Security in Wireless Ad Hoc Networks-free thesis

In wireless ad hoc networks, since all nodes share the common communication medium, attackers can easily launch a variety of attacks ranging from passive eavesdropping from active interfering. For example, a simple and straight-forward attack is jamming attack, where attackers can disrupt the other nodes’ normal communications by introducing interferences. Various schemes have been proposed to handle jamming attack in the literature. One way to handle jamming attack is to design robust physical layer technologies, such as spread spectrum, which are resistant to RF jamming [7, 7, 77]. By using some spreading codes only known to the communicating peers, nodes have created a secret channel among them. Recently, several new approaches have also been proposed to handle jamming attacks in a more e±cient way, such as those proposed in [0, 55, 8, 8, 8]. In this dissertation we will not focus on jamming attack, and will assume that some 8existing schemes, such as those proposed in [7,8], have been employed to address such attacks.

Besides physical layer attacks, attackers can also try to interrupt the normal
Medium Access Control (MAC) layer behaviors, such as described in . In
this dissertation, we will not focus on speci¯c types of MAC layer attacks. Instead,
we will focus on some general attack models which have incorporated the e®ects
of MAC misbehavior. In [], Cagalj et. al. have also studied the possible MAC
layer sel¯sh and cheating behaviors in wireless CSMA/CA networks.
To secure wireless ad hoc network, we can ¯rst try to prevent attackers from
entering the networks. This can be achieved by applying necessary access con-
trol and authentication , such as secure key distribution and
secure neighbor discovery , etc. For example, Zhou and Haas investigated
distributed certi¯cate authorities in ad hoc networks using threshold cryptogra-
phy [0]. Hubaux et al. developed the idea of self-organized public-key infras-
tructure similar to PGP in the sense that public-key certi¯cates are issued by
the users [,]. The di®erence with PGP is that in their system, certi¯cates are
stored and distributed by the users. Capkun et al. have also discussed how to build
security associations with the help of mobility in mobile ad hoc networks [].
Since in ad hoc network nodes relies on each other to forward packets, routing
has become one of the most active research topics during the last decade, and
various routing protocols have been proposed, such as DSR [7, 8], AODV [70],
OLSR [5], and TBRPF [6]. Some performance comparison among various rout-
ing protocols have been demonstrated in []. However, in order to work properly,
these protocols need trusted working environments, while in reality the environ-
ments is usually adversarial. Some examples of routing attacks are: black hole,
9gray hole, wormhole, rushing attack, and frame-up [{7]. For example, the at-
tackers can create a wormhole through collusion in the network to short circuit
the normal °ow of routing packets [5], or can apply rushing attack to disseminate
route request quickly through the network [6]. By creating a wormhole or apply-
ing rushing attacks, the attackers can prevent good routes from being discovered,
and increase their chance of being on discovered routes. Once an attacker is on a
certain route, it can create a black hole by dropping all the packets passing through
it, or create a gray hole by selectively dropping some packets passing through it.
If the protocols have the mechanism to track malicious behavior, an attacker can
also try to frame up good nodes. In addition, an attacker can modify the packets
passing through it, which has similar e®ects as dropping packets, but a little bit
more severe because more network resources will be wasted when the following
nodes on this route continue forwarding this corrupted packet.

In the literature, various secure routing protocols have been proposed, such as For example, Papadimitratos and Haas [65] have proposed a secure routing protocol for mobile ad hoc networks that guarantees the discovery of correct connectivity information over an unknown network in the presence of malicious nodes. Sanzgiri et al [76] have considered a scenario that nodes authenticate routing information coming from their neighbors while not all the nodes on the route will be authenticated by the sender and the receiver. Hu, Perrig and Johnson [] have proposed Ariadne, a secure on-demand ad hoc network routing protocol, which can prevent attackers or compromised nodes from tampering with uncompromised routes that (only) consist of uncompromised nodes. In [5, 6], they have described how to defend against rushing attacks through secure neighbor discovery and how to apply packet leashes 0to defend against wormhole attacks. Later, Capkun and Hubaux have investigated secure routing in ad hoc networks in which security associations exist only between a subset of all pairs of nodes [].
However, most of the existing secure routing schemes have focused on prevent-
ing illegitimate nodes from being on the routes. In other words, they have focused
on defending against outside attackers. In ad hoc networks, due to the loose ac-
cess control and weak physical protection, insider attackers can be very common.
In the literature, very few schemes have considered insider attacks. Among them
the most representative one is proposed by Marti et al [58]. They focused on
the case that nodes agree to forward packets but fail to do so, and proposed two
tools that can be applied upon source routing protocols: watchdog and pathrater.
Speci¯cally, each node launches a \watchdog” to monitor its neighbors’ packet
forwarding activities and to make sure that these neighbors have forwarded the
packets according to its requests. Pathrater will be used to prevent misbehaving
nodes from being on the selected routes when performing route discovery. How-
ever, this system su®ers some problems, and many attacks can cause a malicious
behavior not being detected, such as ambiguous collisions, receiver collisions, lim-
ited transmission power, collusion, and partial dropping. Meanwhile, due to noise
and possible attacks, good nodes can also be easily marked as malicious. In other
words, the proposed scheme may su®er both high false alarm ratio and high miss
detect ratio when performing attacker detection.
Following [58], CONFIDANT was proposed to detect and isolate misbehav-
ing node and thus make it unattractive to deny cooperation [6]. Comparing to
the schemes proposed in [58], CONFIDANT allows the reputation to propagate
throughout the network. However, since the scheme still rely on watchdog, they
also su®er the same types of problems as [58]. Furthermore, once reputation is
allowed to propagate, attackers can also collude to frame up or blackmail other
nodes [5]. Besides [6,58], Ning and Sun have also provided a case study of insider
attacks against mobile ad hoc routing protocols by focusing on AODV.
Security in ad hoc networks has also been addressed from the intrusion detection
point of view, such as [0,,96,97]. In these works, the authors have discussed how
to apply intrusion detection techniques to secure wireless ad hoc networks. They
examined the vulnerabilities of a wireless ad hoc network, then introduced multi-
layer integrated intrusion detection and response mechanisms. Such techniques can
also be used to deal with insider attacks. However, in their work they have not
described speci¯c mechanisms to secure ad hoc networks. Furthermore, no formal
analysis of securing ad hoc networks against insider attacks has been provided.
Besides the above mentioned attacks, attackers can also launch various types
of other attacks to disrupt the normal communications. For example, one severe
attack is Sybil attack , where an attacker can behave as if it were a larger
number of nodes, for example by impersonating other nodes or simply by claiming
false identities. In [] the authors have also studied JellyFish attacks. Another
types of severe attacks, which will be thoroughly studied in this dissertation, is
injecting tra±c attacks, that is, the attackers will try to inject an overwhelming
amount of tra±c into the network to consume valuable network resources and de-
grade the network performance. Section . describes the attack model considered
in this dissertation.

Free download research paper