A Hybrid Database Intrusion Detection Algorithm Using Swarm Intelligence and Radial Basis Function Network

Author Name(s): Anitarani Brahma, Suvasini Panigrahi, Jayasmita Mahapatra
Author Email: brahmaanita00@gmail.com

Abstract

Recently, all over the globe, intrusion detection especially in Database gaining lots of attentions from researchers as it serves a security support system to existing security mechanism. This paper presents a hybrid intrusion detection algorithm in Database System by making use of swarm intelligence and radial basis function network. Both these techniques are combined to produce better detection results. The performance of the proposed approach is investigated with synthetic dataset and is compared with other machine learning technique and found to be significantly better.

Introduction

Due to advancement of Information Technology in the field of e-commerce, online banking , online marketing, the existing database expose to the outside world which contains many sensitive information that should not be disclosed. Hence, intensive care is needed to the existing security mechanism and Database Intrusion Detection System (DIDS) serves additional protection to the Database. The main objective of DIDS is to discriminate between normal and malicious transactions made to the database. Many statistical mechanism already been applied by researchers to develop an accurate DIDS model [1,2,3,4,5,6]. Most of the researches on DIDS fall in the category of anomaly based detection as they detect intrusive attack if they deviate from the normal behavioral access pattern and they do not rely on signature type attack that is in case of misuse detection category. It is common and general that, anomaly based intrusion detection have to cope with large volume and high dimensional data gathered from multiple sources and has to deal with real dynamic environment. This is the reason, the computational Intelligence come into play in the field of intrusion detection. To achieve real time database intrusion detection, researchers investigated so many methods to improve the accuracy. But this may not be sufficient to secure the dynamic real time database.

Conclusion

In this paper, two soft computing techniques for database intrusion detection were employed and the model was successfully demonstrated its usefulness on a synthetic dataset. RBF network was used as a classifier for intrusion detection in database. PSO is used in RBFN to improve its performance in clarity of decision making process. Experimental results showed that the proposed DIDS in detecting intrusions in Database. Our future work will focus on building accurate DIDS model by combining both misuse and anomaly based detection process by embedding several soft computing techniques

1,075 total views, no views today

Download PDF File

About the author: admin