(IJCNS) International Journal of Computer and Network Security, 71
Vol. 2, No. 10, 2010
Optimization of DTN routing protocols by using
forwarding strategy (TSMF) and queuing drop
policy (DLA)
Sulma Rashid1, Qaisar Ayub1, M. Soperi Mohd Zahid 1 , A.Hanan. Abdullah1
1
Universiti Teknologi Malaysia (UTM),
Faculty of Computer Science & Information System,
Department of Computer System & Communication Skudai - Johor, 81310, Malaysia
sulmaqaiser@gmail.com , sheikhqaisar@gmail.com , soperi@utm.my , hanan@utm.my
Abstract: Delay tolerant Networks (DTN) are wireless networks
where disentanglement may occur repeatedly. In order to attain
the delivery probability in DTN, researchers have proposed the
use of store-carry-forward paradigm where a node may
accumulate messages in its buffer and carry it’s for long period
of time, until a forwarding opportunity arises. In this context,
multiple copies scheme gets popularity which floods multiple
messages to increase the delivery probability. This combination
leads protocol to store message for long time at node buffer and
less time duration of contact limited the message forwarding
and increase overhead of bandwidth. Thus a effective scheduling
for forwarding and drop polices are extremely important to
make decision in which order message is forwarded from queue
when limited time of transmission is available and decide which
drop policy will be use to overcome the full node buffer when a
new message received. In this paper, we propose a combination
of drop policy DLA and forwarding scheduling policy TSMF
called as DLTs that optimizes the DTN routing protocols in term
of overhead and boost the delivery probability and buffer time
average. We evaluate their efficiency and tradeoffs, through
simulation and prove that DLTs is performing better than
FIFO.
Keywords: Store and carry networks. Routing protocols, drop
policies, forwarding policies.
1. Introduction
In conventional routing schemes it is necessary to launch
end-to-end path from source to destination previous to the
transmission of data. Hence most of wireless applications
such as sensor networks for ecological monitoring [16],
ocean sensor networks [18],[17] biological sensor networks
[11] and vehicular networks [19], [20] due to highly
unstable path which may change or break while being
discovered.
Disruption tolerant networks (DTNs) enable the
transmission of data by using intermittently connected
mobile nodes. DTN as [9],[6] refers as work by using store-
carry-forward paradigm, where each node in the network
store the message in buffer, carries the message while
moving and forward when it encounter with another node.
Due to long delays, frequent disruptions between
intemitteltly nodes and limited resources, routing in DTN
[10] has become the prime issue.
Based on the idea of message forwarding, routing
schemes for DTN can be divided in to two major categories,
single copy and multi copy [12]. In single copy schemes
only one copy of message exists in the network, which is
forwarded along single path [5] for example first contact
[5], direct delivery [5]. While in multi copy schemes more
then one copy of same message are forwarded to multiple
paths for example Epidemic router[8], spray&wait[14]
,prophet[1], MaxProp [20] ,probabilistic forwarding[3].
As proved by [13] that multi copy policy has high impact
on message delivery and robustness at the cost of more
bandwidth, energy and memory usage. However an
important issue which was not investigated in the previous
work is the use of an efficient buffer management strategies
and message forwarding polices. A recent work [1], [21],[7]
has proposed few forwarding and buffer management
strategies.
In this paper we combine the buffer management strategy
DLA with forwarding queue mode TSMF to optimize the
performance of DTN routing protocols in term of delivery
probability, overhead ratio and buffer time averages. This
technique is called as DLTs.
The remaining paper is prepared as follows .Section 2
elaborates the existing buffer and forwarding polices.
Section 3 is about routing protocols for optimization and
Section 4 is performance metrics, Section 5 is approach and
simulation results simulates in section 6 by a conclusion at
section 7.
2. Existing drop and forwarding policies
When nodes under DTN resource constrained (buffer)
network communicates the congestion arise frequently.
Hence the issue is which message from congested buffer will
be dropped to continue transmission.
2.1 Queuing drop policies
Following are queuing drop polices for messages discarded
order when a new message received at the full buffer node.
2.1.1 Drop Random (DR)
The selection of message to be dropped is in random order
2.1.2 Drop–Least-Recently-Received (DLR)
The message with the long stay time in buffer will be
dropped. The idea is that the packet with in buffer for long
time has less probability to be passed to other nodes.
72 (IJCNS) International Journal of Computer and Network Security,
Vol. 2, No. 10, 2010

2.2.4 GRTRSort
2.1.3 Drop -Oldest (DOA) “GRTRSort looks at difference P (B-D) – P(A-D) values for each
The message with the shorted remaining life time (TTL) in message between the nodes and forward the message only if
network is dropped. The idea of dropping such packet is that P(B-D)>P(A-D).” [1]
if packet TTL is small, it is in the network for long time and
thus has high probability to be already delivered.
2.2.5 GRTRMax
2.1.4 DL-Drop last (DL) “Select messages in descending order of P (B-D) forward the
It drops the newly received message. message only if P (B-D)> P (A-D).” [1]
2.2.6 TSMF
2.1.5 Drop front (DF)
The message that enters first in the queue is dropped first. In TSMF [7] the forwarding queue the message with small
size is placed on top of queue.
2.1.6 N-Drop
In N-Dropt [2], the message that does N number of 3. Routing protocol for optimization
forwarding will be selected to drop.
2.1. Epidemic
2.1.7 MOFO In epidemic routing[8] application messages are swamped to
The message that has been forwarded to maximum number the relay nodes called carriers , where carrier nodes though
of times will be dropped first. [1] moving comes in contact with another related proton of
network , it forward the message to further island of nodes.
2.1.8 MOPR This redundancy of forwarding formulates sure about the
Each message in node is associated with a forwarding delivery of message to its destination.
predictability FP, initially assigned to 0. When the message
is forwarded the FP value is updated and the message with 2.2. Spray&wait (binary)
highest FP value will be dropped first. [1]
Spray&wait (binary) start with N number of copies. While
2.1.9 SHLI moving when it encounter with a node such that N=0, it
The message having smallest TTL will be selected to drop. spread the (N/2) message copies to new node while keep
[1] (N/2) for itself. When it left with one copy (N=1) it perform
direct transmission. Spray&wait combines the speed of
2.1.10 LEPR epidemic router with the simplicity of direct transmission.
“Since the node is least likely to deliver a message for which
it has a low P-value, Drop the message for which the node 2.3. Direct delivery
has the lowest P value.” [1] The source node transmits the message to other node only
when other node is its destination. Direct delivery[5] can be
2.1.11 Drop Largest (DLA) considered a hybrid technique as one can include it in
In Drop Largest (DLA) [21] large size message will be flooding where Direct Delivery always select the direct path
selected in order to drop. between source and destination.
2.2 Forwarding policies
2.4. First Contact
2.2.1 First In First Out (FIFO) In first contact [5] a message is forwarded along a single
path by selecting the node randomly from available
In FIFO queue mode all messages are arranged according to connections. If connections do not exist the nodes waits and
arrival time and the message which has oldest arrival time transmit the message to first available contact.
will be transmitted first.

2.2.2 Random queue mode (RND) 2.5. Prophet router

The message is randomly selected for the transmission.
2.2.3 GRTR
“Assume A, B are nodes that meet while the destination is
D, P(X, Y) denote the delivery predictability that a node X has
for Destination Y. GRTR forward the message to node only
if P (B-D) >P (A-D)” [1].
The routing protocols perform variants of flooding.
Epidemic [8] replicates messages to all encountered peers,
while Prophet [3] tries to estimate which node has the
highest “likelihood” of being able to deliver a message to
the final destination based on node encounter history.
 (IJCNS) International Journal of Computer and Network Security, 73
Vol. 2, No. 10, 2010

4. Performance oriented metrics

4.1 Delivery probability

It is ratio of the message delivered to the destination to those
generated by the sources. High probability means that more
messages are delivered to the destination.
Number of Nodes 126
Movement model 1 Random Waypoint Figure 3. DLTs (Drop Largest-Transmit smallest)
Movement model 2 Map rout movement
Number of groups 06
Figure 2 depicts the DLTS mechanism. As buffer at Node
A is congested, DLTS will drop the large size message
Buffer size 5MB
which is M190. When B starts transmission it will first sort
Transmission range 10M
out the forwarding queue to pop up small sizes messages.
Transmission speed 250 K
We can see that with available transmission time M101,
Message creation 25-35 (seconds) M920, M126, M137, M115 will be transmitted which
interval
increases the delivery probability.
Table1: Simulation setup
6. Simulation and results
4.2 Overhead ratio
It is the negation of number of messages transmitted to relay
to the number of message delivered. Low value of overhead
In the following section we have examined routing protocols
means less processing required delivering the relayed
(Section 5), with exiting (FIFO-FIFO) and proposed
messages.
(DLTs).
All the experiments were carrying out using ONE
4.3 Buffer time average Simulator. The ONE Simulator is a discrete event simulator
It is Sum of time spend by all message(s) in buffer divided written in Java. The main aim of simulator is to apply DTN
by message delivered. (store-carry-forward) of message for lengthy time, where the
likelihood of disconnections and failures increased
5. Approach DLTs
Consider a scenario of two nodes A and B with buffer size
1000kb,Mid is message identification and must be unique
for each message, SM is size of message, AT represent the
arrival time while TT is transmission time for message.
ATT is available transmission
time.

Figure 4. Delivery probability FIFO-FIFO and DLTs

The above figure 3 depicts the comparison of DLTs and

Figure 2. FIFO-FIFO queue mode
Assume B wants to transmit the message to Node A,
according to forwarding queuing policy (FIFO) M810 will
be selected. We can see that the buffer at node A is
congested, according to Drop policy (FIFO) messages
(M110, M563, M120, M111, M150.) at node A will be
dropped until the space becomes available for M810.With
available ATT only M810 with TT of 5s will be transmitted.
FIFO queue and forwarded mode with respect to delivery
probability. We can observe that epidemic, spray&wait and
prophet router have high delivery probability because
frequency of node encounter is high resulting more
congestion where transmitting smallest message raise the
delivery of messages. FC and DD are single copy cases and
chance of congestion is less as in multi copy schemes result
less delivery as compared to other routers but still DLTs
increases the delivery than existing queue policy. Moreover,
DD router passes the messages to that node that are
destination. However in all the configuration of routers,
message delivery probability of DLTs is improved then
FIFO.
74
Figure 5. Buffer time average FIFO-FIFO and DLTS
Figure 4 observes buffer time averages with DLTs and
FIFO. It can be clearly seen that DLTs has high value of
buffer time average with all router. Buffer time occupancy is
very scared resource in the DTN where store and carry
paradigms are used. As expected, in this architecture buffer
should retain the message as along as it can, so that delivery
of message should increase and eliminate the drop ratio of
messages. DLTs improved the buffer time occupancy for all
router in multi and single copy routers and increase the
delivery ratio as in figure 3.
Figure 6. Overhead ratio FIFO-FIFO and DLTs
Figure 5 represents the impact of DLTs and FIFO with
respect to overhead ratio. We are able to see clearly that
overhead ratio with DLTS is decreases in all routers
irrespective of multi copy or single copy approaches of
routers., DD due to direct transmission overhead is zero so
we excluded that case for both algorithms, while Epidemic,
spray&wait, Prophet, FC overhead is reduced to
considerable level.
7. Conclusion and future work
In this paper, we propose a combination of drop policy DLA
and forwarding scheduling policy TSMF called as DLTs
that optimizes the DTN routing protocols in term of
overhead and boost the delivery probability and buffer time
average. We evaluate their efficiency and tradeoffs, through
simulation and prove that DLTs is performing better than
FIFO.
The results presented in this paper can be used as a
starting point for further studies in this research field, and
give helpful guidelines for future DTN protocol design.
References
[1] A.indgren and K. S. Phanse, “Evaluation of queuing
policies and forwarding strategies for routing in
intermittently connected networks,”in Proc. of IEEE
COMSWARE, pp. 1-10, Jan. 2006.
(IJCNS) International Journal of Computer and Network Security,
Vol. 2, No. 10, 2010
[2] Yun Li, Ling Zhao ,Zhanjun Liu,Qilie Liu.” N-Drop
Congestion Control strategy under Epidemic Routing in
DTN.” Research center for wireless information
networks,chongqing university of posts &
Telecommunications ,chongqing 400065,china, pp.
457-460, 2009
[3] A. Lindgren, A. Doria, and O. Schelen, “Probabilistic
routing in intermittently connected networks,”
SIGMOBILE Mobile Computing and Communication
Review, vol. 7, no. 3, 2003.pp 19-20.
[4] KERÄNEN, A., AND OTT, J. Increasing Reality for
DTN Protocol Simulations. Tech. rep., Helsinki
University of Technology, Networking Laboratory, July
2007.
[5] T. Spyropoulos, K. Psounis, and C. Raghavendra A, C.
S. “Single-copy routing in intermittently connected
mobile networks,” IEEE/ACM Transactions on
Networking (TON), vol. 16, pp. 63-76, Feb. 2008.
[6] Tara Small and Zygmunt Haas.” The shared wireless
infestation model - a new ad hoc networking paradigm
(or where there is a whale, there is a way)”, In
Proceedings of The Fourth ACM International
Symposium on Mobile Ad Hoc Networking and
Computing (MobiHoc 2003), pages 233.244, June 2003.
[7] Qaisar Ayub, Sulma Rashid and Dr.Mohd Soperi Mohd
Zahid. Article: Optimization of Epidemic router by new
forwarding queue mode TSMF. International Journal of
Computer Applications 7(11):5–8, October 2010.
Published By Foundation of Computer Science.
[8] Vahdat and D. Becker. Epidemic Routing for Partially-
connected Ad hoc Networks. Technical Report CS-
2000-06, Duke University, July 2000.
[9] SCOTT, J., HUI, P., CROWCROFT, J., AND DIOT, C.
Haggle: Networking Architecture Designed Around
Mobile Users. In Proceedings of IFIP WONS (2006).
[10] FALL, K. A Delay-Tolerant Network Architecture for
Challenged Internets. In Proc. of ACM SIGCOMM
(2003).
[11] Z. J. Haas and T. Small. A New Networking Model for
Biological Applications of Ad Hoc Sensor Networks.
IEEE/ACM Transactions on Networking, 14, No. 1:27–
40, 2006.
[12] T. Spyropoulos, K. Psounis, and C. Raghavendra.
Efficient Routing in Intermittently Connected Mobile
Networks: The Multi-copy Case. In ACM/IEEE
Transactions on Networking, 2007.
[13] T. Small and Z. J. Haas. Resource and performance
tradeoffs in delay-tolerant wireless networks. In
SIGCOMM Workshop on Delay Tolerant Networking
(WDTN), 2005.
[14] T. Spyropoulos, K. Psounis, and C. S. Raghavendra.”
Spray and wait: an efficient routing scheme for
intermittently connected mobile networks”, In
SIGCOMM Workshop on Delay Tolerant Networking
(WDTN), 2005
[15] J. Scott, P. Hui, J. Crowcroft, and C. Diot, “Haggle: A
networking architecture designed around mobile users,”
in Proc. IFIP Conf. Wireless On-Demand Network
Systems and Services (WONS), 2006.
 (IJCNS) International Journal of Computer and Network Security, 75
Vol. 2, No. 10, 2010

[16] P. Zhang, C. M. Sadler, S. A. Lyon, and M. Martonosi.

Hardware Design Experiences in ZebraNet. In Proc.
ACM SenSys, pages 227–238, Nov. 2004.
[17] Maffei, K. Fall, and D. Chayes. Ocean Instrument
Internet. In Proc. AGU Ocean Sciences Conf., Feb
2006.
[18] J. Partan, J. Kurose, and B. N. Levine. A Survey of
Practical Issues in Underwater Networks. In Proc.
ACMWUWNet, pages 17–24, Sept. 2006.
[19] J. Ott and D. Kutscher. A Disconnection-Tolerant
Transport for Drive-thru Internet Environments. In
Proc. IEEE INFOCOM, pages 1849–1862, Mar. 2005.
[20] J. Burgess, B. Gallagher, D. Jensen, and B. N. Levine.
MaxProp: Routing for Vehicle-Based Disruption-
Tolerant Networks. In Proc. IEEE Infocom, April 2006.
[21] Sulma Rashid,Qaisar Ayub,”Effective buffer
management policy DLA for DTN routing Protocals
under congetion”, International Journal of Computer
and Network Security,Vol 2,NO 9,Sep 2010. pp .118-
121