Table 1 Some Attacks onthe Protocol Stack 3.1. Adversary ModelThere are twotypes of attacker in a network which are called insider and outsider attackers. Fig.1 Insider and Outsider Attacks inMANETs An insider attacker is anauthorized and a part of the routing mechanism on MANETs.An outsider attacker is not a legitimateuser of the network. Routing algorithms are typically distributed and cooperativein nature and affect the whole system.
On the other hand an insider MANET nodecan disrupt the network communications intentionally; there might be otherreasons for its apparent misbehaviours. A node can be failed, unable to perform its function for some reason, such asrunning out of battery, or collusions in the network. The threat of failednodes is particularly serious if they are needed as part of an emergency/secureroute. Their failure can even result in partitioning of the network, preventingsome nodes from communicating with other nodes in the network. A selfish node can also misbehave to preserve itsresources. Selfish nodes avail themselves of the services of the other nodes,but do not reciprocate.
In this paper, we mainly concentrate on attacks carriedout by malicious nodes who intentionally aim to disrupt the networkcommunication.We should also consider themisuse goals of attackers. 3.1.
2Disruption of the network Attackers do not follow the specifications ofrouting protocols in routing attack but aim to disrupt the networkcommunication in the following ways:1. Node Isolation: isolating a nodeor some nodes(s) from communicating with other nodes in the network,partitioning the network, etc.2. Route Disruption: Creating routingloops, modifying existing routes, and causing the packets to be forwarded alonga route that is not optimal, non-existent, or otherwise erroneous.3. ResourceConsumption:Itreduces network performance by consuming network bandwidth or node resources,etc.3.1.
3 PerformanceBottleneck of Attacker The main factors affecting theperformance of an attack are identified below.Deploymentcapability:Adversarydistribution may range from a single node to a pervasive carpet of smartcounter-dust, with a consequent variation in attack capabilities. This sort ofdistinction may affect the ability to eavesdrop, jamming the network effectively,and to escape destruction (e.g. a single powerful jammer can easily be taken out;distributed jamming is harder to extinguish).Locationcontrol:The location of adversary nodes may have a clear impact on what the adversarycan do.
An adversary may be restricted to placing attack nodes at thegeographical boundary of an enemy network (but may otherwise choose the preciselocations), may plant specific nodes or may have the ability post facto tocreate a pervasive carpet of smart dust. Computationalpower:Thisis obvious affects the ability of an attacker to compromise a network. Suchpower need not be localised to the attached network – eavesdropped traffic canbe relayed back to high performance super-computing networks for analysis.Mobility: Mobilitygenerally brings an increase in power. On the other hand, mobility may preventan attacker from continually targeting one specific victim. For example, a nodeon the move might not receive all falsified routing packets initiated by theattacker.
The impact of mobility on detection is a complex matter.Degreeof physical access (including node capture ability andability to carry out physical deconstruction) given the agile nature of MANETsdetermining an applicable adversary model is difficult. However, systems can beevaluated against a range of representative threat models. 4.
TYPES OFATTACKS IN MANET The nature of attacksis of two types’ active attacks and passive attacks. Active attack is an attackwhen misbehaving node has to bear some energy costs in order to perform thethreat. These attacks results unauthorised state changes in the network such asdenial of service, modification of packets, and the like. These attacks aregenerally launched by users or nodes with authorisation to operate within the