JOURNAL OF COMMUNICATIONS AND NETWORKS, VOL. 15, NO.

1, FEBRUARY 2013

31

Enhanced OLSR for Defense against DOS Attack in

Ad Hoc Networks
Mohanapriya Marimuthu and Ilango Krishnamurthi
Abstract: Mobile ad hoc networks (MANET) refers to a network
designed for special applications for which it is difficult to use a
backbone network. In MANETs, applications are mostly involved
with sensitive and secret information. Since MANET assumes a
trusted environment for routing, security is a major issue. In this
paper we analyze the vulnerabilities of a pro-active routing protocol called optimized link state routing (OLSR) against a specific
type of denial-of-service (DOS) attack called node isolation attack.
Analyzing the attack, we propose a mechanism called enhanced
OLSR (EOLSR) protocol which is a trust based technique to secure the OLSR nodes against the attack. Our technique is capable of finding whether a node is advertising correct topology information or not by verifying its Hello packets, thus detecting node
isolation attacks. The experiment results show that our protocol is
able to achieve routing security with 45% increase in packet delivery ratio and 44% reduction in packet loss rate when compared to
standard OLSR under node isolation attack. Our technique is light
weight because it doesnt involve high computational complexity
for securing the network.
Index Terms: Ad hoc networks, denial-of-service (DOS) attack,
node isolation attack, optimized link state routing (OLSR), routing protocols.

I. INTRODUCTION
A mobile ad hoc networks (MANET) is a collection of mobile devices which are connected by wireless links without the
use of any fixed infrastructures or centralized access points. In
MANET, each node acts not only as a host but also as a router to
forward messages for other nodes that are not within the same
direct wireless transmission range. Each device in a MANET
is free to move independently in any direction, and will therefore change its links to other devices frequently. MANETs are
much more vulnerable and are susceptible to various kinds of
security attacks [1] because of its cooperating environment. In
the absence of a fixed infrastructure that establishes a line of defense by identifying and isolating non-trusted nodes, it is possible that the control messages generated by the routing protocols
are corrupted or compromised thus affecting the performance
of the network. Routing protocols in MANET can be classified
into two categories: reactive protocol and proactive protocol. In
proactive routing protocols, all nodes need to maintain a consistent view of the network topology. When a network topology changes, respective updates must be propagated throughout
the network to notify the change. In reactive routing protocols
Manuscript received December 29, 2011; approved for publication by JangWon Lee, Division II Editor, May 20, 2012.
The authors are with Sri Krishna College of Engineering and Technology,
Coimbatore, Tamilnadu, India, email: {mohanapriya, deancse}@skcet.ac.in.
Digital Object Identifier 10.1109/JCN.2013.000007

for mobile ad hoc networks, which are also called on-demand

routing protocols, routing paths are searched for, when needed.
Even though many research works had been carried out for routing attacks in MANET, most of it concentrated mainly on reactive routing protocols. Optimized link state routing (OLSR)
routing protocol which is a proactive routing protocol [2] offers promising performance in terms of bandwidth and traffic
overhead but it does not incorporate any security measures. As
a result, OLSR is vulnerable to various kinds of attacks [3], [4]
such as flooding attack, link withholding attack, replay attack,
denial-of-service (DOS) attack and colluding misrelay attack.
In this paper, we analyze a specific DOS attack called node isolation attack [5] and propose a solution for it. Node isolation
attack can be easily launched on OLSR after observing the network activity for a period of time. We propose a solution called
enhanced OLSR (EOLSR) that is based on verifying the hello
packets coming from the node before selecting it as a multipoint
relay (MPR) node for forwarding packets.
The remainder of this paper is organized as follows: Section II
describes an overview of OLSR protocol. In Section III an attack model for node isolation attack is discussed. In Section IV,
the previous works on OLSR MANET security are discussed. In
Section V, we propose a novel algorithm called EOLSR. In Section VI, we discuss the simulation model and results. In Section VII, brief conclusions are given.
II. OLSR OVERVIEW
Optimized link state routing (OLSR) [2], [5] is one of the
most important proactive routing protocols designed for
MANET. It employs periodic exchange of messages to maintain topology information of the network at each node. The key
concept of OLSR is the use of multipoint relay (MPR) to provide efficient flooding mechanism by reducing the number of
transmissions required. Each node selects a set of its neighbor
nodes as MPR. Only nodes selected as MPR nodes are responsible for advertising as well as forwarding topology information
into the network. Fig. 1 illustrates a node broadcast its messages
throughout the network using standard flooding where all neighbors relay message transmitted by the leftmost node and MPR
flooding where only MPR nodes relay the message. The protocol is best suitable for large and dense network as the technique
of MPRs works well in this context.
A node selects MPRs from among its one hop neighbors with
symmetric, i.e., bi-directional, links. Therefore, selecting the
route through MPRs automatically avoids the problems associated with data packet transfer over uni-directional links.
In OLSR protocol, two types of routing message are used,
namely, HELLO message and TC message. A HELLO message
is the message that is used for neighbor sensing and MPR se-

c 2013 KICS
1229-2370/13/$10.00 

32

JOURNAL OF COMMUNICATIONS AND NETWORKS, VOL. 15, NO. 1, FEBRUARY 2013

(a)

(b)

Fig. 1. Node broadcasting messages: (a) Regular flooding and (b) MPR
flooding.

lection. In OLSR, each node generates HELLO message periodically (every HELLO INTERVAL). A nodes HELLO message contains its own address and the list its 1-hop neighbors.
A TC message is the message that is used for route calculation.
In OLSR, each MPR node advertises TC message periodically
(every TC INTERVAL). A TC message contains the list of the
senders MPR selector. The protocol functioning of OLSR is as
follows:
A. Neighborhood Discovery
Neighborhood discovery is the process, whereby each router
discovers the routers which are in direct communication range
of itself (1-hop neighbors), and detects with which of these it can
establish bi-directional communication [3]. Each router sends
HELLOs, listing the identifiers of all the routers from which it
has recently received a HELLO, as well as the status of the
link.
B. MPR Flooding
MPR Flooding is the process whereby each router is able
to, efficiently, conduct network-wide broadcasts [3], [5]. Each
router designates, from among its bi-directional neighbors, a
subset (MPR set) such that a message transmitted by the router
and relayed by the MPR set is received by all its 2-hop neighbors. Nodes may express, in their HELO messages, their willingness to be selected as MPR, which is taken into consideration for the MPR calculation. Each node selects its MPR set
from among its 1-hop neighbors such that they can reach all its
2-hop neighbors.
Each node maintains information about the set of neighbors
that have selected it as an MPR. The set of nodes having selected
a given node as MPR is the MPR-selector-set of that node. A
node obtains this information from periodic HELLO messages
received from the neighbors. In OLSR, each MPR node must
forward the data and routing message coming from any of its
MPR selectors.
C. Link State Advertisement
Link state advertisement is the process whereby nodes are determining which link state information to advertise through the
network [3]. Each node must advertise, at least, all links between itself and its MPR-selector-set, in order to allow all nodes
to calculate shortest paths. Such link state advertisements are
carried in TCs, broadcast through the network using the MPR
flooding process. As a node selects MPRs only from among

Fig. 2. Topology perceived by node H before the attack.

its bi-directional neighbors, links advertised in TC are also bidirectional and routing paths calculated by OLSR contain only
bi-directional links. TCs are sent periodically, however certain
events may trigger non-periodic TCs.
III. NODE ISOLATION ATTACK
Node isolation attack is a kind of DOS attack launched by malicious nodes against OLSR protocol [5]. The goal of this attack
is to isolate a node from communicating with other nodes in the
network. More specifically, this attack prevents a victim node
from receiving data packets from other nodes in the network.
The idea of this attack is that attacker(s) prevent link information of a specific node or a group of nodes from being spread
to the whole network. Thus, other nodes who could not receive
link information of these target nodes will not be able to build
a route to these target nodes and hence will not be able to send
data to these nodes.
In this attack, attacker creates virtual links by sending fake
HELLO messages including the address list of target nodes 2hop neighbors, (the attacker can learn victims 2-hop neighbors
by analyzing TC message of its 1-hop neighbors). According to
OLSR protocol, the MPR selection is based on the maximum
coverage of any nodes 2-hop neighbors. So the target node will
select the attacker to be its only MPR node because it assumes
that it can reach all its 2-hop neighbors through the attacker itself. Thus, the only node that must forward and generate TC
messages for the target node is the attacking node. By dropping TC messages received from the target and not generating
TC messages for the target node, the attacker can prevent the
link information of target node from being disseminated to the
whole network. As a result, other nodes would not be able to receive link information of a target node and will conclude that a
target node does not exist in the network thus launching DOS
attack on the victim. Therefore, a target nodes address will
be removed from other nodes routing tables. Since in OLSR,
through HELLO messages each node can obtain only information about its 1-hop and 2-hop neighbors, other nodes that are
more than two hops away from a target node will not be able
to detect the existence of the target node. As a consequence, the
target node will be completely prevented from receiving data
packets from nodes that are three or more hops away from it.
In Fig. 2, node C is the attacking node, and node B is the target node. Instead of sending correct HELLO message that contain {B, F} in neighbor address list, the attacker sends a fake

MARIMUTHU AND KRISHNAMURTHI: ENHANCED OLSR FOR DEFENSE AGAINST...

Fig. 3. Topology perceived by node H after the attack.

HELLO message that contains {B, F, G, Z} which includes the

target nodes all 2-hop neighbors {F, G}and one non-existent
node {Z} [5]. According to the protocol, the target node B will
select the attacker C as its only MPR. Here node Z is announced
only by the attacker and not by any other neighbor nodes of the
victim. This is to improve the possibility of attacker being selected as a MPR. So the victim node B assumes that its 2-hop
neighbor node Z can be reached only via node C (attacker) and
all the other 2-hop neighbors also can be reached through node C
itself. So it selects node C as its only MPR. Being node Bs only
MPR, the attacker refuses to forward and generate TC message
for node B. Since the link information of node B is not propagated to the entire network, other nodes whose distance to node
B is more than two hops (e.g., node H) would not be able to build
route to node B. Fig. 3 shows the topology perceive by node H
after the node isolation attack [5]. As a result, other nodes would
not be able to send data to node B. Despite being in the network,
the target node B will be isolated from the network. An attacker
can launch this attack, as long as the target node is within its
transmission range.
IV. RELATED WORK
Recently, several cryptographic based techniques had been
contributed for securing OLSR [6][9].
In [6], a cryptographic based approach has been proposed
for protecting the network. This technique classifies the OLSR
nodes into either trusted or un-trusted nodes with an assumption
that trusted nodes are not compromised. It integrates a timestamp and a signature with each routing control message: The
signature is used to authenticate messages from trusted nodes,
and timestamps are used to prevent replay attacks. The drawback of this approach is that it does not deal with defense against
compromised trusted nodes. But in our scheme, the hello packets generated by the authenticated nodes are also verified that
enable us to detect the authenticated but compromised nodes.
In [7], the authors consider the compromise of trusted nodes.
It is assumed that a public key infrastructure (PKI) and a timestamp algorithm are in place. Apart from routing control packets
additionally this technique uses a message ADVSIG that contains time stamp and signature information. Each node maintains a table where information received in ADVSIGs is kept.
Based on this information, each node verifies the correctness
of the link state information in subsequent messages. In [8],
the authors employed distributed key management techniques

33

to prevent wormhole and message replay attacks. The technique

proposed in [9] uses signature and timestamp schemes to ensure authentication and protection against replay attacks. The
techniques in [7][9] imposes a large overhead to the network
in terms of additional traffic and signature computations which
results in high energy consumption at each node. Since our
scheme does not depend on any encryption and decryption techniques, it does not add any computational complexity at each
node.
A fully distributed certificate authority (CA) based on threshold cryptography is proposed in [10]. In this technique a node
can requests a certificate from any k nodes (shareholders) of the
network that are authorized CAs. Each of the share holders determine whether to serve the request based on whether the node
in question is well behaving. But this technique does not employ any monitoring system to determine the good behavior of
network nodes, so that it does not deal with compromised trusted
nodes.
In [5] and [11], the authors proposed a simple mechanism
to detect the link withholding and misrelay launched by MPR
nodes based on overhearing of traffic generated by 1-hop neighbors. But this technique requires promiscuous listening of neighbor nodes which result in energy drop at this node whereas we
do not use any neighbor monitoring approach.
In [12], Vilela et al. proposed a cooperative security scheme
using a complete path message (CPM) and rating table. This approach requires each node which receives a TC packet to send
CPM back to the TC source. Based on the path information from
the CPM, the TC source can detect the link spoofing attack. But
this technique incurs a large overhead in terms of additional traffic, since it requires all nodes which receive TC message to generate a CPM message. Since CPM contains complete path it traversed, the size of the message increases as network grows. Our
technique uses additionally three control messages which does
not pass the network more than 3-hops, so that it does not incurs
a large overhead in terms of traffic.
A formal approach to handle the MPR selection and defense
against the security attacks in OLSR is suggested in [13]. This
approach validates the routing table and the topology information using trust based reasoning. Hence, each node can verify
the validity of the received HELLO and TC messages simply by
correlating the information provided by these messages.
V. PROPOSED WORK
The proposed solution called EOLSR is an enhancement of
the basic OLSR routing protocol, which will be able to detect
the presence of malicious nodes in the network. Even though
our proposed solution is based on our previous work [16], we
have modified the approach of detecting the malicious node.
This is to eliminate any malicious node from giving the false
information about any normal node that wants to become MPR.
Our solution assumes that all the nodes are authenticated and
can participate in communication i.e., all nodes are authorized
nodes. In our approach, we assume the authentication mechanism [15] is applied in order to identify the exact origin of each
packet which prevents malicious node from sending forged reply packets using spoofed address. Our proposed work is based

34

JOURNAL OF COMMUNICATIONS AND NETWORKS, VOL. 15, NO. 1, FEBRUARY 2013

Table 1. As ONE_HOP.

HELLO message sender

B
C
D

2-hop neighbors of node A

E
F
G

Table 3. As ONE_HOP table after receiving Xs HELLO message.

HELLO message sender

B
C
D
X

2-hop neighbors of node A

E
F
G
E, F, G, Z

Table 2. Xs neighbors.

Originator
X

Neighbors
E, F, G, Z

Fig. 6. A send 2-hop request to B, C, and D then B, C, and D send

request to E, F, and G.
Fig. 4. OLSR node A selects B, C, and D as MPR.

Fig. 5. X advertises its neighbor nodes to A.

on verifying the correctness of the received Hello message from

a neighbor node before designating it as MPR for this node. Our
work on trust based analysis to detect the malicious node is inspired from [13]. In OLSR routing protocol, every node build
its routing table and learn the network topology based on the
HELLO and TC messages it receives from its neighbors. Here
we analyze the pattern of HELLO message of the node that advertise all the nodes 2-hop neighbors as its 1-hop neighbors
and verify whether that node is malicious or not. In our technique along with HELLO and TC messages, three other control packets called 2-hop request, 2-hop reply and node exist
query (NEQ) message are used to verify the information disseminated through the Hello messages. Also, each node maintains a
ONE_HOP table which consists of HELLO message sender (all
nodes reached by one hop from the given node) and the receiving nodes 2-hop neighbors announced in the message. In Fig. 4,
node A selects B, C, and D as MPR to broadcast packets to E, F,
and G and maintains ONE_HOP table shown in Table 1.
As shown in Fig. 5, if new node X sends HELLO message,
advertising all the target nodes 2-hop neighbors as its 1-hop
neighbors along with a new neighbor Z. Then, A add Xs 1-hop

information in As ONE_HOP table as shown in Table 3.

After including Xs information, as shown in Fig. 6 node A
send a 2-hop request to the nodes E, F, and G through its one
hop neighbors B, C, and D to verify whether node X is present
in the ONE_HOP table of nodes E, F, and G.
When the 2hop request is received by nodes E, F, and G, they
send a 2-hop reply to A that contains their one hop neighbors as
shown in Fig. 7. Instead of directly querying whether node X
is the one hop neighbor of nodes E, F, and G, in our algorithm,
the respective nodes have to send their one hop neighbor table in the 2- hop reply. This is to prevent any malicious node
from falsely claiming that the queried node X is not their one
hop neighbor. If node X is present in all the 2-hop reply it receives, A will select X as a MPR and broadcast all the data and
TC packets through node X to the other part of the network.
Otherwise, A add node X in Malicious node list and discard its
HELLO message. Node A then informs about the presence of
malicious node X to the network through HELLO and TC messages. The nodes on receiving the malicious node information
then delete the entire route involving that node from their routing table. It also ignores all the Hello and TC messages coming
from that node. In other case, if node X is actually be in the coverage area of E, F, and G nodes, then the target node A queries
about the existence of node Z in the network through the NEQ
message forwarded through its current MPR nodes. If any designated MPR node in the network confirms the existence of node
Z, then node X will be selected as MPR, otherwise, it will be
confirmed as a malicious node. Moreover, colluding attacks are
not possible because our technique doesnt employ any neighbor
node monitoring except explicit verification of the Hello messages it receives. The processing takes place at each node after
receiving a Hello packet is described in Algorithm 1. Algorithm
2 depicts the behavior of a node after receiving a 2-hop request.

MARIMUTHU AND KRISHNAMURTHI: ENHANCED OLSR FOR DEFENSE AGAINST...

Fig. 7. E, F, and G send 2-hop reply to A through B, C, and D.

Due to congestion in the network, or due to node mobility,

if any of the two-hop reply is lost, the source node after a time
out period resend the 2-hop request packet to the corresponding
node from which the reply is not received. Only if 2-hop reply
is received from all the 2 hop neighbors, and after verifying the
trustworthiness of the node in question, it will be selected as the
new MPR node. Otherwise, data forwarding will be continued
using the existing MPR nodes only.
Algorithm 1 HELLO reception.
1: if originator_node not in malicious list then
2:
Add the hello packet information in ONE_HOP table
3:
if 2-hop reply received then
4:
Verify the proof of correctness advertised by the
5:
hello packet sender node
6:
if correct then
7:
Select that node as its MPR if required
8:
else
9:
Move the hello packet sender to malicious list
10:
end if
11:
end if
12:
Inform the network about the presence of the attacker
13: end if

Algorithm 2 2-Hop request reception.

1: if 2-hop request received then
2:
Send a 2-hop reply containing all its one hop neighbors
3:
information
4: end if

VI. SIMULATION MODEL AND RESULTS

In this section, we present the performance evaluation on
our technique using extensive simulations conducted with the
network simulator GLOMOSIM [14]. We generated random
topologies with a maximum of 50 nodes over a rectangular field.
The terrain dimension is fixed as 750 1000 m. The maximum
transmission range of each node is 250 m. The duration of the

35

simulation is 600 s. Random waypoint model is used as the mobility model for each node. Node speed is varied from 2 m/s
to 25 m/s. The node pause time is varied from 0 second to 300
seconds. The default settings as in the specifications of OLSR
[2] were used for HELLO and TC messages. In our simulation,
we used 35% of malicious nodes out of the normal nodes to
launch the attack. The malicious nodes are chosen randomly
and also one of the neighbors of the nodes that are generating
the data traffic is chosen as malicious nodes. The traffic load is
simulated using 15 user datagram protocol-case based reasoning
(UDP-CBR) connections (30 nodes) generating traffic of 5 kB
UDP packets (data payload 512 Bytes) with an inter departure
time of 1 s. To eliminate the randomness in the result, for each
metric, simulation is done for ten different seed values with different random movement of nodes and the average value is taken
for the result. Also our approach is compared with another existing approach [5].
A. Performance Evaluation
We used the following metrics to evaluate the performance of
our proposed solution EOLSR against OLSR under attack and
the results obtained are shown in Figs. 810.
1. Packet delivery ratio: The ratio between the number of packets originated by the CBR sources of source nodes and the
number of packets received by the CBR sink at the destination node.
2. Packet loss rate: It is the number of data packets dropped by
the malicious nodes that are selected as MPR nodes.
3. Control packet overhead: This is the ratio of number of control packets generated to the data packet received.
Fig. 8 shows the packet delivery ratio in the presence of node
isolation attack. Here 1 to 5 malicious nodes are randomly selected to launch the attack. They select any one of the neighbor nodes as their victim and after analyzing the TC messages
and hello messages coming from that node; they create a fake
hello message containing all the 2-hop neighbors of the victim
and send it to the victim. Once the victim selects it as its MPR,
they drop all the data packets and TC packets coming from the
victim. As shown in the figure, The throughput achieved by
OLSR was approximately 25%, while the throughput achieved
in EOLSR under the same scenario was approximately 70%, increased by 45% i.e., EOLSR improved the throughput achieved
by OLSR under attack. When the number of attackers increases,
the throughput nearly drops to zero in normal OLSR whereas in
our scheme, even though the number of attackers increases, the
throughput achieved is more or less in steady state because the
MPR selection is made only after verifying the correctness and
trustworthiness of the node. Similarly, the throughput achieved
by the existing approach [5] is 65% which is 5% less than our
scheme. This is because the existing solution in [5] does not verify the trustworthiness of a node before selecting it as an MPR.
Instead after selecting the MPR node, it overhears the packet
forwarded by that MPR node and compares it with the packets send by itself to verify whether the MPR node is forwarding
the packets or not. Since the detection of malicious MPR node
is possible after the dropping of some TC and data packets by
the MPR node, the throughput achieved in [5] is lesser than our
scheme.

36

JOURNAL OF COMMUNICATIONS AND NETWORKS, VOL. 15, NO. 1, FEBRUARY 2013

&RQWUROSDFNHWV

3DFNHWORVV





(/265



6RORLQ>@






7KURXJKSXW

Fig. 9. Packet loss ratio.

Fig. 9 shows the number of packets dropped by the malicious

nodes in OLSR and EOLSR. The packet loss rate of OLSR under attack was approximately 74%, while the packet loss rate
of EOLSR was approximately 30%, reduced by 44%. Similarly
the packet loss rate of existing solution in [5] was approximately
37%, which was increased by 7% when compared to our solution. This is because the existing solution [5] is a detection
technique, which detects the attack after it has been launched
whereas our technique verifies the trustworthiness of a node before selecting it as an MPR. So packet drop ratio of our approach
is less when compared to the solution in [5]. Moreover, the existing approach in [5] employs promiscuous listening to overhear
packets forwarded by the MPR nodes which results in energy
dropping at the individual nodes and also this technique cannot
withstand colluding attackers. Whereas our technique does not
employ promiscuous listening so colluding attacks are not possible and also energy consumption at each node will be much
lesser than in [5].
The control packet ratio of EOLSR is 57% which is 11%
higher than the control packet ratio of the solution in [5] which
is 46%. This is because of the additional control packets introduced in EOLSR to prevent the node isolation attack by verifying MPR nodes.
VII. CONCLUSION
This paper proposes a solution for node isolation attack
launched against OLSR routing protocol. Here, we have discussed through an attack model, that it is easy for a malicious

Fig. 10. Control packet overhead.

Fig. 8. Packet delivery ratio.

1RRIDWWDFNHUV

1RRIDWWDFNHUV

1RRIDWWDFNHUV

node to launch the node isolation attack to isolate an OLSR

MANET node. This attack allows at least one attacker to prevent a specific node from receiving data packets from other
nodes that are more than two hops away. The proposed solution
called EOLSR, which is based on OLSR, uses a simple verification scheme of hello packets coming from neighbor nodes to
detect the malicious nodes in the network. The experiment results show that the percentage of packets received through our
proposed work is better than OLSR in presence of multiple attacker nodes. The simulation is done using GloMoSim and our
scheme is found to achieve routing security with 45% increase
in packet delivery ratio than standard OLSR and also achieves
44% reduction in packet loss rate than OLSR. Compared to
other related works, the proposed protocol has more merits; the
most important merit is that it achieves degradation in packet
loss rate without any computational complexity or promiscuous
listening. Moreover, cooperative or colluding attack cannot be
launched, because our technique doesnt employ any promiscuous listening of neighbor nodes for detecting the attackers.
REFERENCES
[1]

B. Kannhavong, H. Nakayama, and A. Jamalipour, A survey of routing attacks in mobile ad hoc networks, IEEE trans. Wireless Commun., vol. 14,
no. 5, pp. 8591, Oct. 2007.
[2] T. Clausen and P. Jacquet, IETF RFC3626: Optimized link state routing
protocol (OLSR), Experimental, 2003.
[3] T. Clausen and U.Herberg, Security issues in the optimized link state
routing protocol version 2 (OLSRv2), Int. J. Netw. Security Appl., 2010.
[4] B. Kannhavong, H. Nakayama and A. Jamalipour, A study of routing
attack in OLSR-based mobile ad hoc networks, Int. J. Commun. Syst.,
2007.
[5] B. Kannhavong, H. Nakayama, N. Kato, Y. Nemoto, and A. Jamalipour,
Analysis of the node isolation attack against OLSR-based mobile ad hoc
network, in Proc. ISCN, 2006, pp. 3035.
[6] D. Raffo, C. Adjih, T. Clausen, and P. Muhlethaler, Securing the OLSR
protocol, in Proc. Med-Hoc-Net, 2003.
[7] D. Raffo, C. Adjih, T. Clausen, and P. Muhlethaler, An advanced signature system for OLSR, in Proc. ACM SASN, 2004.
[8] D. Raffo, C. Adjih, T. Clausen, and P. Muhlethaler, Attacks against
OLSR: Distributed key management for security, in Proc. OLSR Interop
and Workshop, 2005.
[9] C. Adjih, T. Clausen, A. Laouiti, P. Muhlethaler, and D. Raffo, Securing the OLSR routing protocol with or without compromised nodes in the
network, HIPERCOM Project, INRIA Rocquencourt, Tech. Rep. INRIA
RR-5494, Feb. 2005.
[10] D. Dhillon, T. S. Randhawa, M. Wang, and L. Lamont, Implementing a
fully distributed certificate autorithy in an OLSR MANET, in Proc. IEEE
WCNC, 2004.
[11] D. Dhillon, J. Zhu, J. Richards, and T. Randhawa, Implementation &

MARIMUTHU AND KRISHNAMURTHI: ENHANCED OLSR FOR DEFENSE AGAINST...

[12]
[13]
[14]
[15]
[16]

evaluation of an IDS to safeguard OLSR integrity in MANETs, in Proc.

IWCMC, 2006.
A. J. P. Vilela and J. Barros, A cooperative security scheme for optimized
link state routing in mobile ad-hoc networks, in Proc. IST MWCS, 2006.
A. Adnane, R. de Sousa, C. Bidan, and L. M, Analysis of the implicit
trust within the OLSR protocol, in Proc. IFIP, 2007.
X. Zeng, R. Bagrodia, and M. Gerla, GloMoSim: A library for parallel
simulation of large-scale wireless networks, in Proc. PADS, 1998.
D. Raffo, Security schemes fo the OLSR protocol for ad hoc networks,
Ph.D. dissertation, Univ. Paris, 2005
M. Mohanapriya and S. Urmila, A novel technique for defending routing
attacks in OLSR MANET, in Proc. IEEE ICCIC, 2010.

Mohanapriya Marimuthu is currently a Ph.D. candidate in the Computer Science Department, at Anna
university of Technology, Coimbatore. She received
her B.E. degree in Computer Science and Engineering
from Bharathiar University in 2002 and M.E. degree
in Computer Science and Engineering from Anna University in 2004. She is working as an Assistant Professor in the Department of Computer Science and Engineering at Sri Krishna College of Engineering and
Technology, Coimbatore. Her research interests are
Network Security and Ad hoc Networks.

37

Ilango Krishnamurthi is a Professor Dean at the Department of Computer Science and Engineering of
Sri Krishna College of Engineering and Technology,
Coimbatore. Ilango received his Ph.D. degree in Computer Science and Engineering from the Indian Institute of Technology, Chennai. He graduated from
BITS, Pilani and received M.S. degree from Iowa
State University, USA both in the fields of Computer
Science & Engineering. He spent 15 years at NIT,
Trichirapalli in the capacities of Lecturer, Assistant
Professor and coordinator of the part time B. Tech programme. Since July 2006, he is with SKCET as a Professor & Head. Since June
2008, he has been promoted as Dean, CSE Department. He has published 20
research papers in National, International journals and conferences. His current
research interests are in the areas of semantic web, mobile computing, and data
mining.