You are on page 1of 2

2017 IEEE 22nd Pacific Rim International Symposium on Dependable Computing

Erasure-Code-Based DTN Multi-Path Routing for Contact Avoidance

Hironori Arai† Masayuki Arai†,‡


†Graduate School of Industrial Technology ‡College of Industrial Technology
Nihon University, Chiba, Japan
E-mail: arai.masayuki@nihon-u.ac.jp

Abstract— Delay (or disruption) tolerant networks (DTNs) have of the connecting nodes which is expected to communicate
widely studied as promising technology for providing with the receiver or closer-to-receiver nodes to store the data
communications in environments where all nodes are not to relay the communication. Which and how many nodes are
connected with each other. Assuming the existence of adversary selected are depending on the routing protocol applied. In the
nodes, in this study we evaluate effectiveness of erasure-code-
figure, the sender requests to two nodes painted with dark
based DTN routing in terms of secure transmission. We combine
secret-sharing-based multi-path routing with several routing color to relay the data. After the upper node leaves from the
strategies, and evaluate the ratio of successful transmission by sender, it connects to another relaying node, and when the
simulation. second relaying node connects to the receiver, transmission
between the sender and receiver completes.
Keywords-Delay Tolerant Network, contact avoidance, multi-
path routing, erasure code

I. INTRODUCTION sender
receiver

Delay (or disruption) tolerant networks (DTNs) have


widely studied as promising technology for providing sender
sender
communications in environments where all nodes are not receiver receiver
connected with each other [1,2]. Since DTNs utilize wireless
transmissions between nodes and store-and-forward routing in
disconnected environments, nodes might suffer from sniffing Figure 1. Example of data transmission in DTN.
and other attacks, and thus the systems have to provide secure
communications even under existence of adversaries, B. Erasure-code-based DTN routing for delay reduction
especially for critical applications such as military and so on. In DTN environment, end-to-end transmission delay is
In this study we investigate effectiveness of erasure-code- strongly affected by contact interval of two relaying nodes.
based multi-path routing for DTN in terms of secure Wang et al. proposed application of erasure coding to DTNs,
transmissions under existence of adversary nodes. In the aiming at reducing average transmission delay [3]. Figure 2
proposed routing protocol, encoded fragments are transmitted illustrates the basic idea of erasure-code-based transmission.
via different routes from each other. Then, not only the target First, a sender divides the original data into k fragments, and
receiver but also the set of adversary nodes cannot reconstruct then generates (n – k) redundant fragments generated based on
original data unless they receive at least k out of n fragments. (n, k) erasure coding. After that, total n fragments are
Assuming the locations of adversary nodes are known, we transmitted in the network via different routes from each other.
establish several routing strategies, and evaluate data Even if some fragments suffer from large transmission delay,
reachability without being stolen by simulations. the receiver can reconstruct the original data when it receives
any combination of k out of n fragments, due to characteristics
of erasure coding.
II. PRELIMINARIES
divide into k fragments
A. Delay tolerant networks add (n - k) redundant fragments
Figure 1 illustrates data transmission under DTNs. If two
nodes are in the range of wireless transmission, they can
communicate directly with each other. A node which can send total n fragments via different paths
communicate with two or more nodes can relay the
communication between connecting nodes, and thus a node
can communicate with a subset of nodes in the system,
directly or indirectly, at a given instant. In the figure the dotted original data can be reconstructed by k or more fragments
line represents the link that is currently unavailable.
If a sender finds that the receiver is not in the subset of Figure 2. Data fragmentation and reconstruction by erasure-code-based
directly or indirectly connecting nodes, it then requests to any DTN transmission

978-1-5090-5652-1/17 $31.00 © 2017 IEEE 215


DOI 10.1109/PRDC.2017.37
C. Attack model and contact avoidance routing longcut avoidance, regardless of application of adversary and
Sakai et al. suggested existence of adversaries in mobile adjacent avoidance.
A+D
ad-hoc networks (MANETs) and proposed avoidance routing 1 C+L+A+D
C+L+A
[4]. In Ref. [5], the protocol was then extended to contact C+L
C+A+D
avoidance, targeting DTN. Adversaries are assumed to have 0.8
C+A

Success Ratio
infinite computational ability, thus existing data encryption 0.6
C

strategies are not valid, once the data is transmitted through


0.4
adversaries' wireless communication area. Assuming that
adversaries' locations are known to all nodes, avoidance and 0.2
contact avoidance routings try to avoid selecting relaying node
0
that are adjacent to adversaries. 0 2 4 6 8 10
# adversaries

III. EVALUATION OF ERASURE-CODE-BASED DTN Figure 3. Simulation results of ratio of successful transmission for
ROUTING FOR CONTACT AVOIDANCE different number of adversaries. (100 nodes, (4, 8) secret sharing)

We evaluate the effectiveness of erasure-code-based Figure 4 compares the transmission delay for successful
routing in terms of not only delay reduction but also secure transmission, in the same setup as Fig. 3. In exchange of
transmission under existence of adversary nodes. We assume improved reachability, delay is increased.
to apply (k, n) secret sharing which is information- 50 A+D
theoretically secure, that is, even the set of adversaries with C+L+A+D
C+L+A
infinite computational ability cannot reconstruct the original 40 C+L

Average of Success_time
C+A+D
data unless they acquire at least k out of n fragments. C+A
30 C
In addition to erasure coding (C), we applied the
combination of the following three strategies to routing 20
protocol:
10
Longcut avoidance (L): avoids to select nodes which are not
0
closer to the receiver 0 2 4 6 8 10
Advsersary avoidance (A): avoids to select adversary nodes # adversaries

as relay Figure 4. Simulation results of delay for succesfull transmission different


Advsersary's adjacent avoidance (D): avoids to select nodes number of adversaries. (100 nodes, (4, 8) secret sharing)
which might connect to adversaries
IV. CONCLUSIONS
We performed Monte-Carlo simulations, where In this study we investigate effectiveness of erasure-code-
connections of the nodes are simply represented by a contact based multi-path routing for DTN in terms of secure
graph, without considering physical locations of the nodes and transmissions under existence of adversary nodes. We applied
wireless transmission ranges. When a sender initiates a (k, n) secret sharing-based erasure coding and several routing
transmission, the transmission is determined as successful if strategies to DTN. Simulation results indicated that the
at least k fragments are delivered to the receiver until the combination of erasure coding, longcut avoidance, and
specified deadline while the set of adversaries fails to collect adversary and adversary's adjacent avoidances significantly
k or more fragments, or as failed otherwise. According to the improved the ratio of successful transmissions.
routing strategies applied, the sender or relaying nodes might Further tuning of the protocol considering the trade-offs
sometimes find that there are no safe routes and give up between transmission delay and delivery ratio is one of the
relaying. In this case the transmission is also determined as open issues.
failed.
Figure 3 shows the simulation results for different REFERENCES
numbers of adversary nodes. Total number of nodes in the [1] T. Spyropoulos, K. Psounis and C. S Raghavendra, "Spray and Wait:
An Efficient Routing Scheme for Intermittently Connected Mobile
network is 100, and (4, 8) secret sharing is applied as erasure Networks," SIGCOMM Workshop on Delay-tolerant Networking, pp.
coding, that is, receiver and adversaries can reconstruct 252-259, 2005.
[2] A. Balasubramanian, B. Levine and A. Venkataramani, "DTN Routing
original data if 4 out of 8 fragments are delivered. The label as a Resource Allocation Problem," SIGCOMM, pp. 373-384, 2007.
"A+D" stands for the protocol where adversaries and [3] Y. Wang, S. Jain, M. Martonosi, K. Fall, “Erasure-Coding Based
adversary's adjacent nodes are not select as relaying node but Routing for Opportunistic Networks”, ACM SIGCOMM workshop on
Delay-tolerant networking, pp.229-236, 2005.
erasure coding is not applied, which corresponds to the [4] K. Sakai, M.-T. Sun, W.-S. Ku, J. Wu and T. H Lai, Multi-Path-Based
conventional contact avoidance routing. The figure indicates Avoidance Routing in Wire-less Networks," International Conference
on Distributed Computing Systems, pp. 706-715, 2015.
that the ratio of successful transmission can be improved by [5] T. Osuki, K. Sakai, S. Fukumoto, and M. Arai, "Contact Avoidance
applying erasure coding, when there are three or more Routing for Delay Tolerant Networks," IEICE Technical Report, No.
IN2015-13, pp. 25-28, 2015 (in Japanese).
adversaries. The figure also indicates the effectiveness of

216

You might also like