Proceedings of the IEEE SoutheastCon 2015, April 9 - 12, 2015 - Fort Lauderdale, Florida

  Performance Evaluation of Unmanned Aerial

 
Vehicle Ad hoc Networks
 
Robin Grodi Danda B. Rawat Chandra Bajracharya
 
Department of Electrical Engineering Department of Electrical Engineering Department of Electrical Engineering
Georgia Southern University Georgia Southern University Georgia Southern University
  Statesboro, GA 30460, USA Statesboro, GA 30460, USA Statesboro, GA 30460, USA
[email protected]. [email protected] [email protected]
 

 
Abstract—Unmanned Aerial Vehicles (UAVs) can form an ad- network regularly to maintain connectivity, thus reducing the
hoc network to communicate and cooperate with each other amount of available bandwidth within the network. Reactive
 
via wireless links. UAV ad hoc networks have the potential to protocols only require that paths be found when data needs
provide reliable and real-time services using single-hop or multi-
hop communications. However, the network topology of a UAV to be sent which means that there is little protocol overhead.
 
ad-hoc network is highly dynamic because of the fast flying While reactive protocols do require a large amount of protocol
UAVs that result in short lived communication links between overhead, knowing paths ahead of time allows for a low end-
UAVs.  In mission critical applications, small delays could result to-end delay time for sending data. Reactive routing protocols
in catastrophic outcomes. Thus it is required for UAVs to do not have this advantage. When choosing a routing protocol,
exchange their information as quickly as possible (with shortest
  delay). In this paper, we evaluate the performance of
possible a tradeoff must be made between delay time and available
UAV ad hoc networks by considering the effect of network size bandwidth.
 
on throughput, energy consumption and effective transmission In a UAV network all nodes are connected to a ground
range (total coverage range). Numerical results obtained from station via single or multi-hop communication. The goal of
simulations are presented to evaluate the performance of UAV this network is to relay information from mobile UAV nodes
ad hoc  networks. Our results show that when network size (the
number of UAVs) increases, the throughput per UAV and energy back to the stationary ground node. This type of technology
consumption
  decrease while the total network transmission range can be used in disaster situations where continuous updates
(effective coverage range) increases. are needed but rescue workers cannot reach the target area in
Index Terms—UAV ad hoc networks, aerial MANETs, wireless a timely manner (due to debris or other obstructions). UAVs
ad hoc  networks, performance evaluation. would be able to fly over the target area and relay information
(pictures, videos, etc.) back to the rescue workers to keep them
  I. I NTRODUCTION updated. For a UAV network to be useful in the described
Unmanned Aerial Vehicles (UAVs) are emerging for dif- situation, it must maintain both connection and throughput
ferent  applications in many fields [1]–[4]. UAVs can be between individual UAVs within the network.
used for everything from reconnaissance in military uses to A typical ad hoc routing protocol works well for 2-
 
photography in the civilian world. These UAVs are equipped dimensional stationary networks. A new routing protocol was
with a micro-controller for processing inputs and commands proposed in [8] that would work well in a 3-dimensional het-
  be externally controlled via remote control if desired.
and can erogeneous network where communication is needed between
UAVs can also be equipped with wireless transceivers to allow highly mobile aerial and ground nodes. The authors consider
for communication
  with other UAVs or with units on the hierarchical clustering where similar nodes are grouped into
ground. If multiple UAVs are communicating with each other, clusters. These clusters are then grouped into levels. Within
they can
  form an ad hoc network. An ad hoc network allows these levels, a cluster head is appointed to facilitate communi-
for data to be collected at one node within the network and sent cation between levels. This organization of the network allows
to any  other node within the network [5], [6]. Data is routed for reduced bandwidth as only the cluster heads need to worry
through the network and the path taken depends on the routing about communicating with other clusters.
protocol used. In [7] the pros and cons of different routing Besides for routing, throughput is also extremely important
 
protocols are discussed. They consider two kinds of routing in UAV networks. The authors of [9] examine the effects
protocols: proactive and reactive. Proactive routing protocols of lossy links on throughput and energy consumption within
  that each UAV within the network knows the shortest
require a network. The paper mentions that lossy links can have a
path to any other UAV within the network. This data is stored huge impact on throughput - in some cases reducing network
into a  table and is updated frequently. By having to update throughput by up to half. The effects of lossy links on energy
shortest paths within a network periodically, proactive routing consumption was seen as less significant as compared to
  require a large amount of protocol overhead. This
protocols throughput. In [10], a Load-Carry-and-Deliver (LCAD) model
means that a large amount of data is needed to be sent by the is created for use in UAV networks where delay-tolerant bulk
 

7300-5/15/$31.00 ©2015 IEEE

978-1-4673-7300-5/15/$31.00
978-1-4673-7300-5/15/$31.00 ©2015 IEEE
©2015 IEEE
 Proceedings of the IEEE SoutheastCon 2015, April 9 - 12, 2015 - Fort Lauderdale, Florida

 
data needs to be sent. An analysis is done to show how the
LCAD  model performs as compared to multi-hop transfer of
data. Similarly, an adaptive medium access control protocol
can be  used if the UAVs are equipped with two directional
antennas and two omnidirectional antennas [11]. UAVs us-
  type of protocol can receive information using the
ing this
omnidirectional antenna while being able to simultaneously
  information from either directional antenna. An in-
transmit UAV
Link to ground station
depth throughput analysis for 802.11 multi-hop networks with
Object of Interest Radio range of a UAV Ground station
regards
  to hidden nodes is done in [12]. Optimum node density
is analyzed in [13] for mobile ad hoc networks relating to
Fig. 1. A sample network of 7 UAVs and a ground station where UAVs sense
transmission
  power. It was seen that there is no optimal node the target region to find a object of interest while maintaining single-hop and
density for ad hoc networks. Instead, it was noted that as node multi-hop communication link with the ground station [6].
mobility increases, the number of nodes should also increase
 
to achieve the highest throughput.
time T . We consider Pf to be the probability of failure. The
The goal of this paper is to analyze the effect of network
  goodput differs from the throughput as throughput considers
size on throughput, energy consumption and transmission
all data, not just the useful information. Goodput is a better
range. While throughput and energy consumption have been
  in the referenced papers, none of them have taken measurement as it does not consider the network overhead.
studied
Eq.(1) shows the total network goodput. If we want to find
into account how the size of the network would affect the
  parameters. Simulations are done to show a graphical the goodput per UAV we need to take into account all of the
various
UAVs in the network. In this type of network, we consider all
representation of the results.
users to be equal and share bandwidth evenly. Eq.(2) shows
The  remainder of the paper is organized as follows: A
the goodput per UAV by dividing the total network goodput
system model is presented in Section II which describes what
by the number of UAVs. Nuav represents the number of UAVs
 
our analysis will be based off of. Section III consists of a
using the same channel within the network.
performance analysis. Simulation results are shown in Section
IV and  we conclude the paper in Section V. G
Gp = [bps] (2)
Nuav
II. S YSTEM M ODEL AND P ROBLEM F ORMULATION
  Each UAV in the network will need to use a certain amount
In this paper, we consider a network of UAVs equipped
of energy to achieve the specified goodput. We represent that
with cameras/sensors
  for sensing, GPS unit, and with wire-
energy as E. We can create an equation to represent the
less device (a transceiver with omnidirectional antenna) for
amount of goodput achievable per unit energy by dividing
networking and transferring information to another UAV or a
  goodput per UAV, Gp , by the needed energy E.
ground station using single-hop or multi-hop communication
as shown in Fig. 1. We have considered a disaster situation Gp
where  rescue workers are not able to reach a target area in a Ge = [bps/J] (3)
E
timely manner. This network of UAVs are able to configure
 
themselves in such a way that they create a temporary link If we look back to our example of using a UAV network
between a ground station and the target area. This temporary in a disaster situation, it is important to know the maximum
 
link provides a crucial data link to rescue workers as they possible transmission range for the network. This is achievable
approach the target area. when each UAV is put into a straight line. Each individual
In  this paper, using this system model, our goal is to UAV in the network will have a max transmission range. That
look into how throughput is affected by network size, energy transmission range might vary from UAV to UAV. The trans-
consumption
  and transmission range. mission range between any two UAVs needs to be the smaller
transmission range achievable between those two UAVs. From
III. P ERFORMANCE A NALYSIS Fig.2, we can see that there are three UAVs in the network.
 
If we consider goodput to be the total amount of useful data To get our transmission range of the network we must add the
transmitted from one node to another over a certain period transmission range of UAV 1, T1 , to the smaller of the two
 
of time, the goodput can be calculated using the following transmission ranges between UAV 1 and UAV 2. We add that
equation: sum to the smaller of the two transmission ranges between
  UAV 2 and UAV 3 and then add it to the transmission range
(d − h) × Np of UAV 3. This sum gives us our total network transmission
G= (1 − pf ) [bps] (1)
  T range, which we will define as Tr . Eq.(4) shows how we
This equation takes the useful information (d data bits) and can define this mathematically. We must add the first and last
  out the h header bits. This is then multiplied by
subtract transmission ranges, T1 and Tn respectively, to the sum of all
Np number of packets and divided by the total transmission the intermediate transmission ranges, Ai . The equation for Ai
 

7300-5/15/$31.00978-1-4673-7300-5/15/$31.00
©2015 IEEE ©2015 IEEE
 Proceedings of the IEEE SoutheastCon 2015, April 9 - 12, 2015 - Fort Lauderdale, Florida

 
can be seen in (5) and will sum for all UAVs in the network,
NU AV  .
800

NXU AV
700
  Tr = T1 + Tn + Ai (4)
i=1 600

Transmission Range (meters)

where  the value of Ai is computed as
500

Ai = min{Ti , Ti+1 } (5)

  400

300
 
200

  T1 min{T2,T3}
100
UAV 1 UAV 3 1 2 3 4 5 6 7 8 9 10
UAV 2 Number of UAVs

  min{T1,T2} T3 Fig. 4. Transmission Range vs. Number of UAVs in the network

  14

Fig. 2.   A network of 3 UAVs showing transmission range and distance

12

between UAVs where distance between UAVs can be computed using GPS 10
locations. Note that the distance between them depends on their relative
 

Propogation Delay (microseconds)

velocity and direction of travel [6].
8

  P ERFORMANCE E VALUATION AND N UMERICAL

IV.
6
R ESULTS
  eq.(1), we can simulate a one megabit data packet
Using 4

being sent over various time periods. We use timer periods

  from one second to sixty seconds at one second
ranging 2

intervals. Fig. 3 shows the needed data rate to send that one
  data packet over the specified time period. From the
megabit 0
1 2 3 4 5 6
Number of UAVs
7 8 9 10

graph, it can be seen that as the amount of available time

Fig. 5. Propogation Delay vs. Number of UAVs in the network
increases,
  the needed datarate decreases.
1

  6
10
0.9
Amount of data sent successfully (megabits)

0.8
  0.7

0.6
 
Data Rate (bits/second)

0.5
5
10
0.4
 
0.3

  0.2

0.1

  0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
4
10 probability of failure
10 20 30 40 50 60
Time (seconds) Fig. 6. Goodput vs probability of failure
  Fig. 3. Datarate needed versus total transmission time

Using eq.(4) we simulated the maximum network trans- the number of UAVs in the network increases, the maximum
 
mission range versus the number of UAVs in the network. possible transmission range increases as well.
This simulation considers that the UAVs are all at the same Considering the same model as the transmission range
  and that the maximum transmission range between
altitude graph, we also plot the affect on propagation delay. Fig.
any two UAVs is the shortest Fig. 4 shows max network 5 plots propagation delay as a function of the number of
 
transmission range as a function of the number of UAVs in the UAVs in the network. We must consider that as you increase
network. We consider the furthest possible transmission range the number of UAVs in the network, the propagation delay
 
by simulating the UAVs in a straight line at maximum range increases exponentially as each new UAV adds a new point of
from each other. The result is a linear graph showing that as possible failure.
 

7300-5/15/$31.00978-1-4673-7300-5/15/$31.00
©2015 IEEE ©2015 IEEE
 Proceedings of the IEEE SoutheastCon 2015, April 9 - 12, 2015 - Fort Lauderdale, Florida

 
6
unit energy for each UAV in the network. If you increase the
  10
number of UAVs in the network, the amount of throughput
available to each UAV decreases, which in turn decreases
  goodput per unit energy for each UAV. From Fig.7 we can see
1 UAV
Goodput per unit energy (bps/J)

5
10
10 UAV that the goodput per unit energy for 10 UAVs in the network
  is lower than when only one UAV is in the network.
The simulations performed in this paper can be used to
  analyze and to add a more detailed explanation of what factors
4
10 most affect a UAV network.
  Our future research includes showing the effect of different
routing protocols (proactive vs reactive) on the throughput
  along with the number of UAVs and transmission ranges of
3
10
10 20 30 40 50 60 UAVs.
Time (seconds)

  Goodput per unit energy vs the number of UAVs in the network

Fig. 7. ACKNOWLEDGMENT
This work was supported in part by the National Science
As   mentioned, as more UAVs enter the network, there Foundation (NSF) under grant CNS-1405670. Any opinion,
are more points of possible failure. Fig. 6 shows a one finding, and conclusions or recommendations expressed in this
  packet being sent through a UAV network with
megabit material are those of the authors and do not necessarily reflect
varying probabilities of failure. Adding additional UAVs to the views of NSF.
the network
  increases the amount of possible points of failure.
This probability of failure can range anywhere from 0% to R EFERENCES
100%.  It can be seen that at 100% probability of failure, no [1] R. L. Finn and D. Wright, “Unmanned aircraft systems: Surveillance,
data is transmitted and at 0% probability of failure, all of the ethics and privacy in civil applications,” Computer Law & Security
Review, vol. 28, no. 2, pp. 184–194, 2012.
data is transmitted. [2] D. M. Marshall, R. K. Barnhart, S. B. Hottman, E. Shappee, and M. T.
 
We have shown the relationship between network size, Most, Introduction to unmanned aircraft systems. CRC Press, 2011.
[3] R. Austin, Unmanned aircraft systems: UAVs design, development and
transmission range and propagation delay. Each UAV in the
  deployment. John Wiley & Sons, 2011, vol. 54.
network will require a certain amount of energy to send data. [4] A. C. Watts, V. G. Ambrosia, and E. A. Hinkley, “Unmanned aircraft
By summing the energy used by each UAV we know the total systems in remote sensing and scientific research: Classification and
energy  consumption of the network. We would like to know considerations of use,” Remote Sensing, vol. 4, no. 6, pp. 1671–1692,
2012.
how energy efficiency is affected by increasing the UAVs in [5] E. W. Frew and T. X. Brown, “Networking issues for small unmanned
 
the network. Fig. 7 shows goodput per unit energy. It can be aircraft systems,” Journal of Intelligent and Robotic Systems, vol. 54,
no. 1-3, pp. 21–37, 2009.
seen that as the number of UAVs in the network increases, [6] D. B. Rawat, R. Grodi, and C. Bajracharya, “Enhancing connectivity
 
the goodput per unit energy decreases exponentially. This is for communication and control in unmanned aerial vehicle networks,” in
to be expected since as the number of UAVs in the network Proc. 2015 IEEE Radio and Wireless Week (IEEE RWW 2015), January
25 – 28 2015.
  the total available bandwidth decreases. With limited
increases, [7] C. Mbarushimana and A. Shahrabi, “Comparative study of reactive and
bandwidth to send data, the amount of energy needed will also proactive routing protocols performance in mobile ad hoc networks,”
decrease.
  in 21st International Conference on Advanced Information Networking
and Applications Workshops 2007, vol. 2, 2007, pp. 679–684.
[8] D. L. Gu, G. Pei, H. Ly, M. Gerla, B. Zhang, and X. Hong, “Uav aided
V. CONCLUSION
  intelligent routing for ad-hoc wireless network in single-area theater,”
in IEEE Wireless Communications and Networking Confernce 2000
We have performed an analysis for throughput of UAV ad (WCNC 2000), vol. 3, 2000, pp. 1220–1225.
hoc network
  with regards to network size. Numerical results [9] Y. Li, J. Harms, and R. Holte, “Impact of lossy links on performance
obtained from simulations have shown that the performance of multihop wireless networks,” in 14th International Conference on
Computer Communications and Networks, 2005, 2005, pp. 303–308.
of the  network depends on number of UAVs in the networks, [10] C.-M. Cheng, P.-H. Hsiao, H. Kung, and D. Vlah, “Maximizing through-
transmission range of UAVs and energy consumptions. We put of uav-relaying networks with the load-carry-and-deliver paradigm,”
have shown that as the amount of available transmission time in IEEE Wireless Communications and Networking Conference 2007
  (WCNC 2007). IEEE, 2007, pp. 4417–4424.
increases, the amount of datarate needed to send a specified [11] A. I. Alshbatat and L. Dong, “Performance analysis of mobile ad
amount of data decreases exponentially. With the knowledge hoc unmanned aerial vehicle communication networks with directional
  antennas,” International Journal of Aerospace Engineering, vol. 2010,
of the needed data rate for a specified time and data size, we
2011.
then looked into transmission range and propagation delay. [12] P. C. Ng and S. C. Liew, “Throughput analysis of ieee802. 11 multi-
 
We show that as we increase the number of UAVs in the hop ad hoc networks,” IEEE/ACM Transactions on Networking (TON),
network, our total transmission range increases but so does the vol. 15, no. 2, pp. 309–322, 2007.
 
propagation delay. We then considered different probabilities [13] E. M. Royer, P. M. Melliar-Smith, and L. E. Moser, “An analysis of the
optimum node density for ad hoc mobile networks,” in Communications,
of failure. It was shown that probability of failure has a 2001. ICC 2001. IEEE International Conference on, vol. 3, 2001, pp.
linear  correlation with the amount of data received. The last 857–861.
simulation shows the goodput over time of the network per
 

7300-5/15/$31.00978-1-4673-7300-5/15/$31.00
©2015 IEEE ©2015 IEEE