Poster: Probabilistic Routing in Intermittently Connected
Networks
Anders Lindgren
Lule
˚
a University of Technology
S-971 87 Lule
˚
a, Sweden
dugdale@sm.luth.se
Avri Doria
Lule
˚
a University of Technology
S-971 87 Lule
˚
a, Sweden
avri@acm.org
Olov Schel
´
en
Lule
˚
a University of Technology
S-971 87 Lule
˚
a, Sweden
olov@sm.luth.se
ABSTRACT
We consider the problem of routing in intermittently con-
nected networks. In such networks there is no guarantee
that a fully connected path between source and destination
exist at any time, rendering traditional routing protocols
unable to deliver messages between hosts. We propose a
probabilistic routing protocol for such networks.
Categories and Subject Descriptors
C.2.1 [
Computer-Communication Networks
]: Network
Architecture and Design
General Terms
Algorithms, Design
Keywords
Delay Tolerant Networks, routing, intermittently connected
networks
1. INTRODUCTION
Normally, one of the most basic requirements for enabling
two nodes to communicate through a network is that there
exist a fully connected path between them. However, there
are scenarios where this is not the case, but where it still
would be desirable to allow communication between nodes.
Such scenarios include communication between villages of
the Saami population of reindeer herders in the north of Swe-
den [4], and other aboriginal populations and populations in
p oor regions [6], but also satellite communication [7], sensor
networking [1], and other areas where the Delay Tolerant
Networking (DTN) architecture [2] is of interest. In such
networks, the mobility of nodes can be used to enable the
delivery of messages – for example, even if there never is a
path between nodes A and C, A might meet node B and give
a message to it, and B might later encounter C, delivering
the message to its final destination. In subsequent sections,
Permission to make digital or hard copies of all or part of this work for
personal or classroom use is granted without fee provided that copies are
not made or distributed for profit or commercial advantage and that copies
bear this notice and the full citation on the first page. To copy otherwise, to
republish, to post on servers or to redistribute to lists, requires prior specific
permission and/or a fee.
MobiHoc’03,
June 1–3, 2003, Annapolis, Maryland, USA.
Copyright 2003 ACM 1-58113-684-6/03/0006 ...
$5.00.
some possible approaches for routing in a network where
connectivity is intermittent are discussed, and a probabilis-
tic routing protocol for such networks is outlined.
2. RELATED WORK
Vahdat and Becker present a protocol for epidemic routing
in intermittently connected networks [8]. When two nodes
encounter each other, they exchange messages being car-
ried (subject to buffer space), thus causing the messages to
spread through the network like an epidemic of a disease.
This approach ensures that a message reaches its destina-
tion as soon as possible, but it also wastes a lot of resources
through unnecessary message transfers.
Due to the ambiguity in deciding what the best next hop
is in the networks discussed here, Chen and Murphy propose
that applications should be able to affect that through the
introduction of a utility function [3], allowing applications to
sp ecify weights of several factors influencing the function. In
the proposed solution, called Disconnected Transitive Com-
munication, a discovery protocol is used to find the best next
hop within the cluster of currently connected nodes.
Grossglauser and Tse approach this kind of routing from
a slightly different point of view [5]. One major problem
with ad hoc networks is that due to interference of concur-
rent transmissions between nodes they scale badly. By only
doing local communications between neighbors and instead
relying on the movement of nodes to bring a message to its
destination, it is shown that this problem can be mitigated.
3. PROBABILISTIC ROUTING
Most users usually do not move around completely ran-
domly, and movement patterns are thus likely to be pre-
dictable, such that if a lo cation has been frequently visited
in the past, it is likely that it will be visited again in the
future. We would like to make use of this observation to
improve routing performance by doing probabilistic routing
.
A probabilistic metric called
delivery predictability, is estab-
lished at each node for each known destination indicating the
predicted chance of that node delivering a message to that
destination. When a node encounters another node, they ex-
change information about the delivery predictabilities they
have and update their own information accordingly. Based
on the delivery predictabilities, a decision is then made on
whether or not to forward a certain message to this node.
3.1 Delivery predictability calculation
The protocol relies on the delivery predictability metric,
P
∈ [0,
1], that should reflect the probability of encountering
a certain node. That metric should be used to support the
decision of whether or not to forward a message to a certain
no de.
Whenever a node is encountered, the metric should be
up dated according to (1), where
P
(a,b)
is the delivery pre-
dictability node a
has for node b, and P
init
∈
[0
, 1] is an
initialization constant. This ensures that nodes that are of-
ten encountered have a high delivery predictability.
P
(a,b
)
=
P
(
a,b
)
old
+
1
−
P
(
a,b)
old
× P
init
(1)
If a pair of nodes do not encounter each other in a while,
they are less likely to be good forwarders of messages to each
other, thus the delivery predictability values must
age
, being
reduced in the process. The aging equation is shown in (2),
where
γ
∈
[0
,
1) is the aging constant
, and
k is the number of
time units that have elapsed since the last time the metric
was aged. The time unit used can differ, and should be
defined based on the application and the expected delays in
the targeted network.
P
(
a,b
)
= P
(
a,b)
old
× γ
k
(2)
The delivery predictability also has a
transitive
property,
that is based on the observation that if node A frequently
encounters node B, and node B frequently encounters node
C, then node C probably is a good node to forward messages
destined for node A to. Equation 3 shows how this transi-
tivity affects the delivery predictability, where
β ∈ [0
, 1] is
a scaling constant that decides how large impact the transi-
tivity should have on the delivery predictability.
P
(
a,c)
=
P
(a,c)
old
+
1
−
P
(a,c)
old
× P
(
a,b)
×
P
(
b,c
)
× β (3)
3.2 Decision making
In traditional routing protocols, choosing where to for-
ward a message is usually a simple task; the message is sent
to the neighbor with the lowest cost path to the destina-
tion (usually meaning least number of hops). Normally the
message is also only sent to a single node since the relia-
bility of paths is relatively high. However, in the settings
we envision here, things are completely different. When a
message arrives there might not be a path to the destination
so the node have to keep the message for a while and each
time it encounters another node it must make a decision if
it should forward the message to that node or not. It might
also be sensible to forward a message to multiple nodes to
increase the probability that a message is really delivered to
its destination.
Unfortunately, these decisions are not trivial to make. In
some cases it might be reasonable to select a fixed thresh-
old and only give a message to nodes that have a delivery
predictability over that threshold for the destination of the
message. On the other hand, when encountering a node
with a low delivery predictability, it is not certain that a
no de with a higher metric will be encountered within reason-
able time, thus there can also be situations where we might
want to be less strict in deciding who to give messages to.
Further, there is the problem of deciding how many nodes
to give a certain message to. Distributing a message to a
large number of nodes increases the probability of deliver-
ing the message to its destination, but more resources are be
wasted. On the other hand, giving a message to only a few
no des (maybe just a single node) will use little resources,
but the probability of delivering a message might be lower,
and the incurred delay high. A number of trade-offs must
b e considered upon making the decision. Thus, we think the
strategy adopted for a particular message should be chosen
on a per-message basis to allow applications to cater for the
sp ecial needs of their data. One situation that must be dealt
with is the case where a system have just been started such
that no delivery predictability information is available (or if
mobility is such that prediction is difficult). In such a case,
the protocol should fall back to some (possibly randomized)
variant of epidemic routing.
4. FUTURE WORK
The use of acknowledgements to remove messages from
the network after delivery to their destination will be inves-
tigated. Neglecting to do this will result in unnecessary
transfers of messages, wasting valuable system resources.
Extensive simulations of the proposed protocol will b e run
to evaluate the protocol, and to investigate the impact of
different parameter settings and decision making strategies.
Further, it would also be good to analytically evaluate the
proto col to determine the suitability of the equations chosen
for calculation of the delivery predictability.
5. REFERENCES
[1] G. W. Boehlert, D. P. Costa, D. E. Crocker, P. Green,
T. O’Brien, S. Levitus, and B. J. L. Boeuf.
Autonomous pinniped environmental samplers; using
instrumented animals as oceanographic data collectors.
Journal of Atmospheric and Oceanic Technology,
18(11):1882–1893, 2001. 18 (11).
[2] V. G. Cerf, S. C. Burleigh, A. J. Hooke, L. Torgerson,
R. C. Durst, K. L. Scott, K. Fall, and H. S. Weiss.
Delay-tolerant network architecture. Internet Draft
draft-irtf-dtnrg-arch-02.txt, March 2003.
[3] X. Chen and A. L. Murphy. Enabling disconnected
transitive communication in mobile ad hoc networks. In
Proc. of Workshop on Principles of Mobile Computing,
colocated with PODC’01, Newport, RI (USA), pages
21–27, aug 2001.
[4] A. Doria, M. Ud´en, and D. P. Pandey. Providing
connectivity to the saami nomadic community. In
Proceedings of the 2nd International Conference on
Open Collaborative Design for Sustainable Innovation
(dyd 02), Bangalore, India, Dec 2002.
[5] M. Grossglauser and D. Tse. Mobility increases the
capacity of ad-hoc wireless networks. IEEE/ACM
Transactions on Networking
, 10(4), Aug 2002.
[6] A. Pentland, R. Fletcher, and A. A. Hasson. A road to
universal broadband connectivity. In Proceedings of the
2nd International Conference on Open Collaborative
Design for Sustainable Innovation (dyd 02), Bangalore,
India
, Dec 2002.
[7] G. E. Prescott, S. A. Smith, and K. Moe. Real-time
information system technology challenges for NASA’s
earth science enterprise. In Proceedings of The 20th
IEEE Real-Time Systems Symposium, Phoenix,
Arizona, Dec 1999.
[8] A. Vahdat and D. Becker. Epidemic routing for
partially connected ad hoc networks. Technical Report
CS-200006, Duke University, April 2000.