Ankur Khetrapal

Ad hoc Networks

There exist a number of routing protocols for ad hoc networks, each based on a specific algorithm, each with its own advantages and disadvantages. A survey of the currently existing approaches is of utmost importance in this field. 

1. ANT BASED DISTRIBUTED CERTIFICATE REVOCATION IN VEHICULAR AD HOC NETWORKS

Security is one of the most important issue with such networks. Together with my colleague Mr. Anvay Lonkar, we proposed a novel approach for temporal Distributed Certificate Revocation in vehicular ad hoc networks based upon the 'honesty' parameter determined using the Ant colony optimization' scheme.

Ant Colony Optimization

Fig.1 Ants using pheromone deposits to find food

On detection a malicious node, ants containing this information are sent out to all nodes within the communication range of the sending node. The ants travel based on a revocation table (RT) which is contained by each node. The RT (which looks similar to a distance vector routing table) holds information regarding each node and the amount of maliciousness related to that node. The ant updates this information and then travels in the direction of the node which, according to the RT, considers the accused node (B in fig 2) to be the most malicious.

 Ant Based distributed Certificate Revocation in VANETs

Fig.2 Distributing maliciousness information using ants

Publications

POSTER: A bio-inspired approach to distributed certificate revocation in Vehicular Ad hoc Networks.
Anvay Lonkar and Ankur Khetrapal
To appear at 26th Annual IEEE Conference on Computer Communications (IEEE Infocom) 2007.

^Top

 

2. DENIAL OF SERVICE ATTACKS IN AODV BASED MANETs

A Denial of Service attack may be launched by flooding the network with fake RREQs leading to the congestion of the network and loss of data for the genuine nodes. Assuming RREQ SEND LIMIT to be 10, we get the following scenario

RREQ Network Flooding

Fig.1 RREQ Network Flooding

We propose upper bounds on the RREQs sent and received by a node to avoid network clogging. The approach also takes into account the elimination of the malicious node (sending more than 10 RREQs per unit time) by monitoring the sent RREQs for a period known as STOP LIMIT TIME.  The evaluation of these bounds is still under investigation.

(UNDER CONSTRCUTION)

^Top