Study of VDTN routing protocols performances in sparse and dense traffic in the presence of relay nodes

Abstract

Vehicular Delay Tolerant Networks (VDTN) are growing challenging field of Delay Tolerant Networks
(DTN) containing mobile nodes (vehicles) that communicate using the paradigm of store carry and
forward. In this model, nodes store the bundle waiting for the opportunity to transfer it to another node in
the same transmission range, and in a limited duration. VDTN enables communications in a sparse
network characterized by low density, high mobility of nodes, intermittent connectivity, no end-to-end
communication and no information on the path of nodes, which make routing in such cases difficult and
challenging. This paper studies the performances of some well-known VDTN routing protocols in sparse
and dense traffic. The proposed scenario considers stationary nodes representing sensors measuring
different types of data to be transferred to stationary destination nodes through the VDTN network
(vehicles). The destination nodes are the only part of the network that is connected to the Internet. The
transmission of collected data is also made by fixed nodes representing stationary relays such as audio signal
traffic lights, placed in the crossroads of Rennes city streets. The map of Rennes has been used and
the position of the stoplights has been carried out using Open Data. The aim of our simulation is to study
the impact of mobile nodes and stationary relays on the routing protocols performances in dense and
sparse traffic.

Full text

Journal of Mobile Multimedia, Vol. 10, No. 1&2 (2014) 078-093
© Rinton Press
STUDY OF VDTN ROUTING PROTOCOLS PERFORMANCES IN SPARSE AND DENSE
TRAFFIC IN THE PRESENCE OF RELAY NODES
MARIA BENAMAR, SARA AHNANA, FATIMA ZAHRAE SAIYARI,
NABIL BENAMAR, MOULAY DRISS EL OUADGHIRI
MACS Laboratory, Faculty of sciences, Meknes, Morocco
{mariabenamar, sara.ahnana, saiyari.f.z, benamar73, dmelouad}@gmail.com
JEAN-MARIE BONNIN
Telecom Bretagne, Rennes, France
jm.bonnin@telecom-bretagne.eu
Vehicular Delay Tolerant Networks (VDTN) are growing challenging field of Delay Tolerant Networks
(DTN) containing mobile nodes (vehicles) that communicate using the paradigm of store carry and
forward. In this model, nodes store the bundle waiting for the opportunity to transfer it to another node in
the same transmission range, and in a limited duration. VDTN enables communications in a sparse
network characterized by low density, high mobility of nodes, intermittent connectivity, no end-to-end
communication and no information on the path of nodes, which make routing in such cases difficult and
challenging. This paper studies the performances of some well-known VDTN routing protocols in sparse
and dense traffic. The proposed scenario considers stationary nodes representing sensors measuring
different types of data to be transferred to stationary destination nodes through the VDTN network
(vehicles). The destination nodes are the only part of the network that is connected to the Internet. The
transmission of collected data is also made by fixed nodes representing stationary relays such as audio-
signal traffic lights, placed in the crossroads of Rennes city streets. The map of Rennes has been used and
the position of the stoplights has been carried out using Open Data. The aim of our simulation is to study
the impact of mobile nodes and stationary relays on the routing protocols performances in dense and
sparse traffic.
Key words: Vehicular Delay-Tolerant Networks, Routing, Relay, Performance Evaluation, ONE
1 Introduction
Vehicular delay tolerant network (VDTN) is a class of networks that extends the features of delay
tolerant network (DTN) to the vehicular environment [1]. It was the subject of many researchers in the
recent decade because of its potential applications such as safety related information, information
advertisement, multimedia sharing content, weather reports and email access[2][3]. VDTN enables
communications in a network characterized by intermittent connectivity, no end to end path between

M
. Benamar, S. Ahnana, F. Z. Saiyari, N. Benamar, M. D. El Ouadghiri, and J.-M. Bonnin 79
communicating nodes, high change of network topology, high error rate, limited duration of nodes
contact [4] for these reasons, routing in such environment is a challenging task. Indeed, routing
protocols for such networks should be designed to guarantee message transmission even when the
traffic is very low. VDTN is characterized by its paradigm of store carry and forward that motivate the
use of opportunistic routing protocols. The idea behind this concept is that, the node store message
called bundle in its buffer size till they find the opportunity to transfer it to its destination[4]. In this
paper we study the performance of some well-known VDTN routing protocols, in the case of dense
and sparse traffic. Our previous work [5] studied the suitability of VDTN routing protocols to collect
data in smart cities and focused on the impact of various parameters on the routing protocols
performances. The current work is the continuity of the previous one where we want to get closer to
real cases. For this reason, we have used for our simulations, the map of the city of Rennes and its
traffic information provided by open data. The study focuses on stationary nodes collecting data and
transferring it to other stationary nodes connected to the Internet. The transmission of data is made
through mobiles nodes that are vehicles, and stationary relays representing audio-signal traffic lights.
The main goal is the study of the impact of those parameters on the routing protocols performances.
The remainder of this paper is organized as follows. Section 2 is an introduction to DTN, its
architecture, VDTN and the routing protocols used in both DTN and VDTN. Section 3 describes the
scenario used in the current study with its different parameters. Section 4 contains the results of the
simulations and the interpretations of graphs. Section 5 concludes the paper.
2 Background
2.1 Delay Tolerant Network (DTN)
Delay tolerant network is an opportunistic network[6] which has attracted large attention in the last
few years[7]. This kind of network is characterized by intermittent connectivity, high error rates, long
and variables delays/latencies, and high node mobility [8][9]. It was designed for the first time in 1998
in partnership with Nasa’s Jet Propulsion Laboratory[10] to establish communication between
different nodes located in the solar system[11]after the failure of the traditional routing protocols used
in the internet communication because of the frequent disruptions, and the excessive propagation
delays. The project was named interplanetary networking (IPN), and then it spread to different
challenging applications as wireless sensor networking and the vehicular networking [11].
To deal with the new situation where the traditional routing protocols are not suitable
anymore, the group of research DTN (DTNRG), which has taken part of Internet Research Task Force
(IRTF) had proposed new architecture designed to DTN [12]. The aim was to support new routing
protocols, features related to DTN, and also to provide structure dealing with the heterogeneity of
different communicating regions, by the addition of a layer between the transport and the application
layer named bundle layer (figure 1). It connects the specific lower layers so that application programs
can communicate across several regions of different natures. The bundle layer also stores and relays
the bundles/messages hop by hop between nodes[13].

Study of VDTN Routing Protocols Performances in Sparse and Dense Traffic in the Presence of …
80
Figure 1: Difference between TCP/IP and DTN architecture
2.2 The Approach of Store-Carry and Forward
Contrary to the approach used in the Internet, which relies on the store and forward principle, the DTN
use the paradigm of store, carry and forward implemented in its new architecture.
This approach allows the intermediate nodes to store the bundles in their buffers for remarkable
amount of time that can be minutes, hours, or even days; waiting for the opportunity to get in touch
with the next node until the messages reach their final destinations or their time to live (TTL)
expires.[14].
2.3 Vehicular Delay Tolerant Network (VDTN)
The Vehicular Delay Tolerant Network (VDTN) is a growing field of DTN that has, recently, led to
numerous research works (figure 2). It extends the principle of VANETs, by using the characteristics
and the capabilities of DTN in order to deal with the intermittent connectivity, the high error rates, and
the high mobility of nodes[12]…etc. It is considered one of the most challenging applications of DTN.
There are many applications of VDTN, ranging from road safety improvement to the traffic
flow and road capacity optimization. For more details related to VDTN applications, one can consult
one of the comprehensive surveys in the literature [12].

M
. Benamar, S. Ahnana, F. Z. Saiyari, N. Benamar, M. D. El Ouadghiri, and J.-M. Bonnin 81
Figure 2 : Example of VDTN scenario
2.4 Routing Protocols
First Contact is a single copy DTN routing protocol, i.e. only one copy of each message exists in the
network. In this protocol, a message is transmitted along a single path by selecting the node randomly
from the available links. If the connections do not exist, nodes store data, and transmit it to the first
contact available. However, this node will not delete the message directly transferred, but it will
remove the old data when it is stored entirely[15][16].
Direct Delivery is a single copy DTN routing protocol. Its simplicity lies in the fact that a
given source node can deliver its own message. In the strategy of direct delivery, the source node
stores the message and sends it directly to the destination node; it always chooses the route between
the source and destination. If the buffer is full and the message has not been sent, it will be
dropped[15].
In the epidemic routing protocol [17], which is based on the flooding strategy, the node
replicates the message and transfer it to every node that gets in touch with and that doesn’t have the
message.Its strategy decreases the delay of delivery and increases the hop count and the overhead ratio
because of the number of resources used.

Study of VDTN Routing Protocols Performances in Sparse and Dense Traffic in the Presence of …
82
Contrary to epidemic, Spray And Wait routing protocol [18] uses n-copies to transfer the
messages. This protocol contains two phases: (1) Spray phase, where the message spread from the
source nodes to the others nodes until the n copies get exhausted, (2) Wait phase, where the nodes that
receive the message transfer it to its final destination [19]. This protocol had two versions, the normal
one, where the node gives one copy to every contacted node, and the binary one which is considered as
an improvement of this protocol and consists on transferring half of the messages to every contacting
node[20].
Furthermore, another routing protocol has gained much attention by researchers in DTN field.
This is Probabilistic Routing Protocol using History of Encounters and Transitivity routing protocol
(Prophet), which is based on the strategy of forwarding. This routing protocol calculates probabilistic
metric named delivery predictability, to estimate which node has the higher probability to be the best
node for transferring the message to its final destination with less latency[21][22].
MaxProp routing protocol tries to transfer all the messages not retained by the other nodes
when they are in the same transmission range, to guarantee successful transfer of the messages. This
protocol manages the buffer size of nodes by using the acknowledgements to clear the copy of
messages remaining in the other nodes when they are transferred to their destinations. In this protocol,
nodes exchange messages by depending on the delivery likelihood to a destination to give the priority
to a specific node between others[23]. Table 1 summarizes different routings protocols in VDTN.
Routing
Protocols Type Type of
Copy copy Estimation
based Description
Direct
Delivery Direct Single No Source moves and delivers the bundle directly
First
Contact Random Single No Use any available contact
Epidemic
(ER) Flooding Unlimited-
copy No Rapid propagation of data
Spray &
Wait Controlled
flooding n-copy No Sets a limit on the number of copies
PROPHET Probabilistic Unlimited-
copy Yes Probabilistic
MaxProp flooding Unlimited-
copy Yes use of the delivery likelihood as a cost assigned
to each destination
Table 1 Summarization of different routing protocols known in VDTN

M
. Benamar, S. Ahnana, F. Z. Saiyari, N. Benamar, M. D. El Ouadghiri, and J.-M. Bonnin 83
2.5 About One Simulator
Opportunistic Network Environment (ONE) is a Java-based open source simulator used for the
implementation and testing of routing protocols. The main functions of the ONE simulator are the
modelling of node movement, inter-node contacts using various interfaces, routing, message handling
and application interactions [24]. It was designed especially for delay tolerant network. By the use of
its Graphic User Interface (GUI), ONE allows the visualization of nodes movement during the
simulations, which facilitates the understanding of the simulation [25]. Results collection and analysis
are done through visualization, reports and post-processing tools [24]. The six routing protocols
mentioned above, are implemented by default in ONE Simulator. Also it uses the map of Helsinki city
and contains different models of mobility such as random way point model, map based movement
model, and shortest path map based movement.
3 Performance Assessment
In this section, we define the scenarios of our comparative study of the well known routing protocols in
VDTN: Epidemic, Spray And Wait, Prophet, and Maxprop. We did not compare Direct Delivery and
First Contact because they are not suitable for our study.
Moreover, we compare the routing protocols using four performance metrics. Delivery
Probability, which is the percentage of successfully, delivered bundles among the bundles sent.
Average latency, which defines the time between the generation of the bundles and their delivery. The
overhead ratio, which measures how many transfers are needed for each bundle delivery. And finally,
the contacts per hour, which defines the number of contacts per hour for the relay nodes.
3.1 Network Setting
In the current study, we did not use the default map of ONE. Indeed, the network scenario is based on
a part of the city of Rennes (France) as shown in fig.3. Ten stationary sources nodes and ten stationary
destinations nodes are placed on the extremities of the map. The coordinates of these nodes are not
arbitrary, but in accordance with the information provided by the open data related to the city of
Rennes. The aim is to transfer bundles from those sources, representing sensors, to destinations
representing gateways connected to the Internet. This communication is enabled through mobile nodes
representing vehicles and relay nodes representing auto signal traffic light that we get from the traffic
information of the city of Rennes. The aim is to simulate VDTN routing protocols using real data.
In the first set of simulations, the number of mobile nodes is 30 to specify sparse network,
then we change it to 80 nodes to simulate the case of dense network. The number of relay nodes
deployed in the network was changed from 0 to 14 nodes to show their impact on the performances.
The position of all nodes is shown in figure 4.

Study of VDTN Routing Protocols Performances in Sparse and Dense Traffic in the Presence of …
84
Figure 3 The part of Rennes’s city map
Figure 4 Position of all nodes before the start of simulation

M
. Benamar, S. Ahnana, F. Z. Saiyari, N. Benamar, M. D. El Ouadghiri, and J.-M. Bonnin 85
The time of simulation is 12 hours. All nodes in the network have a 5M in their buffer size. The group
of mobile nodes (vehicles) moves along the map roads, with a speed ranging from 10 km/h to 50 km/h
by using shortest path map based movement model. Data bundles are originated at specific stationary
source nodes and are destined to specific stationary terminal nodes. No random transmissions have
been used in all simulations. Furthermore, we have used external event generator to create 1000
messages with sizes varying from 100KB to 1MB. These messages are created in an interval time
between the start of the simulation and 900s. The TTL was fixed 60min for all simulations. All nodes
in the network use wifi802.11, with transmission range of 30 meters and transmission speed of
11Mbps. Table 2 summarizes the different parameters used in our study.
Simulation time 43200 secondes
Buffer size 5 M
Movement model Shortest Path Map Based Movement
Time to live (TTL) 60 min (1h)
Routing Protocols [MaxProp; Epidemic; Prophet; Spray And Wait]
Interface type Simple Broadcast Interface
Number of nodes
Sources nodes : 10 nodes
Destinations nodes : 10 nodes
Mobiles nodes : [30;80] nodes
Relay nodes : [0;5;14] nodes
Speed of mobile nodes 10-50 km/h
Size of bundles 100KB, 1MB
The event genertor used External Event Queue
Size of map 10000,10000
Transmission range 30 meters
Transmission speed 11 Mbps
Number of copies (L) 15, the binary mode = true
Table 2 Different parameters used in the scenario configuration
We ran the simulations several times using different seeds for each protocol, and then we
calculate the average of all results to reach the confidence interval. The simulation takes in average 30
to 40 min when the number of nodes in the network is small, contrary to the case when the number is
more important. It also depends on the protocol used because some routing protocols take more time

Study of VDTN Routing Protocols Performances in Sparse and Dense Traffic in the Presence of …
86
than the others. After the end of simulation, ONE provides the possibility of viewing the path of
different messages from their sources to their destinations. In figure 5, we illustrate an example of
messages path from sources Src3 and Src6 to destinations Dest10, Dest11, Dest13, and Dest18.
Figure 5 Example of paths messages from sources nodes to destinations nodes
Maxprop offers the best chance of delivery through its buffer management mechanisms of the
queue, acknowledgments, followed by spray and wait. Epidemic does not provide good results in
terms of the delivery probability because it creates an unlimited number of messages, the same for
prophet, because of its calculated probability, which can cause message loss due to expiration of its
TTL before its transmission.
4 Results And Discussion
4.1 Simulation Scenario With 30 Mobile Nodes
We start our evaluation by simulating a scenario with 30 mobile nodes. The delivery probability
increases with the number of relays because if we have more stationary relays we will have more
possibility that the message reaches its destination because it plays the role of intermediary between
nodes especially in the case where density is very small (Figure 6). In addition, the latency decreases
when increasing the number of fixed relays because there is more chance that the message arrives
promptly (Figure7).
At average latency, Maxprop is not very efficient because, the transmission of the message
requires more time to choose the shortest path to the message, but the Epidemic protocol transmits the
message promptly because, by creating an unlimited number of messages, it explores all available
paths in order to convey the message with minimal delay.

M
. Benamar, S. Ahnana, F. Z. Saiyari, N. Benamar, M. D. El Ouadghiri, and J.-M. Bonnin 87
Figure 6 Delivery probability depending on relay nodes in sparse traffic
Figure 7 Average latency depending on relay nodes in sparse traffic

Study of VDTN Routing Protocols Performances in Sparse and Dense Traffic in the Presence of …
88
Figure 3 Overhead ratio depending on relay nodes in sparse traffic
The multitude of copies in Epidemic justifies valuable overhead ratio with this protocol. With
spray and wait this amount of overhead ratio is very small because this protocol created a limited
number of copies that can optimize the use of bandwidth. (Figure 8)
Figure 9 Number of contacts per hour in sparse traffic
The number of contact time increases when the number of relays increases, as the relay
function is to act as an intermediary between two or more mobile nodes, it receives a message from a
node and forwards it to another which causes an increase in the number of contacts. (Figure 9)

M
. Benamar, S. Ahnana, F. Z. Saiyari, N. Benamar, M. D. El Ouadghiri, and J.-M. Bonnin 89
4.2SimulationScenariowith80MobileNodes
The second scenario has 80 mobile nodes in the network, in an environment with medium density
Spray and Wait is the best in terms of delivery probability followed by Maxprop. Identically to the
other case where the traffic is sparse, Epidemic is not very effective at this metric compared to other
protocols. (Figure 10)
Figure 10 Delivery probability depending on relay nodes in dense traffic
In this environment, the message in Maxprop take less time to reach the destination compared
to the others routing protocols, as opposed to spray and wait protocol that offers the greatest time for
the transmission of messages. (Figure 11)
Figure 12 shows that the overhead is very large in this case with epidemic routing protocol,
which uses very large memory resources followed by the prophet protocol given the number of copies
used by the first and the exchange of predictability vectors used by the second. Unlike spray and wait
protocol that uses fewer resources because of the limited number of bundle copies being generated.
As mentioned earlier, the number of contacts increases when the relay nodes are present in
the network because of their role. Moreover, the simulations shows that in dense traffic the contacts
between nodes increase remarkably, as a result of the number of nodes moving in the network. (Figure
13)

Study of VDTN Routing Protocols Performances in Sparse and Dense Traffic in the Presence of …
90
Figure 41 Average latency depending on relay nodes in dense traffic
Figure 5 Overhead ratio depending on relay nodes in dense traffic

M
. Benamar, S. Ahnana, F. Z. Saiyari, N. Benamar, M. D. El Ouadghiri, and J.-M. Bonnin 91
Figure 6 Number of contacts per hour for dense traffic
5 Conclusion
This study compared the performances of different VDTN routing protocols in sparse and dense
environment and with different values of relay nodes. Our goal was to compare between various
routing protocols using real data of the city of Rennes in order to bring out the impact of relay nodes in
such environment.
In this paper, we have used the ONE simulator for our simulations. The results demonstrate
that there is no best routing protocol suitable for all cases, and all metrics; indeed each protocol has its
strengths and weaknesses depending on the scenario under consideration, which includes the density of
the traffic.
The use of relay nodes in this study demonstrates their important role as intermediates
between mobile nodes to store the bundles for a maximum amount of time, especially when the
number of nodes in the network is very small.
From the analysis of the routing protocols in our study, we conclude that it is necessary to develop and
design a robust routing protocol that could be more efficient in various situations and can be easily
implemented.

Study of VDTN Routing Protocols Performances in Sparse and Dense Traffic in the Presence of …
92
References
[1] V. N. G. J. Soares, F. Farahmand, and J. J. P. C. Rodrigues, “Improving Vehicular Delay-
Tolerant Network Performance with Relay Nodes,” 2009 Next Generation Internet Networks,
pp. 1–5, Jul. 2009.
[2] V. N. G. J. Soares, J. J. P. C. Rodrigues, and F. Farahmand, “GeoSpray: A Geographic
Routing Protocol for Vehicular Delay-Tolerant Networks,” Information Fusion, Nov. 2011.
[3] N. Magaia, P. R. Pereira, M. P. Correia, I. I. S. T. Utl, and R. A. Redol, “Nodes ’ Misbehavior
in Vehicular Delay-Tolerant Networks,” 2013.
[4] V. N. G. J. Soares, J. J. P. C. Rodrigues, and F. Farahmand, “Performance assessment of a
geographic routing protocol for vehicular delay-tolerant networks,” 2012 IEEE Wireless
Communications and Networking Conference (WCNC), pp. 2526–2531, Apr. 2012.
[5] N. Benamar, M. Benamar, S. Ahnana, F. Z. Saiyari, M. Driss, E. L. Ouadghiri, and J. M.
Bonnin, “ARE VDTN ROUTING PROTOCOLS SUITABLE FOR DATA COLLECTION IN
SMART CITIES: A PERFORMANCE ASSESMENT,”. Journal of Theoretical and Applied
Information Technology, 31th December 2013. Vol. 58 No.3
[6] R. S. Mangrulkar and M. Atique, “Procedia Computer Science Heterogeneous Highly Dense
Mobile Environment,” vol. 00, pp. 1–13, 2013.
[7] D. Mishra and M. Chawla, “Minimax Theory Based Scheme to Detect Selfish Node and
Reduce Latency in Delay Tolerant Network,” vol. 2013, no. Cac2s, pp. 78–82, 2013.
[8] N. Magaia, P. R. Pereira, M. P. Correia, I. I. S. T. Utl, and R. A. Redol, “Selfish and Malicious
Behavior in Delay-Tolerant Networks.”
[9] Y. Wang, W. Yang, and J. Wu, “Analysis of a Hypercube-based Social Feature Multi-Path
Routing in Delay Tolerant Networks,” vol. X, no. X, 2012.
[10] M. Karimzadeh, M. Of, and S. Thesis, “EFFICIENT ROUTING PROTOCOL IN DELAY
TOLERANT NETWORKS ( DTNs ),” 2011.
[11] M. J. Khabbaz, C. M. Assi, and W. F. Fawaz, “Disruption-Tolerant Networking: A
Comprehensive Survey on Recent Developments and Persisting Challenges,” IEEE
Communications Surveys & Tutorials, vol. 14, no. 2, pp. 607–640, 2012.
[12] A. Casaca, J. J. P. C. Rodrigues, V. N. G. J. Soares, and J. Triay, “From Delay-Tolerant
Networks to Vehicular Delay-Tolerant Networks,” Architecture, vol. PP, no. 99, pp. 1–17,
2011.
[13] V. N. G. J. Soares, F. Farahmand, and J. Rodrigues, “A layered architecture for Vehicular
Delay-Tolerant Networks,” 2009 IEEE Symposium on Computers and Communications, pp.
122–127, 2009.
[14] J. Zhou, J. Li, Y. Qian, S. Member, S. Roy, and K. Mitchell, “Quasi-Optimal Dual-Phase
Scheduling for Pigeon Networks,” vol. 61, no. 9, pp. 4157–4169, 2012.

M
. Benamar, S. Ahnana, F. Z. Saiyari, N. Benamar, M. D. El Ouadghiri, and J.-M. Bonnin 93
[15] H. Wang, X. Liu, and X. Hu, “The mobile scenario influence on DTN routing.”
[16] S. Rashid, Q. Ayub, M. S. M. Zahid, and A. H. Abdullah, “E-DROP: An Effective Drop
Buffer Management Policy for DTN Routing Protocols,” vol. 13, no. 7, pp. 8–13, 2011.
[17] A. Vahdat and D. Becker, “Epidemic routing for partially connected ad hoc networks,”
Technical Report,, 2000.
[18] C. S. Raghavendra, “Spray and Wait: An Efficient Routing Scheme for.”
[19] N. T. M. Case, T. Spyropoulos, S. Member, and K. Psounis, “Efficient Routing in
Intermittently Connected Mobile,” vol. 16, no. 1, pp. 77–90, 2008.
[20] C. B. Desai, V. N. Pandya, and P. M. Dolia, “Comparative Analysis of Different Routing
Protocols in delay tolerant networks,” vol. 4, no. 03, pp. 249–253, 2013.
[21] Z. J. Z. Jin, X. Z. X. Zhao, Y. L. Y. Luo, and D. Z. D. Zhao, “Adaptive Priority Routing with
Ack_Mechanism for DTN networks,” 2009 International Conference on Wireless
Communications Signal Processing, vol. 3, no. 3, pp. 1–5, 2009.
[22] A. Lindgren and A. Doria, “Probabilistic Routing in Intermittently Connected Networks,” pp.
1–8.
[23] J. Burgess, B. Gallagher, D. Jensen, and B. N. Levine, “MaxProp: Routing for Vehicle-Based
Disruption-Tolerant Networks,” Proceedings IEEE INFOCOM 2006. 25TH IEEE
International Conference on Computer Communications, pp. 1–11, 2006.
[24] A. Keränen, T. Kärkkäinen, and J. Ott, “Simulating Mobility and DTNs with the ONE (Invited
Paper),” Journal of Communications, vol. 5, no. 2, pp. 92–105, Feb. 2010.
[25] “A DTN Study: Analysis of Implementations and Tools,” 2010.