Abstract: This paper addresses survey of work done on relevance feedback in the past few years. Till date, various techniques have been proposed for relevance feedback. Some techniques used Bayesian classifier, while others are log based and Neuro-fuzzy based. Some of them use human computer interaction for providing feedback. However, existing methods require a large number of iterations and also it is difficult to converge the query towards user intention. Hence, in this paper we present some of the recent techniques that overcome the above disadvantages and their comparative analysis.

• Publication Date: 28-Apr-2013
• DOI: 10.15224/978-981-07-6260-5-54
• Views: 0
• Downloads: 0

Abstract: The use of wireless links makes a Mobile Ad hoc Network(MANET) susceptible to malicious attacks, ranging from passive eavesdropping to active interference. In wired networks an attacker has to pass through a set of firewalls and gateways to access the network. Whereas MANETS does not have firewalls or gateways so attacks can take place from all directions. Every node in a MANET is an autonomous unit in itself and free to move independently. So any node without adequate protection is very much prone to be captured or compromised. Intrusion prevention techniques like encryption and authentication can reduce the risk of intrusion but cannot completely eliminate them so a second level of defense is needed. We propose a new quantitative method of intrusion detection system to detect intrusion in MANETS with mobile nodes. The proposed method is a behavioral anomaly based system which is dynamic, scalable, configurable and robust. For simulating the proposed system we use AODV routing protoco

• Publication Date: 28-Apr-2013
• DOI: 10.15224/978-981-07-6260-5-59
• Views: 0
• Downloads: 0