Abstract—Vehicular Delay-Tolerant Networking (VDTN) is an
extension of the Delay-Tolerant Network (DTN) architecture
concept to transit networks. VDTN architecture handles non-real
time applications, exploiting vehicles to enable connectivity under
unreliable scenarios with unstable links and where an end-to-end
path may not exist. Intuitively, the use of stationary
store-and-forward devices (relay nodes) located at crossroads
where vehicles meet them and should improve the message
delivery probability. In this paper, we analyze the influence of the
number of relay nodes, in urban scenarios with different
numbers of vehicles. It was shown that relay nodes significantly
improve the message delivery probability on studied DTN
Index Terms—Vehicular Delay-Tolerant Networks; Relay
Nodes; Delay-Tolerant Networks
Delay Tolerant Networks (DTNs)  are a class of
networks designed to address several challenging connectivity
issues such as sparse connectivity, long or variable delay,
intermittent connectivity, asymmetric data rate, high latency,
high error rates and even no end-to-end connectivity. The
DTN architecture adopts a store-and-forward paradigm and a
common bundle layer located on the top of region-specific
network protocols in order to provide interoperability of
heterogeneous networks (regions). In this type of network, a
source node originates a message (bundle) that is forwarded to
an intermediate node (fixed or mobile) thought to be more
close to the destination node. The intermediate node stores the
message and carries it while a contact is not available. Then
the process is repeated, so the message will be relayed hop by
hop until reaching its destination.
The concept of Delay-Tolerant Networking has been widely
applied to scenarios like interplanetary networking , data
MULEs , underwater networks , and wildlife tracking
sensor networks like ZebraNet . Vehicular networks [6, 7]
are another example for an application of the DTN concept.
In this paper we exemplify the use of a Vehicular DTN
(VDTN) to provide asynchronous communication between
mobile nodes and relay nodes, on an old part of a city with a
large area and restricted vehicular access (Fig. 1). Mobile
nodes (e.g., vehicles) physically carry the data, exchanging
information with one another. They can move along the roads
randomly (e.g. cars), or following predefined routes (e.g.
buses and trams). Relay nodes are stationary devices located at
crossroads, with store-and-forward capabilities. They allow
mobile nodes passing by to pickup and deposit data on them.
We can also envision the possibility for the relay nodes to be
able to exchange data with each other, and at least one of them
may have a direct access to the Internet.
Some of the potential non-real time applications for this
scenario are: notification of blocked roads, accident warnings,
free parking spots, advertisements, and also gathering
information collected by vehicles such as road pavement
Fig. 1. Example of the use of a VDTN in an urban scenario.
Improving Vehicular Delay-Tolerant Network
Performance with Relay Nodes
Vasco N. G. J. Soares1,2,3, Farid Farahmand4, and Joel J. P. C. Rodrigues1,2
1Instituto de Telecomunicações, NetGNA Group, Covilhã, Portugal
2Department of Informatics, University of Beira Interior, Covilhã, Portugal
3Superior School of Technology, Polytechnic Institute of Castelo Branco, Castelo Branco, Portugal
4Department of Engineering Science, Sonoma State University, CA, USA
firstname.lastname@example.org; email@example.com; firstname.lastname@example.org
The use of relay nodes should create a greater number of
connectivity opportunities, improving the performance of the
VDTN network in terms of message delivery probability. The
key contribution of this paper is the evaluation of the impact
of the number of relay nodes on DTN routing protocols, in
scenarios with different numbers of mobile nodes.
The remainder of this paper is organized as follows. Section
II briefly reviews the related work on fixed relay node
deployment, identifying our contribution. Section III presents
the simulation scenario and discusses the results. Section IV
concludes the paper and provides guidelines for future work.
II. RELATED WORK
The usage of stationary nodes to improve the overall
performance of mobile DTNs has been studied in [8-12]. In
 the authors suggest the use of throwboxes in mobile DTNs,
in order to increase the number of contact opportunities thus
improving the network capacity. They propose algorithms to
deploy the throwboxes that consider both placement and
routing. This work is complemented in , where the authors
present an energy efficient hardware and software architecture
The work in  considers the cases where the throwboxes
are fully disconnected or mesh connected, analyzing for each
case the impact of the number of throwboxes over the
performance of routing protocols. The work in  evaluates
the relation of adding relay nodes to the overall network
performance of a Vehicular Wireless Burst Switching
Network (VWBS). It proposes and compares the performance
of heuristic algorithms whose objective is to maximize the
network performance in terms of delay or network cost,
providing a solution to the relay node placement problem. In
 the authors study the tradeoffs of mobile networks
enhanced with the deployment of relays, meshes, and wired
base stations infrastructure.
Our work considers Vehicular Delay-Tolerant Network as a
particular application for a mobile DTN characterized by the
opportunistic contacts, where end-to-end connectivity may not
exist, and intermittent connectivity is common. We are
interested in the study of the impact analyses produced by the
relay nodes in scenarios with different numbers of mobile
III. PERFORMANCE EVALUATION
To demonstrate how relay nodes improve the performance
of a VDTN network, we run several simulations using the
Opportunistic Network Environment (ONE) Simulator .
In the networks scenario, the number of mobile nodes, and the
number of deployed relay nodes on the network was changed.
The overall message delivery ratio (measured as the relation of
the number of unique delivered messages to the number of
messages sent), and the message delivery delay (measured as
the time between message creation and delivery) were
analyzed for the following four DTN routing protocols:
Epidemic , MaxProp , PRoPHET , and
Spray-and-Wait (binary and normal variants, with 12 message
Epidemic is a flooding-based scheme where the nodes
exchange the messages they don’t have. MaxProp prioritizes the
schedule of messages transmitted to other nodes and also the
schedule of messages to be dropped. PRoPHET is a
probabilistic routing protocol that considers a history of
encounters and transitivity. Finally, the Spray-and-Wait
protocol creates a number of copies to be transmitted per
message. At each message transfer the number of copies
remaining is reduced in one unit in the normal mode, or in the
case of the binary mode the number of copies left is reduced in
For the simulation scenarios we use the map-based model of
a part of the city of Helsinki (Fig. 2) available on the ONE
Simulator. We simulate a 12-hour period and measure the
differences in performance, when 0, 5, or 10 relay nodes are
deployed in (different) network scenarios with 20, 40 or 60
Fig. 2. Helsinki simulation area with the locations of the relay nodes.
Mobile nodes (vehicles) move between random map
locations. Once a mobile node reaches a destination, it
randomly waits 5 to 15 minutes. Then, it selects a new random
map location, and a random speed between 10 and 50 km/h.
The mobile node moves to the new destination using the
shortest path available. Each of the mobile nodes has a 150
Mbytes FIFO message buffer.
Messages are exchanged between random source and
destination mobile nodes. It is used an inter-message creation
interval in the range [15, 30] (seconds) of uniformly
distributed random values. Message size is in the range [500
KB, 1 MB] of uniformly distributed random values. All the
messages exchanged have a time to live (TTL) of 1 hour.
We assume that the traffic matrix is not provided in
advance, and the mobile nodes routes are not pre-assigned and
fixed, so there isn’t any knowledge about the transfer
opportunities. Therefore, we choose the places for the relay
nodes using a non-uniform strategy, positioning them at the
crossroads of the main roads of the simulation scenario (Fig.
2). Each of the relay nodes has a 500 Mbytes FIFO message
Network nodes connect to each other using IEEE 802.11b
with a data rate of 6 Mbit/s (the IEEE 802.11b approximate
throughput according to ), and a transmission range of 30
meters. Relay nodes coverage cells do not intersect, so they
are not able to communicate directly with each other, only
with the vehicles. In addition, vehicles exchange data between
We run series of simulations for each combination of the
parameters: number of vehicles, and number of relay nodes.
We use different random seeds, and report the mean values.
A. Simulation Scenario with 20 Mobile Nodes
We start our evaluation by simulating a scenario with 20
mobile nodes. Because of the low node density, few
transmission opportunities are registered when no relay nodes
are deployed in the network (Fig. 3). Deploying relay nodes
augments the number of contact opportunities per hour
between all network nodes. Introducing 10 relay nodes
increase the number of contacts at a rate of roughly a factor of
two per hour. This effect suggests that relay nodes will
contribute to increase the number of messages exchanged
#"$"%" &"'"(" )"*"+"#!" ##"#$"
Fig. 3. Number of contacts per hour between all network nodes.
Figure 4 shows that all routing protocols increase message
delivery ratio when relay nodes are deployed. When analyzing
the simulation results with the introduction of 5 relay nodes,
we observe that Epidemic and PRoPHET protocols that
perform variants of flooding increase their message delivery
probabilities in 7% and 8%, respectively. MaxProp, that also
floods but implements explicit message clearing after
delivering, improves 9%. Spray-and-Wait that creates a
number of copies per message presents gains of 8% and 7% in
its binary and normal variants.
Increasing the number of relay nodes to 10 augments the
delivery ratio even more. Epidemic and PRoPHET register the
least improvements, 2% and 1% respectively. MaxProp
increases its message delivery probability further in 4%.
Spray-and-Wait binary variant augments 5%, whereas the
normal variant has a gain of 4%. Finally, it can be observed
that MaxProp is the routing protocol that takes more benefits
from the introduction of the stationary relay nodes, registering
the best delivery probabilities.
Fig. 4. Message delivery probability.
Fig. 5. Message average delay.
The message average delay is an interesting metric, since
minimizing it reduces the time that messages spend in the
network and reduces the contention for resources in the
network (e.g. buffer). In the context of this work the messages
have a small TTL and size, and the nodes have a sufficient
large buffer. Therefore, we can focus on message delivery
probability as the main performance metric. Fig. 5 shows that
all routing protocols register similar values for the message
average delay, and that relay nodes do not significantly affect
B. Simulation Scenario with 40 Mobile Nodes
The second scenario has 40 mobile nodes in the network,
therefore the number of contact opportunities increases
(Fig. 6). As a result, we observe that all routing protocols
perform better than in the previous scenario (Fig. 7).
In this new scenario, based on fact that having the double of
mobile nodes and, consequently, a much larger number of
opportunistic contacts, it could be expected that relay nodes
would not affect the performance of the network considerably.
However, in Fig. 7 it may be observed that 5 relay nodes
provides up to 9% of gain in message delivery probability for
Epidemic routing protocol, 13% for PRoPHET, 13% for
MaxProp, 12% for Spray-and-Wait binary variant, and 10%
for the normal variant. Notice that these performance gains are
superior to the ones presented on the first scenario.
Nevertheless, deploying 10 relay nodes instead of 5, does not
bring more benefits.
Furthermore, it can be seen that MaxProp and Spray-and-
Wait binary variant perform better than the other protocols,
independently of the number of relay nodes.
#"$" %"&"'"(")"*"+"#!"##" #$"
Fig. 6. Number of contacts per hour between all network nodes.
Fig. 7. Message delivery probability.
Epidemic and PRoPHET routing protocols approximately
maintain the message average delay across the simulations
(Fig. 8). The other routing protocols register a very slight
decrease on the average delay.
Fig. 8. Message average delay.
C. Simulation Scenario with 60 Mobile Nodes
In this last scenario, we augment the number of mobile
nodes to 60. This results in an increase of the number of
transmission opportunities, and in the reduction between
inter-contact times (Fig. 9). Therefore, routing protocols will
perform even better.
#"$"%" &" '"(" )"*"+" #!"##"#$"
Fig. 9. Number of contacts per hour between all network nodes.
Nevertheless, relay nodes still have a positive impact on the
message delivery probability. As Fig. 10 shows, when 5 relay
nodes are deployed, Epidemic and MaxProp increase their
message delivery probability approximately in 6% and 4%,
respectively. PRoPHET improves 5% The same is observed
with Spray and Wait variants, that improve 2% and 5%
respectively. However, increasing the number of relay nodes
to 10 decreases the message delivery ratio. This is due to the
problems caused by storage constraints. Having a large node
density will cause more data to be exchanged and stored on
the network nodes.
Finally, it can be observed that MaxProp is the routing
protocol again registers the best delivery probabilities.
Fig. 10. Message delivery probability.
Fig. 11 illustrates the message average delay for the five
protocols in the scenario under study, with 60 mobile nodes.
As may be seen in the previous scenario, Epidemic and
PRoPHET approximately maintain the same message average
delay across the simulations. The other routing protocols also
register a very slight decrease on the average delay.
Fig. 11. Message average delay.
IV. CONCLUSIONS AND FUTURE WORK
This paper studied the performance impact (in terms of
message delivery probability and average delay) of relay
nodes on a VDTN applied to an urban scenario. It was
assumed a cooperative opportunistic environment without
knowledge of contact opportunities and traffic matrix. The
motivation for this work comes from the idea that placing
relay nodes at crossroads allows data deposit and pickup by
passing mobile nodes, which will increase the delivered
messages (probability) to the final destination.
Several experiments were conducted varying the number of
mobile nodes, and deploying a different number of relay nodes
in predefined map locations (over the considered scenario). It
was observed that relay nodes significantly improve the
message delivery probability on the routing protocols.
Our interests for future work are focused on the
performance evaluation of VDTN architecture on isolated and
dispersed regions without network infrastructure, studying the
impact of node cooperation , geographical routing
protocols , and the relay node placement problem.
Part of this work has been supported by the Instituto de
Telecomunicações, Next Generation
Applications Group, Portugal, in the framework of the Project
VDTN@Lab, and by the Euro-NF Network of Excellence of
Seven Framework Programme of EU.
 V. Cerf, S. Burleigh, A. Hooke, L. Torgerson, R. Durst, K. Scott, K.
Fall, and H. Weiss, "Delay-Tolerant Networking Architecture," RFC
4838, April 2007, [Online]. Available: ftp://ftp.rfc-editor.org/in-
 S. Burleigh, A. Hooke, L. Torgerson, K. Fall, V. Cerf, B. Durst, K.
Scott, and H. Weiss, "Delay-Tolerant Networking: An Approach to
Interplanetary Internet," in IEEE Communications Magazine, vol. 41,
2003, pp. 128-136.
 S. Jain, R. Shah, W. Brunette, G. Borriello, and S. Roy, "Exploiting
Mobility for Energy Efficient Data Collection in Wireless Sensor
Networks," ACM/Kluwer Mobile
(MONET), vol. 11, pp. 327-339, June 2006.
 J. Partan, J. Kurose, and B. N. Levine, "A Survey of Practical Issues in
Underwater Networks," in 1st ACM International Workshop on
Underwater Networks, in conjunction with ACM MobiCom 2006, Los
Angeles, California, USA, Sep. 25, 2006, pp. 17 - 24.
 P. Juang, H. Oki, Y. Wang, M. Martonosi, L. S. Peh, and D. Rubenstein,
"Energy-Efficient Computing for Wildlife Tracking: Design Tradeoffs
and Early Experiences with ZebraNet," ACM SIGOPS Operating
Systems Review, vol. 36, pp. 96 - 107, 2002.
 W. Zhao, M. Ammar, and E. Zegura, "A Message Ferrying Approach
for Data Delivery in Sparse Mobile Ad Hoc Networks," in The Fifth
ACM International Symposium on Mobile Ad Hoc Networking and
Computing (MobiHoc 2004), Roppongi Hills, Tokyo, Japan, May 24-26,
2004, pp. 187–198.
 A. Pentland, R. Fletcher, and A. Hasson, "DakNet: Rethinking
Connectivity in Developing Nations," in IEEE Computer, vol. 37, 2004,
 W. Zhao, Y. Chen, M. Ammar, M. Corner, B. Levine, and E. Zegura,
"Capacity Enhancement using Throwboxes in DTNs," in IEEE
International Conference on Mobile Adhoc and Sensor Systems (MASS)
 M. Ibrahim, A. A. Hanbali, and P. Nain, "Delay and Resource Analysis
in MANETs in Presence of Throwboxes," Performance Evaluation, vol.
64, pp. 933-947, October 2007.
 F. Farahmand, A. N. Patel, J. P. Jue, V. G. Soares, and J. J. Rodrigues,
"Vehicular Wireless Burst Switching Network: Enhancing Rural
Connectivity," in The 3rd IEEE Workshop on Automotive Networking
and Applications (Autonet 2008), Co-located with IEEE GLOBECOM
2008, New Orleans, LA, USA, December 4, 2008.
 N. Banerjee, M. D. Corner, and B. N. Levine, "An Energy-Efficient
Architecture for DTN Throwboxes," in INFOCOM 2007 - 26th IEEE
International Conference on Computer Communications, 2007, pp. 776-
 N. Banerjee, M. D. Corner, D. Towsley, and B. N. Levine, "Relays,
Base Stations, and Meshes: Enhancing Mobile Networks with
Infrastructure," in 14th ACM International Conference on Mobile
Computing and Networking (ACM MobiCom), San Francisco,
California, USA, September, 2008, pp. 81-91.
 A. Keränen, J. Ott, and T. Kärkkäinen, "The ONE Simulator for DTN
Protocol Evaluation," in SIMUTools'09: 2nd International Conference
on Simulation Tools and Techniques, Rome, March 2-6, 2009.
 A. Vahdat and D. Becker, "Epidemic Routing for Partially-Connected
Ad Hoc Networks," Duke University, Technical Report CS-2000-06,
 J. Burgess, B. Gallagher, D. Jensen, and B. Levine, "MaxProp: Routing
for Vehicle-Based Disruption-Tolerant Networks," in INFOCOM 2006 -
The 25th IEEE International Conference on Computer Communications,
Barcelona, Catalunya, Spain, April 23-29, 2006, pp. 1-11.
 A. D. Anders Lindgren, "Probabilistic Routing Protocol for
Intermittently Connected Networks,"
February 2008, [Online]. Available: http://tools.ietf.org/html/draft-irtf-
 T. Spyropoulos, K. Psounis, and C. S. Raghavendra, "Spray and Wait:
An Efficient Routing Scheme for Intermittently Connected Mobile
Networks," in ACM SIGCOMM 2005, August, 2005.
 Cisco Systems, Inc., "Capacity Coverage & Deployment Considerations
for IEEE 802.11g,"
ite_paper09186a00801d61a3.shtml [Accessed: April, 2009].
 A. Panagakis, A. Vaios, and I. Stavrakakis, "On the Effects of
Cooperation in DTNs," in COMSWARE, Bangalore, India, January 7-12,
 I. Leontiadis and C. Mascolo, "GeOpps: Geographical Opportunistic
Routing for Vehicular Networks," in IEEE International Symposium on
a World of Wireless, Mobile and Multimedia Networks 2007 (WoWMoM
2007), Espoo, Finland, 18-21 June, 2007, pp. 1-6.
Networks and Applications