Efficient Secure Techniques for Grid Based Peer-to-Peer Detection System

Abstract

As distributed system architectures such as peer-to-peer, grid computing and MANET become more popular, there is an increasing need for robust and scalable mechanisms to establish trust between entities. The primary objective of Grid computing is to support the sharing of resources and service spanning across multiple administrative domains. Due to the inherently dynamic and multi organizational nature maintaining security of both users and resources is the challenging aspect of Grid. Grid uses internet as an infrastructure to build communication, with the fusion of web services and grid technologies further increases the security concerns for their complex nature. This paper takes a look at the vulnerability of Grid environment on denial of service attack. We found that deploying an efficient intrusion detection system to Grid can significantly improve its security and it can detect denial of service attack before it affects the victim. But due to the special characteristics and requirement of Grids, the existing traditional intrusion detection system cannot work properly in that environment. The focus of this paper is to investigate and design an anomaly detection system which can detect DoS and DDoS attack with high attack detection and low false alarm rate to achieve high performance. We have extensively surveyed the current literatures in this area; the main stress is put on feature selection for the Grid based anomaly detection system. An entropy based anomaly detection system has been proposed; also we have discussed the advantage of taking entropy as the metric. Finally the performance of the system has been analyzed using NS2 network simulator.

Keywords- MANET, Grid Computing, Anomaly Detection Technique.