructure. If a UAV cannot com-
municate with the infrastructure, it cannot operate. On the
other hand, FANET is based on the UAV-to-UAV data links
instead of UAV-to-infrastructure data links, and it can ex-
tend the coverage of the operation. Even if a FANET node
cannot establish a communication link with the infrastruc-
ture, it can still operate by communicating through the
other UAVs. This scenario is illustrated in Fig. 1.
There are several FANET designs developed for extend-
ing the scalability of multi-UAV applications. In [19], a FA-
NET design was proposed for the range extension of multi-
UAV systems. It was stated that forming a link chain of
UAVs by utilizing multi-hop communication can extend
the operation area.
It should be noticed that the terrain also affects the
communication coverage of the infrastructure. There may
be some obstacles on the terrain, such as mountains, walls
or buildings, and these obstacles may block the signals of
the infrastructures. Especially in urban areas, buildings
and constructions block the radio signals between the
ground base and UAVs. FANET can also help to operate be-
_
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270 1255

hind the obstacles, and it can extend the scalability of mul-
ti-UAV applications [20].
2.2. Reliable multi-UAV communication
In most of the cases, multi-UAV systems operate in a
highly dynamic environment. The conditions at the begin-
ning of a mission may change during the operation. If there
is no opportunity to establish an ad hoc network, all UAVs
must be connected to an infrastructure, as illustrated in
Fig. 2a. However, during the operation, because of the
weather condition changes, some of the UAVs may be dis-
connected. If the multi-UAV system can support FANET
architecture, it can maintain the connectivity through the
other UAVs, as it is shown in Fig. 2b. This connectivity fea-
ture enhances the reliability of the multi-UAV systems [16].
2.3. UAV swarms
Small UAVs are very light and have limited payload
capacity. In spite of their restricted capabilities, the swarm
behavior of multiple small UAVs can accomplish complex
missions [21]. Swarm behavior of UAVs requires coordi-
nated functions, and UAVs must communicate with each
other to achieve the coordination. However, because of
the limited payloads of small UAVs, it may not be possible
to carry heavy UAV-to-infrastructure communication
hardware. FANET, which needs relatively lighter and
cheaper hardware, can be used to establish a network be-
tween small UAVs. By the help of the FANET architectures,
swarm UAVs can prevent themselves from collisions, and
the coordination between UAVs can be realized to com-
plete the mission successfully.
In [22], Cooperative Autonomous Reconfigurable UAV
Swarm (CARUS) is proposed with FANET communication
architecture. The objective of CARUS is the surveillance of
a given set of points. Each UAV operates in an autonomous
manner, and the decisions are taken by each UAV in the air
rather than on the ground. Ben-Asher et al. have intro-
duced a distributed decision and control mechanism for
multi-UAV systems using FANET [23].In[24], a FANET
based UAV swarm architecture is proposed to convey UAVs
to a target location with cooperative decision-making.
Quaritsch et al. have developed another FANET based
UAV swarm application for disaster management [25].
During a disaster situation, rescue teams cannot rely on
fixed infrastructures. The aim of the project is to provide
quick and accurate information from the affected area.
2.4. FANET to decrease payload and cost
The payload capacity problem is not valid only for small
UAVs. Even High Altitude Low Endurance (HALE) UAVs
must consider payload weights. The lighter payload means
the higher altitude and the longer endurance [16]. If the
communication architecture of a multi-UAV system is fully
based on UAV-to-infrastructure communication links, each
UAV must carry relatively heavier communication hard-
ware. However, if it uses FANET, only a subset of UAVs
use UAV-to-infrastructure communication link, and the
other UAVs can operate with FANET, which needs lighter
communication hardware in many cases. In this way, FA-
NET can extend the endurance of the multi-UAV system.
3. FANET design characteristics
Before discussing the characteristics of FANETs, we pro-
vide a formal definition of FANET and a brief discussion
about the definition to understand FANET clearly.
FANET can be defined as a new form of MANET in which
the nodes are UAVs. According to this definition, single-
UAV systems cannot form a FANET, which is valid only
for multi-UAV systems. On the other hand, not all multi-
UAV systems form a FANET. The UAV communication must
be realized by the help of an ad hoc network between
UAVs. Therefore, if the communication between UAVs fully
relies on UAV-to-infrastructure links, it cannot be classified
as a FANET.
In the literature, FANET related researches are studied
under different names. For example, aerial robot team is
a collaborative and autonomous multi-UAV system, and
generally, its network architecture is ad hoc [26]. In this
sense, ad hoc based aerial robot teams can also be viewed
as a FANET design. However, aerial robot team studies
mostly concentrate on the collaborative coordination of
multi-UAV systems, not on the network structures, algo-
rithms or protocols [27]. Another FANET related topic is
Fig. 1. A FANET scenario to extend the scalability of multi-UAV systems.
1256 _
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270

aerial sensor network [28–30]. Aerial sensor network is a
very specialized mobile sensor and actor network so that
the nodes are UAVs. It moves around the environment,
senses with the sensors on the UAVs and relays the col-
lected data to the ground base. In addition, it can act with
its actors on the UAVs to realize its mission. It is a percep-
tion issue to name the problem as flying ad hoc network or
aerial sensor network. The basic design challenges of a tra-
ditional sensor network are energy consumption and node
density [31], and none of them is related with multi-UAV
systems. Generally, UAVs have enough energy to support
its communication hardware, and node density of a mul-
ti-UAV system is very low when it is compared to tradi-
tional sensor networks. Under the light of these
discussions, it is better to classify the multi-UAV commu-
nication system based on UAV-to-UAV links as a special-
ized ad hoc network, instead of a specialized sensor
network. UAV ad hoc network [32] is another topic, which
is closely related to FANETs. In fact, there is no significant
difference between the existing UAV ad hoc network re-
searches and the above FANET definition. However, FANET
term immediately reminds that it is a specialized form of
MANET and VANET. Therefore, we prefer calling it as Flying
Ad-Hoc Network, FANET.
3.1. Differences between FANET and the existing ad-hoc
networks
Wireless ad hoc networks are classified according to
their utilization, deployment, communication and mission
objectives. By definition, FANET is a form of MANET, and
there are many common design considerations for MANET
and FANET. In addition to this, FANET can also be classified
as a subset of VANET, which is also a subgroup of MANET.
This relationship is illustrated in Fig. 3. As an emerging re-
search area, FANET shares common characteristics with
these networks, and it also has several unique design chal-
lenges. In this subsection, the differences between FANET
and the existing wireless ad hoc networks are explained
in a detailed manner.
3.1.1. Node mobility
Node mobility related issues are the most notable dif-
ference between FANET and the other ad hoc networks.
MANET node movement is relatively slow when it is com-
pared to VANET. In FANET, the node’s mobility degree is
much higher than in the VANET and MANET. According
to [16], a UAV has a speed of 30–460 km/h, and this situa-
tion results in several challenging communication design
problems [33].
3.1.2. Mobility model
While MANET nodes move on a certain terrain, VANET
nodes move on the highways, and FANET nodes fly in the
sky. MANETs generally implement the random waypoint
mobility model [34], in which the direction and the speed
of the nodes are chosen randomly. VANET nodes are re-
stricted to move on highways or roads. Therefore, VANET
mobility models are highly predictable.
In some multi-UAV applications, global path plans are
preferred. In this case, UAVs move on a predetermined
path, and the mobility model is regular. In autonomous
multi-UAV systems, the flight plan is not predetermined.
Even if a multi-UAV system uses predefined flight plans,
because of the environmental changes or mission updates,
the flight plan may be recalculated. In addition to the flight
plan changes, the fast and sharp UAV movements and dif-
ferent UAV formations directly affect the mobility model of
multi-UAV systems. In order to address this issue, FANET
mobility models are proposed. In [35], Semi-Random Cir-
cular Movement (SRCM) mobility model is presented,
and the approximate node distribution function is derived
within a two dimensional disk region. In [36], two new
mobility models are proposed for multi-UAV systems. In
random UAV movement model, UAVs move indepen-
dently. Each UAV decides on its movement direction,
according to a predefined Markov process. In the second
model, the UAVs maintain a pheromone map, and the
pheromones guide their movements. Each UAV marks the
areas that it scans on the map, and shares the pheromone
map with broadcasting. In order to maximize the coverage,
UAVs prefer the movement through the areas with low
pheromone smell. It was shown that the use of a typical
MANET mobility model may result in undesirable path
plans for cooperative UAV applications. It was also ob-
served that the random model is remarkably simple, but
it leads to ordinary results. However, the pheromone based
model has very reliable scanning properties.
Fig. 2. A FANET application scenario for reliable multi-UAV communication network.
_
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270 1257

3.1.3. Node density
Node density can be defined as the average number of
nodes in a unit area. FANET nodes are generally scattered
in the sky, and the distance between UAVs can be several
kilometers even for small multi-UAV systems [37]. As a re-
sult of this, FANET node density is much lower than in the
MANET and VANET.
3.1.4. Topology change
Depending on the higher mobility degree, FANET topol-
ogy also changes more frequently than MANET and VANET
topology. In addition to the mobility of FANET nodes, UAV
platform failures also affect the network topology. When a
UAV fails, the links that the UAV has been involved in also
fail, and it results in a topology update. As in the UAV fail-
ures, UAV injections also conclude a topology update. An-
other factor that affects the FANET topology is the link
outages. Because of the UAV movements and variations
of FANET node distances, link quality changes very rapidly,
and it also causes link outages and topology changes [38].
3.1.5. Radio propagation model
Differences between FANET and the other ad hoc net-
work operating environments affect the radio propagation
characteristics. MANET and VANET nodes are remarkably
close to the ground, and in many cases, there is no line-
of-sight between the sender and the receiver. Therefore,
radio signals are mostly affected by the geographical struc-
ture of the terrain. However, FANET nodes can be far away
from the ground and in most of the cases, there is a line-of-
sight between UAVs.
3.1.6. Power consumption and network lifetime
Network lifetime is a key design issue for MANETs,
which especially consist of battery-powered computing
devices. Developing energy efficient communication proto-
cols is the goal of efforts to increase the network lifetime.
Especially, while the battery-powered computing devices
are getting smaller in MANETs, system developers have
to pay more attention to the energy efficient communica-
tion protocols to prolong the lifetime of the network. How-
ever, FANET communication hardware is powered by the
energy source of the UAV. This means FANET communica-
tion hardware has no practical power resource problem as
in MANET. In this case, FANET designs may not be power
sensitive, unlike most of the MANET applications. How-
ever, it must be stated that power consumption still can
be a design problem for mini UAVs [39].
3.1.7. Computational power
In ad hoc network concept, the nodes can act as routers.
Therefore, they must have certain computation capabilities
to process incoming data in real-time. Generally, MANET
nodes are battery powered small computers such as lap-
tops, PDAs and smart phones. Because of the size and en-
ergy constraints, the nodes have only limited
computational power. On the other hand, both in VANETs
and FANETs, application specific devices with high compu-
tational power can be used. Most of the UAVs have enough
energy and space to include high computational power.
The only limitation about the computational power is the
weight. By the help of the hardware miniaturization ten-
dency, it is possible to put powerful computation hardware
in UAV platforms. However, the size and weight limitation
can still constitute serious constraints for mini UAVs, that
have very limited payload capacity.
3.1.8. Localization
Accurate geospatial localization is at the core of mobile
and cooperative ad hoc networks [40]. Existing localization
methods include global positioning system (GPS), beacon
(or anchor) nodes, and proximity-based localization [41].
In MANET, GPS is generally used to receive the coordi-
nates of a mobile communication terminal, and most of
the time, GPS is sufficient to determine the location of
the nodes. When GPS is not available, such as in dense fo-
liage areas, beacon nodes or proximity-based techniques
can also be used.
In VANET, for a navigation-grade GPS receiver, there is
about 10–15 m accuracy, which can be acceptable for route
guidance. However, it is not sufficient for cooperative
safety applications, such as collision warnings for cars.
Some researchers use assisted GPS (AGPS) or differential
GPS (DGPS) by using some type of ground-based reference
stations for range corrections with accuracy about 10 cm
[42,43].
Because of the high speed and different mobility models
of multi-UAV systems, FANET needs highly accurate local-
Fig. 3. MANET, VANET and FANET.
1258 _
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270

ization data with smaller time intervals. GPS provides po-
sition information at one-second interval, and it may not
be sufficient for certain FANET protocols. In this case, each
UAV must be equipped with a GPS and an inertial measure-
ment unit (IMU) to offer the position to the other UAVs at
any time. IMU can be calibrated by the GPS signal, and
thus, it can provide the position of the UAV at a quicker
rate [44,45].
Because of the above-mentioned differences between
FANET, MANET and VANET; we prefer to investigate FANET
as a separate ad hoc network family. The differences be-
tween MANET, VANET and FANET are outlined in Table 1.
3.2. FANET design considerations
The distinguishing features of FANET impose unique de-
sign considerations. In this subsection, the most prominent
FANET design considerations; adaptability, scalability, la-
tency, UAV platform constraints, and bandwidth require-
ment are discussed.
3.2.1. Adaptability
There are several FANET parameters that can change
during the operation of a multi-UAV system. FANET nodes
are highly mobile and always change their location. Be-
cause of the operational requirements, the routes of the
UAVs may be different, and the distance between UAVs
cannot be constant.
Another issue that must be considered is the UAV fail-
ures. Consequent to a technical problem or an attack
against multi-UAV system, some of the UAVs may fail dur-
ing the operation. While UAV failures decrease the number
of UAVs, UAV injections may be required to maintain the
multi-UAV system operation. UAV failures and UAV injec-
tions change the FANET parameters.
Environmental conditions can also affect FANET. If the
weather changes unexpectedly, FANET data links may not
survive. FANET should be designed so that it should be able
to continue to operate in a highly dynamic environment.
The mission may also be updated during the multi-UAV
system operation. Additional data or new information
about the mission may require a flight plan update. For
example, while a multi-UAV system is operated for a
search and rescue mission; after the arrival of a new intel-
ligence report, the mission may be concentrated on a cer-
tain area, and the flight plan update also affects FANET
parameters.
FANET design should be developed so that it can adjust
itself against any changes or failures. FANET physical layer
should adapt according to the node density, distance be-
tween nodes, and environmental changes. It can scan the
parameters and choose the most appropriate physical layer
option. The highly dynamic nature of FANET environment
also affects network layer protocols. Route maintenance
in an ad hoc network is closely related to the topology
changes. Thus, the performance of the system depends
on the routing protocol in adapting to link changes. Trans-
port layer should also be adapted according to the status of
FANET.
3.2.2. Scalability
Collaborative work of UAVs can improve the perfor-
mance of the system in comparison to a single-UAV sys-
tem. In fact, this is the main motivation to use multi-UAV
based systems. In many applications, the performance
enhancement is closely related with the number of UAVs.
For example, the higher number of UAVs can complete a
search and rescue operation faster [12]. FANET protocols
and algorithms should be designed so that any number
of UAVs can operate together with minimal performance
degradation.
3.2.3. Latency
Latency is one of the most important design issues for all
types of networks, and FANET is not an exception. FANET la-
tency requirement is fully dependent on the application.
Especially for real-time FANET applications, such as mili-
tary monitoring, the data packets must be delivered within
a certain delay bound. Another low latency requirement is
valid for collision avoidance of multiple UAVs [14,46].
In [47], an analysis of one-hop packet delay was con-
ducted for IEEE 802.11 based FANETs. Each node was mod-
eled as M/M/1 queue and the mean packet delay was
derived analytically. The numerical results were verified
with a simulation analysis. Based on the data collected
from the simulation analysis, it was observed that packet
delay can be approximated with Gamma distribution. Zhai
et al. studied packet delay performance of IEEE 802.11 for
traditional wireless LANs, and stated that the MAC layer
packet service time can be approximated by an exponen-
Table 1
The comparison of MANET, VANET and FANET.
MANET VANET FANET
Node mobility Low High Very high
Mobility model Random Regular Regular for predetermined paths, but special mobility models for
autonomous multi-UAV systems
Node density Low High Very low
Topology change Slow Fast Fast
Radio propagation model Close to ground, Close to ground, High above the ground,
LoS is not available
for all cases
LoS is not available
for all cases
LoS is available for most of the cases
Power consumption and
network lifetime
Energy efficient
protocols
Not needed Energy efficiency for mini UAVs, but not needed for small UAVs
Computational power Limited High High
Localization GPS GPS, AGPS, DGPS GPS, AGPS, DGPS, IMU
_
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270 1259

tially distributed random variable [48]. It also shows that
the packet delay behaviors are different for MANETs and
FANETs, and the protocols developed for MANET may not
satisfy the latency requirements of FANET. New FANET
protocols and algorithms are needed for delay sensitive
multi-UAV applications.
3.2.4. UAV platform constraints
FANET communication hardware must be deployed on
the UAV platform, and this situation imposes certain con-
straints. The weight of the hardware is an important issue
for the performance of the UAVs. Lighter hardware means
lighter payload, and it extends the endurance. Another
opportunity that comes with the lighter communication
hardware is to deploy additional sensors on the UAV. If
the total payload is assumed as constant and the commu-
nication hardware is lighter, more advanced sensors and
other peripherals can be deployed.
Space limitation is another UAV platform related con-
straints for FANET designs. Especially for mini UAVs, the
space limitation is very important for communication
hardware that must be fitted into the UAV platform [39].
3.2.5. Bandwidth requirement
In most of the FANET applications, the aim is to collect
data from the environment and to relay the collected data
to a ground base [25]. For example, in surveillance, moni-
toring or rescue operations; the image or video of the tar-
get area must be relayed from the UAV to the command
control center with a very strict delay bound, and it re-
quires high bandwidth. In addition, by the help of the tech-
nological advancements on sensor technologies, it is
possible to collect data with very high resolution, and this
makes the bandwidth requirement much higher. The col-
laboration and coordination of multiple UAVs also need
additional bandwidth resource.
On the other hand, there are many constraints for the
usage of available bandwidth such as:
capacity of the communication channel,
speed of UAVs,
error-prone structure of the wireless links,
lack of security with broadcast communication.
A FANET protocol must satisfy the bandwidth capacity
requirement so that it can relay very high resolution
real-time image or video under several constraints.
4. Communication protocols for FANETs
In this section, the FANET communication protocols and
the open research issues are presented. We survey the
existing FANET protocols proposed for the physical layer,
medium access control (MAC) layer, network layer, trans-
port layer, and their cross-layer interactions.
4.1. Physical layer
The physical layer deals with the basic signal transmis-
sion technologies, such as modulation or signal coding.
Various data bit sequences can be represented with differ-
ent waveforms by varying the frequency, amplitude and
phase of a signal. Overall, in the physical layer, the data bits
are modulated to sinusoidal waveforms and transmitted
into the air by utilizing an antenna.
MANET system performance is highly dependent on its
physical layer, and the extremely high mobility puts extra
problematic issues on FANET. In order to develop robust
and sustainable data communication architectures for FA-
NET, the physical layer conditions have to be well-under-
stood and well-defined. Recently, UAV-to-UAV and UAV-
to-ground communication scenarios have been broadly
studied in both simulation and real-time environments.
Radio propagation models and antenna structures are
investigated as the key factors that influence FANET phys-
ical layer design.
4.1.1. Radio propagation model
Electromagnetic waves radiate from the transmitter to
the receiver through wireless channels. The characteriza-
tion of radio wave propagation is expressed as a mathe-
matical function, which is called radio propagation
modeling [49]. FANET environment has several unique
challenges in terms of radio propagation when compared
to the other types of wireless networks. Some of the chal-
lenges are summarized as follows:
Variations in communication distance.
Direction of the communicating pairs in the antenna
radiation pattern.
Ground reflection effects.
Shadowing resulting from the UAV platform and on-
board electronic equipment.
The effect of aircraft attitude (pitch, roll, yaw etc.) on
the wireless link quality.
Environmental conditions.
Interferences and hostile jamming.
Because of the above-mentioned factors, communica-
tion links exhibit varying quality over time in FANETs [50].
Ahmed et al. studied the characterization of UAV-to-
UAV, UAV-to-ground, and ground-to-UAV communication
links [51]. In this study, free space and two-ray ground
approximation models are compared for each link type,
and the presence of gray regions is observed, when the
UAVs are close to the ground. Gray regions showed that
the radio propagation model of UAV-to-UAV links is simi-
lar to two-ray ground model, and FANET protocol design-
ers must be aware of the presence of the gray zones due
to fading.
Zhou et al. investigated the channel modeling problem
for UAV-to-UAV communications [52]. In this study, it
was observed that the error statistics of the wireless chan-
nels between UAVs are non-stationary. Depending on the
changes of the distance between UAVs, a two-state Markov
model was proposed to incorporate the effects of Rician
fading, which is suitable for strong line-of-sight path, as
in FANET. The simulation results showed that the proposed
model is able to simulate packet dropouts with non-sta-
tionary error statistics.
1260 _
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270

A Nakagami-m based radio propagation model was also
proposed for FANET communication in [53]. Nakagami-m
suitably agrees with the empirical data measured for VA-
NET networks [54,55]. This model estimated the received
signal strength for a multi-path environment with cover-
ing fading effects, and it was represented as a function of
two parameters: the average received radio signal strength
and the fading intensity. A mathematical expression for the
outage probability over Nakagami-m fading channel has
been derived for a cooperative UAV network.
In [56], the performance analysis of multi-carrier relay
based UAV network was modeled analytically over fading
channels. A general analytical formula was provided for
the outage probability of UAV-to-UAV and UAV-to-ground
link. It was stated that fading channel model should be
chosen according to the operation environment. For exam-
ple, while Rayleigh fading can be more suitable for low alti-
tude crowded area applications, Nakagami-m and Weibull
fading with high fading parameters best fit for high alti-
tude open space missions.
4.1.2. FANET antenna structure
The antenna structure is one of the most crucial factors
for an efficient FANET communication architecture. The
distance between UAVs is longer than typical node dis-
tance of MANETs and VANETs, and it directly affects the
FANET antenna structure. More powerful radios can be
used to overcome this problem, but high link loss and var-
iation could still arise at longer distances. In order to over-
come this phenomenon, multiple receiver nodes can be
deployed to boost packet delivery rates by exploiting the
spatial and temporal diversity of the wireless channel
[57]. It is shown that UAV receiver nodes exhibit poor
packet reception correlation at short time scales, which
ultimately necessitates the usage of multiple transmitters
and receivers to improve packet delivery rates.
Antenna type is another factor that affects the FANET
performance. In the literature, there are two types of
antennas deployed for FANET applications: directional
and omnidirectional. While omnidirectional antennas radi-
ate the power in all directions, directed antenna can send
the signal through a desired direction.
In highly mobile environments, as in FANET, the node
locations change frequently and omnidirectional antennas
have a natural advantage to transmit and receive signals. In
omnidirectional antennas, node location information is not
needed. However, directional antennas also have several
advantages when compared to omnidirectional antennas.
Firstly, the transmission range of a directed antenna is
longer than the transmission range of an omnidirectional
antenna [58]. It can be an important advantage for FANET,
where the distance between nodes is longer than the dis-
tance between typical MANET nodes [37]. The longer
transmission range decreases hop count, and it can en-
hance the latency performance [59]. Especially, in real time
FANET applications, such as military monitoring, latency is
one of the most dominant design factors.
There is a trade-off between communication range and
spatial reuse for omnidirectional antennas [60]. Directional
antenna based systems can handle communication range
and spatial reuse problem for FANETs, at the same time.
While it can increase communication range, it does not
limit spatial reuse [61]. Depending on the higher spatial
reusability, the capacity of a network with directed anten-
na is higher than the capacity of a network with omnidi-
rectional antenna.
Security is another issue that can be enhanced by the
help of the directed antennas. Omnidirectional antenna
based systems are more prone to jamming than the direc-
ted antenna based systems [62]. A brief comparison of
omnidirectional and directional antennas is provided in
Table 2.
4.1.3. Open research issues
The characteristics of the physical layer affect the de-
sign of the other layers and the overall FANET performance
directly. The existing FANET physical layer related studies,
which are summarized in Table 3, concentrate on the radio
propagation models and antenna structures.
Although the nodes are located in a 3D environment in
real FANET applications, most of the existing studies as-
sume 2D FANET topology structures. The FANET studies
have shown that the antenna behaviors in 3D can be differ-
ent from the antenna behaviors in 2D [51], and it can affect
the physical layer directly. The performance analysis of the
existing physical layer protocols and developing new phys-
ical layer designs for 3D are largely unexplored issues for
FANETs.
4.2. MAC layer
Although MANET, VANET and FANET have different
challenges and characteristics, they have also several com-
mon design considerations. Basically, FANET is a special
subset of MANET and VANET. In this sense, the first FANET
examples use IEEE 802.11 with omnidirectional antennas
[34,32], which is one of the most commonly used MAC lay-
ers for MANETs. By the help of the request-to-send (RTS)
and clear-to-send (CTS) signal exchange mechanism, IEEE
802.11 can handle the hidden node problem [63].
4.2.1. Challenges of FANET MAC layer
High mobility is one of the most distinctive properties
of FANET, and it presents new problems for the MAC layer.
Because of the high mobility and the varying distances be-
tween nodes, link quality fluctuations take place in FANETs
frequently. Link quality changes and link outages directly
affect FANET MAC designs. Packet latency is another design
problem for FANET MAC layer design. Especially for real
time applications, packet latency must be bounded and it
Table 2
The comparison of omnidirectional and directional antennas for FANETs.
Attribute Omnidirectional Directional
Signal direction All Desired
Node orientation Not needed Needed
Transmission range Shorter Longer
Latency Higher Lower
Spatial reusability Lower Higher
Capacity Lower Higher
Prone to jamming Higher Lower
_
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270 1261

imposes new challenges. Fortunately, there are new tech-
nologies that can be used to meet the FANET requirements
in MAC layer. Directional antenna and full-duplex radio
circuits with multi-packet reception are some examples
of promising technological advancements that can be used
in FANET MAC layer [58,64].
4.2.2. Directional antenna based FANET MAC layer
Directional antennas have several advantages over
omnidirectional antennas for FANETs, as it is provided in
the physical layer subsection. Besides the advantages of
directional antennas, it also brings unique design prob-
lems, especially for the MAC layer. An extensive survey
about directional antenna based MAC protocols can be
found in [65].
While most of the existing directional antenna based
MAC layers are proposed for MANET and VANET, there
are also a few researches about FANET MAC layer design
with directional antennas. In [66], Alshbatat and Dong
have proposed Adaptive MAC Protocol Scheme for UAVs
(AMUAV) [66]. While AMUAV sends its control packages
(RTS, CTS, and ACK) with its omnidirectional antenna,
DATA package is sent by directional antennas. It is proved
that directed antenna based AMUAV protocol can improve
throughput, end-to-end delay and bit error rate for multi-
UAV systems.
4.2.3. MAC layer with full-duplex radio and multi-packet
reception
In traditional wireless communication, reception and
transmission cannot be performed at the same time. With
the recent advancements on the radio circuits, it is now
possible to realize full-duplex wireless communication on
the same channel [58]. Another restriction of the tradi-
tional wireless communication is about the packet recep-
tion. If there is more than one sender, the receiver cannot
receive the data correctly. Fortunately, data reception from
more than one source is possible by the help of the multi-
packet reception (MPR) radio circuits [64]. Full-duplex and
MPR radio circuits have significant impacts on the FANET
MAC layer.
Channel state information (CSI) is one of the most
important parameters for full-duplex radios, and it is al-
most impossible to determine the perfect CSI, in highly dy-
namic environments, as in FANETs. In [67], a new token-
based FANET MAC layer was proposed with full-duplex
and multi-packet reception (MPR) radios. It aims frequent
CSI update so that UAVs can have the latest CSI information
at any time. Token-based structure of CSI updates elimi-
nates packet collisions. Performance results have shown
the effectiveness of the proposed MAC layer, even if the
resulting channel knowledge is imperfect.
4.2.4. Open research issues
Providing a robust FANET MAC layer necessitates to ad-
dress and overcome some unique challenging tasks such as
link quality variations caused by high mobility, and longer
distance between nodes. Although the first FANET test
beds have used IEEE 802.11 with omnidirectional anten-
nas, it cannot respond to the requirements of FANETs.
There are only a few studies about FANET MAC layers
which are presented in Table 4.
In order to overcome the unique challenges of FANET,
directed antenna technology, which can send the signal
to a desired direction, is a promising technology. Location
estimation of the nodes and sharing this information are
vital issues for directed antenna based MAC layers, and
they are more challenging for FANETs, where the nodes
are highly mobile. Most of the existing directed antenna
based MAC layers assume that the location information is
maintained by the upper layers and cannot offer a robust
and integrated solution in the MAC layer [65]. Localization
service can be integrated in the MAC layer to find the loca-
tions of the other UAVs that are constantly changing their
coordinates.
Although there are several unique challenges of FANET,
it has also a number of opportunities for MAC layer design.
In most of the MANET designs, energy is one of the most
considerable constraints. However, FANET protocols have
to work on UAVs and there is no practical energy restric-
tion on UAVs. FANET nodes can include and operate more
advanced hardware than the MANET nodes. This opportu-
nity can be used to develop more efficient FANET MAC
layers.
4.3. Network layer
The initial FANET studies and experiments are designed
with the existing MANET routing protocols.
One of the first flight experiments with FANET architec-
ture is performed in SRI International [68]. In this research,
Topology Broadcast based on Reverse-Path Forwarding
(TBRPF) [69], which is basically a proactive protocol, is
used as the network layer to minimize the overhead. In
Table 3
An overview of physical layer related studies for FANETs.
Physical layer study Short description
Characterization of FANET communication
links [51]
The gray regions were observed in FANET experiments and it showed that the radio propagation model
of UAV-to-UAV links is similar to two-ray ground model, rather than free space model
Channel modeling of FANET links [52] Rician fading based two-state Markov model was developed to model wireless channel between UAVs.
The simulations showed that the proposed model can simulate packet dropouts
Nakagami-m based FANET radio
propagation model [54,55,53]
A mathematical expression for the link outage probability over Nakagami-m fading channel was derived
for FANETs
General link outage model for FANETs [56] A general analytical formula was provided for the outage of UAV-to-UAV and UAV-to-ground links over
various fading channels. Rayleigh, Nakagami-m, and Weibull models were studied as fading channels
Multiple transmitters and receivers [57] UAV receiver nodes can achieve poor packet reception correlation at short time scales. The usage of
multiple transmitters and receivers improves packet delivery rates dramatically
1262 _
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270

[70], Brown et al. developed another FANET test bed with
Dynamic Source Routing (DSR) [71] protocol. The main
motivation to choose DSR is its reactive structure. The
source tries to find a path to a destination, only if it has
data to send. There are also some other FANET studies that
use DSR. Khare et al. stated that DSR is more appropriate
than proactive methods for FANETs, where the nodes are
highly mobile, and the topology is unstable [72].
Because of the high mobility of the FANET nodes, main-
taining a routing table, as in proactive methods, is not opti-
mal. However, repetitive path finding before each packet
delivery, as in reactive routing, can also be exhaustive. A
routing strategy only based on the location information
of the nodes can satisfy the requirements of FANET. In
[73], proactive, reactive and position-based routing solu-
tions are compared for FANETs. It was shown that Greedy
Perimeter Stateless Routing (GPSR) [74], which is a posi-
tion-based protocol, outperformed proactive and reactive
routing solutions. Shirani et al. developed a simulation
framework to study the position-based routing protocols
for FANETs [75]. It was stated that greedy geographic for-
warding based routing protocols can be used for densely
deployed FANETs. However, the reliability can be a serious
problem in case of sparse deployments. A combination of
other methods, like face routing, should be used for the
applications that require 100% reliability.
Although the first FANET implementations have used
the existing MANET routing strategies, most of the MANET
routing algorithms are not ideal for FANETs, because of the
UAV specific issues such as rapid changes in the link qual-
ity and very high node mobility [32]. Therefore, FANET spe-
cific routing solutions are developed in recent years.
Alshbatat et al. proposed a novel FANET routing proto-
col with directional antenna called Directional Optimized
Link State Routing Protocol (DOLSR) [76]. This protocol is
based on the well-known Optimized Link State Routing
Protocol (OLSR) [77]. One of the most important factors
that affect the OLSR performance is to choose multipoint
relay (MPR) nodes. The sender node chooses a set of MPR
nodes so that the MPR nodes can cover two hop neighbors.
Through the use of MPRs, the message overheads can be
reduced, and the latency can be minimized. One of the
most decisive design parameters for OLSR is the number
of MPRs, which affects the delay dramatically. Simulation
studies showed that DOLSR can reduce the number of
MPRs with directional antennas, and it results in lower
end-to-end latency, which is an important design issue
for FANETs.
Time-slotted on-demand routing protocol is proposed
in [78] for FANETs. It is basically time-slotted version of
Ad-hoc On-demand Distance Vector Routing (AODV) [79].
While AODV sends its control packets on random-access
mode, time-slotted on-demand protocol uses dedicated
time slots in which only one node can send data packet.
Although it reduces the usable network bandwidth, it mit-
igates the packet collisions and increases the packet deliv-
ery ratio.
Geographic Position Mobility Oriented Routing
(GPMOR) was proposed for FANETs in [80]. The traditional
position-based solutions only rely on the location informa-
tion of the nodes. However, GPMOR predicts the move-
ment of UAVs with Gaussian-Markov mobility model, and
it uses this information to determine the next hop. It is re-
ported that this approach can provide effective data for-
warding in terms of latency and packet delivery ratio
compared to the existing position-based MANET routing
protocols.
Another set of routing solutions for FANETs is the hier-
archical protocols, which are developed to address the net-
work scalability problem. Here, the network consists of a
number of clusters in different mission areas. Each cluster
has a cluster head (CH), and all the nodes in a cluster are
within the direct transmission range of the CH. The CH is
in connection with the upper layer UAVs or satellites di-
rectly or indirectly as they represent the whole cluster.
On the other hand, CH can also disseminate data by broad-
casting to its cluster members. This model can produce
better performance results when the mission area is large,
and the number of UAVs is higher as depicted in Fig. 4.
One of the most crucial design issues for hierarchic
routing is the cluster formation. Mobility prediction clus-
tering is a cluster formation algorithm developed for FA-
NET [81]. The high mobility structure of FANET nodes
results in frequent cluster updates, and the mobility pre-
diction clustering aims to solve this problem with the pre-
diction of the network topology updates. It predicts the
mobility structures of UAVs by the help of the dictionary
Trie structure prediction algorithm [82] and link expiration
time mobility model. It takes a weighted sum of these
models and the UAV with the highest weight among its
neighbors is selected as the CH. The simulation studies
showed that this CH selection scheme can increase the sta-
bility of the clusters and the CHs.
In [83], a clustering algorithm for UAV networking is
proposed. It first constructs the clusters on the ground,
and then updates it during the operation of the multi-
UAV system. Ground clustering planning calculates the
clustering plan, and then chooses the CHs according to
the geographical information. After the deployment of
UAVs, the cluster structure is adjusted according to the
Table 4
An overview of FANET MAC layer protocols.
MAC layer protocol Short description
Adaptive MAC protocol scheme for
UAVs (AMUAV) [57]
It sends its control packages (RTS, CTS, and ACK) with its omnidirectional antenna, and DATA package is
sent by directional antennas. It can improve throughput, end-to-end delay and bit error rate for multi-UAV
systems
Token-MAC [57] It is based on a token-based technique to update channel information and update link states. It eliminates
code collision problem with its token-based structure. It can also decrease the latency and improve the
throughput with the usage of full-duplex and MPR radio circuit
_
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270 1263

mission information. Simulation studies showed that it can
effectively increase the stability and guarantee the ability
of dynamic networking.
Data-centric routing algorithms can also be used for FA-
NETs. UAVs are regularly produced for application-specific
missions, and it is difficult to adapt the multi-UAV system
for different missions. Data-centric routing solutions can
be used in FANETs for different types of applications on
the same multi-UAV system. Publish-subscribe model is
typically used for this type of communication architecture
[84,85]. It automatically connects data producers, which
are called publishers, with data consumers, which are
called subscribers. Data-centric solutions are needed to
perform in-network data aggregation. Unlike flooding, it
only dispatches the registered data types/contents to the
subscribers. In this case, point-to-multipoint data trans-
mission can be preferred to point-to-point data transmis-
sion. Data-centric communications are decoupled in three
dimensions:
Space decoupling: Communicating parties can be
anywhere.
Time decoupling: Data can be dispatched to the sub-
scribers immediately or later.
Flow decoupling: Delivery can be performed reliably.
This model can be preferred when the system includes a
limited number of UAVs on a predetermined path plan,
which requires minimum cooperation.
4.3.1. Open research issues
Routing is one of the most challenging issues for FA-
NETs. Because of the unique FANET challenges, the existing
MANET routing solutions cannot satisfy all the FANET
requirements. The existing FANET routing solutions are
presented in Table 5.
Peer-to-peer communication is essential for collabora-
tive coordination and collision avoidance of multi-UAV
systems. However, it is also possible to use FANET to col-
lect information from the environment as in wireless sen-
sor networks, which generate different traffic pattern. All
the data are routed to a limited set of UAVs that are di-
rectly connected to an infrastructure. Developing new
routing algorithms that can support peer-to peer commu-
nication and converge cast traffic is still an open issue.
Data centric routing is a promising approach for FA-
NETs. By the help of the publish-subscribe architecture of
data centric algorithms, it can be possible to produce mul-
ti-UAV systems that can support different applications. To
the best of our knowledge, data centric FANET algorithms
are totally unexplored.
4.4. Transport layer
The success of FANET designs is closely related to the
reliability of the communication architecture, and setting
up a reliable transport mechanism is essential, especially
in a highly dynamic environment.
The main responsibilities of a FANET transport protocol
are as follows:
Reliability: Reliability has always been the primary
responsibility of transport protocols in communication
networks. Messages should be reliably delivered to
the destination node to ensure proper functionalities.
Data may be simple text/binary in which 100% reliabil-
ity is required, or it may be multimedia streams in
which low reliability is acceptable. FANET transport
protocol should support different reliability levels for
different FANET applications.
Congestion control: The typical consequences of a con-
gested network are the decrease in packet delivery ratio
and the increase in latency. If a FANET is congested, col-
laboration and collision avoidance between UAVs can-
not be performed properly. A congestion control
mechanism is necessary to achieve an efficient and reli-
able FANET design.
Flow control: Because of a fast sender or multiple send-
ers, the receiver may be overloaded. Flow control can be
a serious problem especially for heterogeneous multi-
UAV systems.
The first FANET systems were implemented based on
the existing transport protocols. Elston et al. developed a
multi-UAV system with FANET communication architec-
ture. It was operated on IP-based addressing, and the
transport layer of the system supported both TCP and
UDP transport schemes [86]. However, TCP performs
poorly in MANET environments and it is also unsuited for
FANETs [87,88]. TCP flow control functionality is based
on the framing mechanism and its window size changes
constantly. An accurate estimation of the round trip time
is a challenging issue.
Joint Architecture for Unmanned Systems (JAUS) is an
emerging standard for messaging between unmanned sys-
tems [89]. Although JAUS was firstly produced for ground
systems, as Joint Architecture for Unmanned Ground Sys-
tems, it was later generalized to all kinds of unmanned
vehicles (aerial, ground, surface-of-water and undersea
vehicles). AS5669a [90] defines data communications for
JAUS, and it enables the use of efficient transport protocols,
which have their own packet formats and semantics. In
AS5669a, JTCP/JUDP is designed on top of the TCP/UDP as
a wrapper. JAUS also suggests JSerial protocol for data-
transparent transports, which support variable length data
packets when low bandwidth serial links are employed.
NATO has also a Standardization Agreement (STANAG
4586), which defines a common transport protocol for net-
Fig. 4. Hierarchical routing in FANET.
1264 _
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270

work centric operations/warfare between nodes in a multi-
national UAV network [91]. STANAG 4586, depicted in
Fig. 5, was aimed to promote interoperability between
one or more Ground Control Stations, UAVs and C4I (Com-
mand, Control, Communication, Computer and Intelli-
gence) network, particularly in joint operational settings
[92]. Unlike JAUS, STANAG 4586 is specifically developed
for supporting UAV systems.
4.4.1. Open research issues
Contrary to the wired networks and MANETs, FANETs
are characterized by highly mobile nodes and wireless
communication links with high bit error rates. They have
frequent link outages according to the positions of UAVs
and ground stations. Reliability is a critical issue for FANET
transport layers.
FANET applications use different types of data such as
target images, acoustic signals, or video captures of a mov-
ing target. These applications require different reliability
levels. While typical data communication requires 100%
reliable transport protocol, multimedia application reli-
ability requirement is lower. On the other hand, multime-
dia data traffic has some other strict requirements on
delay, bandwidth, and jitter. Therefore, new transport layer
solutions must be developed to address the requirements
of different FANET applications. To the best of our knowl-
edge, there is no transport layer specially designed for FA-
NETs. Many aspects of FANETs, which affect the reliable
and efficient data transfer protocol, are still unexplored.
4.5. Cross-layer architectures
Although layered architectures have served well for
wired networks, they are not suitable for many wireless
communication applications [93]. Cross-layer architec-
tures are proposed to overcome the performance problems
of the wireless environment. Cross-layer design can be de-
fined as a protocol design by the violation of the layered
communication architecture [94]. There are several ways
for cross-layer architecture design. Unlike the layered de-
sign principles, the adjacent layers can be designed as a
super layer. Another cross-layer protocol is to support
interactions between non-adjacent layers. It is also possi-
ble to share protocol state information across all the layers
to meet the specific requirements [95].
A FANET cross-layer architecture is introduced in [96],
where the interaction between the first three layers of
OSI reference model is facilitated. In this study, a novel
directional antenna based MAC layer protocol, Intelligent
Medium Access Control Protocol (IMAC-UAV), is used.
Directional Optimized Link State Routing (DOLSR) protocol
[76] is the network layer of this system. Cross-layer design
is based on the information sharing between the first three
layers. It is shown that based on the aircraft attitude vari-
ations (pitch, roll and yaw); the performance of a FANET
application can be improved by the help of this cross-layer
architecture.
Huba and Shenoy have proposed meshed-tree algorithm
based on the directed antennas [97]. This solution inte-
grates clustering and scheduling for MAC layer along with
the routing strategy for the network layer. It can handle
MAC layer and network layer with a single algorithm that
can form the clusters, route the data from UAVs to the clus-
ter heads, and schedule the time slots in a TDMA based
MAC layer. This approach results in a robust and scalable
solution. Performance studies have shown that it can nota-
bly enhance packet delivery rate and end-to-end latency.
4.5.1. Open research issues
As in the other types of highly dynamic wireless net-
works, cross-layer architecture is an effective technique to
meet the strict requirements of FANET. Although there are
some studies about cross-layered FANETs, which are pre-
sented in Table 6, the area is largely open for new protocols.
By the help of the interactions between layers, it is possible
to enhance the FANET performance. Especially, link quality
status, which is related with the physical layer, can be an
important parameter for the upper layers. For example,
transport layer can update its operation mode to satisfy
the reliability requirement of the FANET application accord-
ing to the current link qualities. Another cross-layer protocol
opportunity is to combine all layers into a single protocol.
This unified cross-layer approach can help to design more
efficient FANET architectures for multi-UAV systems.
5. FANET test beds and simulators
In this section, the existing FANET test beds and simula-
tors are investigated to provide a quick guideline for new
FANET researchers.
Table 5
An overview of network layer protocols for FANETs.
Network layer related
algorithms
Routing
type
Short description
DOLSR [76] Proactive It utilizes directed antennas in OLSR [77] to enhance packet delivery ratio and to decrease average
latency
Time-slotted on-demand
routing [78]
Reactive It embeds time-slotted reservation schema into AODV [79] to eliminate collisions
GPMOR [80] Geographic It predicts the movement of UAVs with Gauss–Markov mobility model, and uses this information to
determine the next hop
Mobility prediction clustering
[81]
Hierarchical It utilizes the dictionary Trie structure prediction algorithm and link expiration time mobility model
to predict network topology updates. In this way, it can construct more stable cluster formations
Clustering algorithm of UAV
networking [83]
Hierarchical It constructs the clusters on the ground, and then updates the clusters during the operation of the
multi-UAV system
_
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270 1265

One of the first FANET test beds was implemented in
University of Colorado [32]. It was developed and realized
with IEEE 802.11b radio equipment mounted on small
UAVs with Fidelity-Comtech bidirectional amplifier up to
1 W output and a GPS. Dynamic Source Routing (DSR)
was chosen as the network protocol, and a monitoring sys-
tem was embedded into the radios for detailed perfor-
mance characterization and analysis.
Berkley Aerobot Team (BEAR) [98] is another multi-UAV
test bed that can support UAV-to-UAV communication.
BEAR research facility features a fleet of BEAR helicopter
UAVs, fixed-wing UAVs, unmanned ground robots, and a
mobile ground station. Rotorcraft-based Unmanned Aerial
Vehicles (RUAVs) in BEAR include 802.11 wireless network
cards that can be used for FANET.
Xiangyu et al. developed a new multi-UAV system
based on ad hoc networking architecture [99]. The mul-
ti-UAV system successfully validated the effectiveness
and feasibility of wireless ad hoc networking between
UAVs.
Sensing Unmanned Autonomous Aerial Vehicles (SUA-
AVE) project [26] aims to create and control a UAV swarm
with ad hoc networking between UAVs. The project is not
limited with a particular scenario, but the platform was
developed based on a search-and-rescue operation.
Although the first examples of the project were planned
with IEEE 802.11 protocol, SUAAVE can be used to develop
new communication architectures and protocols for UAV
swarms.
The UAV Research Facility (UAVRF) [100] conducts UAV
related researches at Georgia Institute of Technology. The
UAVRF operates different multi-UAV systems and conducts
flight tests to validate research findings. Christmann et al.
developed a FANET implementation with IEEE 802.11 com-
munication hardware in UAVRF [101].
The above-mentioned multi-UAV test beds are de-
signed to work in outdoor conditions. In order to create
a more controlled environment for rapid prototyping
and initial tests, there are also indoor test beds. Indoor
multi-UAV test beds are designed to test UAV perfor-
Fig. 5. UAV system interoperability architecture with STANAG 4586.
Table 6
Cross-layer FANET communication protocols.
Cross-layer protocols Short description
IMAC-UAV with
DOLSR [96]
It uses IMAC-UAV as the MAC layer and DOLSR as the network layer protocol for directed antennas. The first three layers
communicate through the shared data set. In this way, the transmission parameters can be adjusted dynamically. It reduces
end-to-end delay with respect to original OLSR network protocol
Meshed-tree
algorithm [97]
It integrates the MAC layer and the network layer in a single protocol, which can form the clusters, route the data from UAVs
to the cluster heads and schedule the time slots in a TDMA based MAC layer. It enhances packet delivery ratio and end-to-
end latency
1266 _
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270

mances in restricted and controlled large rooms. The
Aerospace Controls Laboratory at MIT utilizes a UAV test
bed facility, Real-time indoor Autonomous Vehicle test
Environment (RAVEN) [102]. RAVEN uses a motion cap-
ture system to enable rapid prototyping of aerobatic
flight controllers for helicopters and airplanes; robust
coordination algorithms for multiple helicopters; and vi-
sion-based sensing algorithms for indoor flight. General
Robotics, Automation, Sensing, and Perception (GRASP)
[103] is another indoor test bed developed in University
of Pennsylvania. It is developed to support research on
coordinated, dynamic flight of micro UAVs with broad
applications to reconnaissance, surveillance, manipulation
and transport.
Another way to investigate FANET designs is to simulate
the developed algorithms with a realistic multi-UAV sys-
tem simulator which can support ad hoc networking.
Although there are many multi-UAV simulators, most of
them do not model UAV-to-UAV communication links.
Real time multi-UAV simulator (RMUS) [104], which is de-
signed to work with IEEE 802.11, is one of the first multi-
UAV simulators that support the direct communication
links between UAVs. It is implemented as both a testing
and validation mechanism for the real demonstration of
multiple UAVs conducting decentralized data fusion and
control [105].
A Simulator and Test bed for Micro-Aerial Vehicle
Swarm Experiments (Simbeeotic) [106] is proposed as an
open source simulator in Harvard University for UAV
swarms that consist of up to thousands of mini or micro
UAVs. It can simulate the physical movements of the
UAV swarm as well as the communication architecture be-
tween UAVs. It is possible to develop algorithms and rapid
prototyping with Simbeeotic. It supports both pure simula-
tion and hardware-in-loop experimentation. Simbeeotic
can cover a complete view of the UAV swarm system,
including actuation, sensing, and communication.
A list of the existing FANET test beds and simulators is
given in Table 7.
5.1. Open research issues
Although the existing multi-UAV test beds and simula-
tors can support a certain variety of UAVs, they enable very
restricted variety of network protocols, like IEEE 802.11.
On the other hand, the existing network simulators, such
as OPNET [107] and ns-2 [108], can simulate different com-
munication protocols with different parameters. However,
they cannot readily model multi-UAV system specifica-
tions and mobility structures. Although there are several
FANET researches simulated on OPNET, it has no built-in
UAV node structure or UAV communication channel model
to simulate FANETs. ns-2, which is one of the common net-
work simulators, cannot model 3D communication, which
is an important parameter for FANET design [51].
In order to simulate new FANET designs, a multi-UAV
simulation tool that can simulate various UAV platforms
and network protocols is needed. The FANET simulator
must be able to model different UAV specifications, differ-
ent multi-UAV formations, different multi-UAV mobility
structures, along with different network protocols.
6. Conclusion
Communication is one of the most challenging design
issues for multi-UAV systems. In this paper, ad hoc net-
works between UAVs are surveyed as a separate network
family, Flying Ad-hoc Network (FANET). We formally de-
fine FANET and present several FANET application scenar-
ios. We also discuss the differences between FANET and
other ad hoc network types in terms of mobility, node den-
sity, topology change, radio propagation model, power
consumption, computational power and localization. FA-
NET design considerations are also investigated as adapt-
ability, scalability, latency, UAV platform constraints, and
bandwidth. We provide a comprehensive review of the re-
cent literature on FANETs and related issues in a layered
approach. Furthermore, we also discuss open research is-
Table 7
FANET test beds and simulators.
Project University/Institution/Lab Type Internet address
Test bed for a wireless network on small UAVs
[32]
University of Colorado, Interdisciplinary
Telecommunications Electrical and Computer
Engineering
Outdoor
test bed
http://itd.colorado.edu/
Berkley Aerobot team (BEAR) [98] University of California, Berkeley, Robotics Lab Outdoor
test bed
http://
robotics.eecs.berkeley.edu/
Multi-UAV system for verification of
autonomous formation [99]
Beihang University, School of Automation Science
and Electrical Engineering
Outdoor
test bed
http://id.buaa.edu.cn/IDO/
English/
Sensing Unmanned Autonomous Aerial Vehicles
(SUAAVE) [26]
SUAAVE consortium (UCL, Oxford, Ulster with
Engineering and Physical Sciences Research
Council)
Outdoor
test bed
http://www.suaave.org/
The UAV Research Facility (UAVRF) [100] Georgia Institute of Technology, UAV Lab Outdoor
test bed
http://controls.ae.gatech.edu/
wiki/UAV_Research_Facility
Real-time indoor Autonomous Vehicle test
ENvironment (RAVEN) [102]
MIT, Aerospace Controls Laboratory Indoor
test bed
http://acl.mit.edu/
General Robotics, Automation, Sensing, and
Perception (GRASP) Micro UAV Test Bed [103]
University of Pennsylvania, GRASP Lab Indoor
test bed
https://
www.grasp.upenn.edu/
Real time multi-UAV simulator (RMUS) [104] The University of Sydney, Australian Center for
Field Robotics
Simulator http://www.acfr.usyd.edu.au/
research/index.shtml
Simulator and Test bed for Micro-Aerial Vehicle
Swarm Experiments (Simbeeotic) [106]
Harvard School of Engineering and Applied Sciences Simulator http://
robobees.seas.harvard.edu
_
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270 1267

sues for FANETs, along with the cross-layer designs. The
existing FANET test beds and simulators are also
presented.
To the best of our knowledge, this is the first article
which surveyed flying ad hoc network as a separate ad
hoc network family. Our main motivation is to define mul-
ti-UAV ad hoc network problem, and to encourage more
researchers to work for the solutions to open research is-
sues as described in this paper.
Acknowledgments
We thank Prof. Ian F. Akyildiz for his constructive feed-
backs and suggestions. We also would like to thank Dr. Su-
zan Bayhan for her valuable comments.
References
[1] J. George, P.B. Sujit, J. Sousa, Search strategies for multiple UAV
search and destroy missions, Journal of Intelligent and Robotics
Systems 61 (2011) 355–367.
[2] Z. Sun, P. Wang, M.C. Vuran, M. Al-Rodhaan, A. Al-Dhelaan, I.F.
Akyildiz, BorderSense: border patrol through advanced wireless
sensor networks, Ad Hoc Networks 9 (3) (2011) 468–477.
[3] C. Barrado, R. Messeguer, J. López, E. Pastor, E. Santamaria, P. Royo,
Wildfire monitoring using a mixed air-ground mobile network, IEEE
Pervasive Computing 9 (4) (2010) 24–32.
[4] E.P. de Freitas, T. Heimfarth, I.F. Netto, C.E. Lino, C.E. Pereira, A.M.
Ferreira, F.R. Wagner, T. Larsson, UAV relay network to support
WSN connectivity, ICUMT, IEEE, 2010, pp. 309–314.
[5] F. Jiang, A.L. Swindlehurst, Dynamic UAV relay positioning for the
ground-to-air uplink, in: IEEE Globecom Workshops, 2010.
[6] A. Cho, J. Kim, S. Lee, C. Kee, Wind estimation and airspeed
calibration using a UAV with a single-antenna GPS receiver and
pitot tube, IEEE Transactions on Aerospace and Electronic Systems
47 (2011) 109–117.
[7] I. Maza, F. Caballero, J. Capitán, J.R. Martínez-De-Dios, A. Ollero,
Experimental results in multi-UAV coordination for disaster
management and civil security applications, Journal of Intelligent
and Robotics Systems 61 (1–4) (2011) 563–585.
[8] H. Xiang, L. Tian, Development of a low-cost agricultural remote
sensing system based on an autonomous unmanned aerial vehicle,
Biosystems Engineering 108 (2) (2011) 174–190.
[9] E. Semsch, M. Jakob, D. Pavlícek, M. Pechoucek, Autonomous UAV
Surveillance in Complex Urban Environments, in: Web Intelligence,
2009, pp. 82–85.
[10] H. Chao, Y. Cao, Y. Chen, Autopilots for small fixed-wing unmanned
air vehicles: a survey, in: International Conference on Mechatronics
and Automation, 2007 (ICMA 2007), 2007, pp. 3144–3149.
[11] B.S. Morse, C.H. Engh, M.A. Goodrich, UAV video coverage quality
maps and prioritized indexing for wilderness search and rescue, in:
Proceedings of the 5th ACM/IEEE International Conference on
Human–Robot Interaction, HRI ’10, Piscataway, NJ, USA, 2010, pp.
227–234.
[12] E. Yanmaz, C. Costanzo, C. Bettstetter, W. Elmenreich, A discrete
stochastic process for coverage analysis of autonomous UAV
networks, in: Proceedings of IEEE Globecom-WiUAV, IEEE, 2010.
[13] L. To, A. Bati, D. Hilliard, Radar cross-section measurements of small
unmanned air vehicle systems in non-cooperative field
environments, in: 3rd European Conference on Antennas and
Propagation, 2009 (EuCAP 2009), IEEE, 2009, pp. 3637–3641.
[14] E.W. Frew, T.X. Brown, Networking issues for small unmanned
aircraft systems, Journal of Intelligent and Robotics Systems 54 (1–
3) (2009) 21–37.
[15] M. Rieke, T. Foerster, A. Broering, Unmanned aerial vehicles as
mobile multi-platforms, in: The 14th AGILE International
Conference on Geographic Information Science, 18–21 April 2011,
Utrecht, Netherlands, 2011.
[16] J. Clapper, J. Young, J. Cartwright, J. Grimes, Unmanned Systems
Roadmap 2007–2032, Tech. rep., Dept. of Defense, 2007.
[17] T.X. Brown, B.M. Argrow, E.W. Frew, C. Dixon, D. Henkel, J. Elston, H.
Gates, Experiments Using Small Unmanned Aircraft to Augment a
Mobile Ad Hoc Network (2007) 123–145. ISBN-13: 9780521895842
(Chapter 28).
[18] J. Elston, E.W. Frew, D. Lawrence, P. Gray, B. Argrow, Net-centric
communication and control for a heterogeneous unmanned aircraft
system, Journal of Intelligent and Robotic Systems 56 (1–2) (2009)
199–232.
[19] P. Olsson, J. Kvarnström, P. Doherty, O. Burdakov, K. Holmberg,
Generating UAV communication networks for monitoring and
surveillance, in: Proceeding of the 11th International Conference
on Control, Automation, Robotics and Vision (ICARCV), Singapore,
2010.
[20] T. Samad, J.S. Bay, D. Godbole, Network-centric systems for military
operations in urban terrain: the role of UAVs, Proceedings of the
IEEE 95 (1) (2007) 92–107.
[21] A. Bürkle, F. Segor, M. Kollmann, Towards autonomous micro UAV
swarms, Journal of Intelligent and Robotics Systems 61 (1–4)
(2011) 339–353.
[22] S. Chaumette, R. Laplace, C. Mazel, R. Mirault, A. Dunand, Y.
Lecoutre, J.-N. Perbet, CARUS, an operational retasking application
for a swarm of autonomous UAVs: first return on experience, in:
Military Communication Conference – MILCOM 2011, 2011, pp.
2003–2010.
[23] Y. Ben-Asher, S. Feldman, P. Gurfil, M. Feldman, Distributed
decision and control for cooperative UAVs using ad hoc
communication, IEEE Transactions on Control Systems Technology
16 (3) (2008) 511–516.
[24] A. Alshbatat, Q. Alsafasfeh, Cooperative decision making using a
collection of autonomous quadrotor unmanned aerial vehicle
interconnected by a wireles communication network, in: Proc. of
2nd Word Conference on Information Technology, WCIT-2011,
2011.
[25] M. Quaritsch, K. Kruggl, D. Wischounig-Strucl, S. Bhattacharya, M.
Shah, B. Rinner, Networked UAVs as aerial sensor network for
disaster management applications, Elektrotechnik und
Informationstechnik 127 (3) (2010) 56–63.
[26] S. Cameron, S. Hailes, S. Julier, S. McClean, G. Parr, N. Trigoni, M.
Ahmed, G. McPhillips, R. de Nardi, J. Nie, A. Symington, L. Teacy, S.
Waharte, SUAAVE: Combining aerial robots and wireless
networking, in: 25th Bristol International UAV Systems
Conference, 2010.
[27] F. Morbidi, C. Ray, G.L. Mariottini, Cooperative active target tracking
for heterogeneous robots with application to gait monitoring, in:
IEEE/RSJ International Conference on Intelligent Robots and
Systems (IROS), 2011, pp. 3608–3613.
[28] A. Purohit, P. Zhang, SensorFly: a controlled-mobile aerial sensor
network, in: Proceedings of the 7th ACM Conference on Embedded
Networked Sensor Systems, SenSys ’09, ACM, New York, NY, USA,
2009, pp. 327–328.
[29] M.I. Akbas, D. Turgut, APAWSAN: actor positioning for aerial
wireless sensor and actor networks, in: Proceedings of the 2011
IEEE 36th Conference on Local Computer Networks, LCN ’11, IEEE
Computer Society, Washington, DC, USA, 2011, pp. 563–570.
[30] J. Allred, A. Hasan, S. Panichsakul, W. Pisano, P. Gray, J. Huang, R.
Han, D. Lawrence, K. Mohseni, Sensorflock: an airborne wireless
sensor network of micro-air vehicles, in: Proceedings of the 5th
International Conference on Embedded Networked Sensor Systems,
ACM, 2007, pp. 117–129.
[31] J. Yick, B. Mukherjee, D. Ghosal, Wireless sensor network survey,
Computer Networks 52 (12) (2008) 2292–2330.
[32] T.X. Brown, S. Doshi, S. Jadhav, J. Himmelstein, Test bed for a
wireless network on small UAVs, in: Proc. AIAA 3rd ‘‘Unmanned
Unlimited’’ Technical Conference, 2004, pp. 20–23.
[33] Z. Han, A.L. Swindlehurst, K.J.R. Liu, Optimization of MANET
connectivity via smart deployment/movement of unmanned air
vehicle, IEEE Transactions on Vehicular Technology 58 (2009)
3533–3546.
[34] T.X. Brown, B. Argrow, C. Dixon, S. Doshi, R.-G. Thekkekunnel, D.
Henkel, Ad Hoc UAV ground network (AUGNet), in: USENIX
Technical Conference, 2004.
[35] W. Wang, X. Guan, B. Wang, Y. Wang, A novel mobility model based
on semi-random circular movement in mobile ad hoc networks,
Information Science 180 (3) (2010) 399–413.
[36] E. Kuiper, S. Nadjm-Tehrani, Mobility models for UAV group
reconnaissance applications, in: Proceedings of International
Conference on Wireless and Mobile Communications, IEEE
Computer Society, 2006, p. 33.
[37] B. Anderson, B. Fidan, C. Yu, D. Walle, UAV formation control:
theory and application, in: V. Blondel, S. Boyd, H. Kimura (Eds.),
Recent Advances in Learning and Control, Lecture Notes in Control
and Information Sciences, Vol. 371, Springer, Berlin/Heidelberg,
2008, pp. 15–33.
1268 _
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270

[38] E. Yanmaz, R. Kuschnig, C. Bettstetter, Channel measurements over
802.11a-based UAV-to-ground links, in: GLOBECOM Wi-UAV
Workshop, 2011, pp. 1280–1284.
[39] A. Purohit, F. Mokaya, P. Zhang, Collaborative indoor sensing with
the sensorfly aerial sensor network, in: Proceedings of the 11th
International Conference on Information Processing in Sensor
Networks, IPSN, ACM, New York, NY, USA, 2012, pp. 145–146.
[40] S. Misra, S. Bhardwaj, Secure and robust localization in a wireless ad
hoc environment, IEEE Transactions on Vehicular Technology 58
(2009) 1480–1489.
[41] J. Wang, R. Ghosh, S. Das, A survey on sensor localization, Journal of
Control Theory and Applications 8 (2010) 2–11.
[42] H.-S. Ahn, C.-H. Won, DGPS/IMU integration-based geolocation
system: airborne experimental test results, Aerospace Science and
Technology 13 (2009) 316–324.
[43] A.K. Wong, T.K. Woo, A.T.-L. Lee, X. Xiao, V.W.-H. Luk, K.W. Cheng,
An AGPS-based elderly tracking system, in: International
Conference on Ubiquitous and Future Networks, 2009.
[44] D. Jung, P. Tsiotras, Inertial attitude and position reference system
development for a small UAV, in: Proc. of 26th AIAA Aeroacoustics
Conference, 2007.
[45] G. Mao, S. Drake, B.D.O. Anderson, Design of an extended Kalman
filter for UAV localization, in: Information, Decision and Control,
2007.
[46] J. Baillieul, P.J. Antsaklis, Control and communication challenges in
networked real-time systems, Proceedings of the IEEE 95 (2007) 9–
28.
[47] J. Li, Y. Zhou, L. Lamont, Packet delay in networked multi-UAV
systems, in: Proc. of the 26th International UAV Systems
Conference, 2011.
[48] H. Zhai, Y. Kwon, Y. Fang, Performance analysis of IEEE 802.11 MAC
protocols in wireless LANs: research articles, Wireless
Communications and Mobile Computing 4 (8) (2004) 917–931.
[49] T. Rappaport, Wireless Communications: Principles and Practice,
second ed., Prentice Hall PTR, Upper Saddle River, NJ, USA, 2001.
[50] D. Hague, H.T. Kung, B. Suter, Field experimentation of COTS-based
UAV networking, in: Proceedings of the 2006 IEEE Conference on
Military Communications, MILCOM’06, IEEE Press, Piscataway, NJ,
USA, 2006, pp. 1942–1948.
[51] N. Ahmed, S. Kanhere, S. Jha, Link characterization for aerial
wireless sensor networks, in: GLOBECOM Wi-UAV Workshop,
2011, pp. 1274–1279.
[52] Y. Zhou, J. Li, L. Lamont, C. Rabbath, Modeling of packet dropout for
UAV wireless communications, in: International Conference on
Computing, Networking and Communications (ICNC), IEEE, 2012,
pp. 677–682.
[53] I.Y. Abualhaol, M.M. Matalgah, Outage probability analysis in a
cooperative UAVs network over nakagami-m fading channels, in:
IEEE Conference on Vehicular Technology, 2006, pp. 1–4.
[54] V. Taliwal, D. Jiang, H. Mangold, C. Chen, R. Sengupta, Empirical
determination of channel characteristics for DSRC vehicle-to-
vehicle communication, in: Vehicular Ad Hoc Networks, 2004, p.
88.
[55] J. Yin, G. Holl, T. Elbatt, F. Bai, H. Krishnan, DSRC channel fading
analysis from empirical measurement, in: Proceedings of the 1st
IEEE International Workshop on Vehicle Communications and
Applications (Vehiclecomm), 2006, pp. 25–27.
[56] I.Y. Abualhaol, M.M. Matalgah, Performance analysis of cooperative
multi-carrier relay-based UAV networks over generalized fading
channels, International Journal of Communication Systems 24 (8)
(2011) 1049–1064.
[57] H.T. Kung, C.-K. Lin, T.-H. Lin, S.J. Tarsa, D. Vlah, Measuring diversity
on a low-altitude UAV in a ground-to-air wireless 802.11 mesh
network, in: IEEE Globecom Workshops, 2010.
[58] J.I. Choi, M. Jain, K. Srinivasan, P. Levis, S. Katti, Achieving single
channel, full duplex wireless communication, in: Proceedings of the
Sixteenth Annual International Conference on Mobile Computing
and Networking, MobiCom ’10, ACM, New York, NY, USA, 2010, pp.
1–12.
[59] R. Ramanathan, On the performance of ad hoc networks with
beamforming antennas, in: Proceedings of the 2nd ACM
International Symposium on Mobile Ad Hoc Networking &
Computing, MobiHoc ’01, ACM, New York, NY, USA, 2001, pp. 95–
105.
[60] O. Bazan, M. Jaseemuddin, On the design of opportunistic MAC
protocols for multihop wireless networks with beamforming
antennas, IEEE Transactions on Mobile Computing 10 (3) (2011)
305–319.
[61] Z. Huang, C.-C. Shen, A comparison study of omnidirectional and
directional MAC protocols for ad hoc networks, in: Global
Telecommunications Conference, GLOBECOM, IEEE, 2002.
[62] G. Noubir, On connectivity in ad hoc networks under jamming
using directional antennas and mobility, in: Wired/Wireless
Internet Communications, Lecture Notes in Computer Science, vol.
2957, Springer, Berlin/Heidelberg, 2004, pp. 521–532.
[63] P. Chatzimisios, A.C. Boucouvalas, V. Vitsas, Effectiveness of RTS/
CTS handshake in IEEE 802.11 a wireless LANs, Electronics Letters
40 (14) (2004) 915–916.
[64] H. Chen, F. Yu, H.C.B. Chan, V.C.M. Leung, A novel multiple access
scheme in wireless multimedia networks with multi-packet
reception, in: Proceedings of the 1st ACM Workshop on Wireless
Multimedia Networking and Performance Modeling, WMuNeP ’05,
ACM, New York, NY, USA, 2005, pp. 24–31.
[65] Bazan, Osama, Jaseemuddin, Muhammad, A survey on MAC
protocols for wireless adhoc networks with beamforming antennas,
IEEE Communications Surveys Tutorials 14 (2) (2012) 216–239.
[66] A.I. Alshbatat, L. Dong, Adaptive MAC protocol for UAV
communication networks using directional antennas, in:
Proceedings of International Conference on Networking, Sensing
and Control (ICNSC), 2010, pp. 598–603.
[67] Y. Cai, F. Yu, J. Li, Y. Zhou, L. Lamont, MAC performance
improvement in UAV ad-hoc networks with full-duplex radios
and multi-packet reception capability, in: Proc. of IEEE
International Conference on Communications (ICC), Ottawa,
Canada, 2012, pp. 523-527.
[68] B.R. Bellur, M.G. Lewis, F.L. Templin, An ad hoc network for teams of
autonomous vehicles, in: Proc. First Annual Symposium on
Autonomous Intelligence Networks and Systems, AINS
Symposium, 2002.
[69] B. Bellur, R.G. Ogier, A reliable, efficient topology broadcast protocol
for dynamic networks, Proceedings of Eighteenth Annual Joint
Conference of the IEEE Computer and Communications Societies,
INFOCOM ’99, vol. 1, IEEE, 1999, pp. 178–186.
[70] T. Brown, S. Doshi, S. Jadhav, D. Henkel, R. Thekkekunnel, A full
scale wireless ad hoc network test bed, in: Proc. of International
Symposium on Advanced Radio Technologies, Boulder, CO, 2005,
pp. 50–60.
[71] D.B. Johnson, D.A. Maltz, Dynamic source routing in ad hoc wireless
networks, in: T. Imielinski, H.F. Korth (Eds.), Mobile Computing,
The Kluwer International Series in Engineering and Computer
Science, vol. 353, Springer, US, 1996, pp. 153–181.
[72] V.R. Khare, F.Z. Wang, S. Wu, Y. Deng, C. Thompson, Ad-hoc network
of unmanned aerial vehicle swarms for search & destroy tasks, in:
Intelligent Systems, IS, International IEEE Conference, 2008.
[73] M.T. Hyland, B.E. Mullins, R.O. Baldwin, M.A. Temple, Simulation-
based performance evaluation of mobile ad hoc routing protocols in
a swarm of unmanned aerial vehicles, in: Proceedings of the 21st
International Conference on Advanced Information Networking and
Applications Workshops – Vol. 02, AINAW ’07, IEEE Computer
Society, Washington, DC, USA, 2007, pp. 249–256.
[74] B. Karp, H.T. Kung, GPSR: greedy perimeter stateless routing for
wireless networks, in: Proceedings of the 6th Annual International
Conference on Mobile Computing and Networking, MobiCom ’00,
ACM, New York, NY, USA, 2000, pp. 243–254.
[75] R. Shirani, M. St-Hilaire, T. Kunz, Y. Zhou, J. Li, L. Lamont, The
performance of greedy geographic forwarding in unmanned
aeronautical ad-hoc networks, in: Proceedings of the 2011 Ninth
Annual Communication Networks and Services Research
Conference, CNSR ’11, IEEE Computer Society, Washington, DC,
USA, 2011, pp. 161–166.
[76] A.I. Alshabtat, L. Dong, J. Li, F. Yang, Low latency routing algorithm
for unmanned aerial vehicles ad-hoc networks, International
Journal of Electrical and Computer Engineering 6 (1) (2010) 48–54.
[77] T. Clausen, P. Jacquet, Optimized Link State Routing Protocol
(OLSR), RFC 3626 (Experimental), October 2003.
[78] J.H. Forsmann, R.E. Hiromoto, J. Svoboda, A time-slotted on-demand
routing protocol for mobile ad hoc unmanned vehicle systems, SPIE
6561 (2007) 65611P.
[79] C.E. Perkins, E.M. Royer, Ad-hoc on-demand distance vector
routing, IEEE Workshop on Mobile Computing Systems and
Applications (1999) 90.
[80] L. Lin, Q. Sun, J. Li, F. Yang, A novel geographic position mobility
oriented routing strategy for UAVs, Journal of Computational
Information Systems 8 (2012) 709–716.
[81] C. Zang, S. Zang, Mobility prediction clustering algorithm for UAV
networking, in: GLOBECOM Workshops, IEEE, 2011, pp. 1158–1161.
_
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270 1269

[82] C. Konstantopoulos, D. Gavalas, G. Pantziou, A mobility aware
technique for clustering on mobile ad-hoc networks, in:
Proceedings of the 8th International Conference on Distributed
Computing and Networking, ICDCN’06, Springer-Verlag, Berlin,
Heidelberg, 2006, pp. 397–408.
[83] L. Kesheng, Z. Jun, Z. Tao, The clustering algorithm of UAV
networking in near-space, in: 8th International Symposium on
Antennas, Propagation and EM Theory, 2008 (ISAPE 2008), 2008,
pp. 1550–1553.
[84] J. Ko, A. Mahajan, R. Sengupta, A network-centric UAV organization
for search and pursuit operations, Aerospace Conference
Proceedings, 2002, vol. 6, IEEE, 2002, pp. 2697–2713.
[85] J. López, P. Royo, E. Pastor, C. Barrado, E. Santamaria, A middleware
architecture for unmanned aircraft avionics, in: Proceedings of the
2007 ACM/IFIP/USENIX International Conference on Middleware
Companion, MC ’07, ACM, New York, NY, USA, 2007, pp. 24:1–24:6.
[86] J. Elston, B. Argrow, A. Houston, J. Lahowetz, Distributed
atmospheric sensing using small UAS and doppler radar, in: AIAA
Infotech@Aerospace Conference, Seattle, WA, 2009.
[87] Z. Fu, H. Luo, P. Zerfos, S. Lu, L. Zhang, M. Gerla, The impact of
multihop wireless channel on TCP performance, IEEE Transactions
on Mobile Computing 4 (2) (2005) 209–221.
[88] C.S.R. Murthy, B. Manoj, Ad Hoc Wireless Networks: Architectures
and Protocols, Prentice Hall PTR, Upper Saddle River, NJ, USA, 2004.
[89] R. Wade, Joint architecture for unmanned systems, in: Proc. of 44th
Annual Targets, UAVs & Range Operations Symposium & Exhibition,
2006.
[90] A. 4b Network Environmental Committee, JAUS/SDP Transport
Specification, <http://standards.sae.org/as5669a/> (last visited on
05.01.12).
[91] NATO Standardization Agreement 4586, <http://www.nato.int/
docu/pr/2002/p02-106e.htm> (last visited on 05.01.12).
[92] M. Cummings, A. Kirschbaum, A. Sulmistras, J. Platts, STANAG 4586
human supervisory control implications, in: Unmanned Vehicle
Systems Canada Conference, 2006.
[93] V. Srivastava, M. Motani, Cross-layer design: a survey and the road
ahead, IEEE Communications Magazine 43 (12) (2005) 112–119.
[94] F. Foukalas, V. Gazis, N. Alonistioti, Cross-layer design proposals for
wireless mobile networks: a survey and taxonomy, IEEE
Communications Surveys Tutorials 10 (1) (2008) 70–85.
[95] B. Jarupan, E. Ekici, A survey of cross-layer design for VANETs, Ad
Hoc Networks 9 (5) (2011) 966–983.
[96] A.I. Alshbatat, L. Dong, Cross-layer design for mobile ad-hoc
unmanned aerial vehicle communication networks, in: ICNSC,
2010, pp. 331–336.
[97] W. Huba, N. Shenoy, Airborne surveillance networks with
directional antennas, in: ICNS 2012, The Eighth International
Conference on Networking and Services, 2012, pp. 1–7.
[98] B. Robotics, I.M. Lab, BEAR: Berkeley Aerobot Team, <http://
robotics.eecs.berkeley.edu/bear/index.html> (last visited on
05.01.12).
[99] J. Xiangyu, W. Sentang, L. Xiang, D. Yang, T. Jiqiang, Research and
design on physical multi-UAV system for verification of
autonomous formation and cooperative guidance, in: Proceedings
of the 2010 International Conference on Electrical and Control
Engineering, ICECE ’10, 2010, pp. 1570–1576.
[100] E. Johnson, Georgia Tech. UAV Research Facility, <http://
controls.ae.gatech.edu/wiki/UAV_Research_Facility> (last visited
on 05.01.12).
[101] H. Christmann, Self-configuring Ad-hoc Networks for Unmanned
Aerial Systems, Master’s thesis, Georgia Institute of Technology,
April 2008.
[102] J.P. How, B. Bethke, A. Frank, D. Dale, J. Vian, Real-time indoor
autonomous vehicle test environment, IEEE Control Systems
Magazine 28 (2) (2008) 51–64.
[103] N. Michael, D. Mellinger, Q. Lindsey, V. Kumar, The GRASP multiple
micro-UAV testbed, IEEE Robotics & Automation Magazine 17 (3)
(2010) 56–65.
[104] A.H. Göktog
˘an, E. Nettleton, M. Ridley, S. Sukkarieh, Real time
multi-UAV simulator, in: International Conference on Robotics and
Automation, vol. 2, 2003, pp. 2720–2726.
[105] A.H. Göktog
˘an, S. Sukkarieh, Distributed simulation and
middleware for networked UAS, Journal of Intelligent & robotics
Systems 54 (1–3) (2009) 331–357.
[106] B. Kate, J. Waterman, K. Dantu, M. Welsh, Simbeeotic: a simulator
and testbed for micro-aerial vehicle swarm experiments, in:
Proceedings of the 11th International Conference on Information
Processing in Sensor Networks, IPSN ’12, ACM, New York, NY, USA,
2012, pp. 49–60.
[107] C.M. Durham, T.R. Andel, K.M. Hopkinson, S.H. Kurkowski,
Evaluation of an OPNET model for unmanned aerial vehicle (UAV)
networks, in: G.A. Wainer, C.A. Shaffer, R.M. McGraw, M.J. Chinni
(Eds.), SpringSim, SCS/ACM, 2009.
[108] Network Simulator—NS (Version 2), <http://www.isi.edu/nsnam/
ns/> (last visited on 05.01.12).
Ilker Bekmezci received B.Sc., M.Sc., and
Ph.D. degrees in computer engineering from
Bogazici University, Istanbul, Turkey, in 1994,
1998, 2008, respectively. He is currently an
associate professor with the Department of
Computer Engineering, Turkish Air Force
Academy, Istanbul, Turkey. His current
research interests are in wireless communi-
cations and ad hoc networks.
Ozgur Koray Sahingoz is currently an assis-
tant professor in the Department of Computer
Engineering at Turkish Air Force Academy. He
graduated from the Computer Engineering
Department of Bogazici University in 1993. He
received his M.Sc. and Ph.D. degree from
Computer Engineering Department of Istan-
bul Technical University, in 1998 and 2006,
respectively. His research interests lie in the
areas of Wireless Sensor Networks, Artificial
Intelligence, Parallel and Distributed Com-
puting, Soft Computing, Information Systems,
Intelligent Agents, Multi Agent Systems.
Samil TEMEL is a Ph.D. researcher/student in
computer engineering at Turkish Air Force
Aeronautics and Space Technologies Institute
(ASTIN). He received his M.S. degree at TUAFA
in 2008 and he holds a B.S. degree in Com-
puter Engineering from Yildiz Technical Uni-
versity, Istanbul, Turkey. His research
interests include directional MAC protocols
and wireless sensor networks.
1270 _
I. Bekmezci et al. / Ad Hoc Networks 11 (2013) 1254–1270