High Impact Factor : 4.396 icon | Submit Manuscript Online icon |

SVM Based Anomaly and Intrusion Detection System in Mobile Ad Hoc Networks

Author(s):

Dinesh Swami , Sri Balaji College of Engineering and Technology; Amit Kumar Mishra, Sri Balaji College of Engineering and Technology

Keywords:

MANETs, SVM, ADS, IDS, Node Behavior, Routing

Abstract

Communication Networks forms the backbone of Information Technology era. With the advancements in technology, several types of networks are designed and implemented ranging from Satellite Communication networks to Household Personal Area Networks. A mobile ad-hoc network is a collection of mobile devices (laptops, smartphones, sensors, etc.) that communicate with each other over wireless links and cooperate in a distributed manner in order to provide the necessary network functionality in the absence of a fixed infrastructure. This type of network, operating as a stand-alone network or with one or multiple points of attachment to cellular networks or the Internet, paves the way for numerous new and exciting applications. Due to the ad-hoc architecture of MANETS, these are susceptible to attack by an intruder. Generally, the nodes are battery powered and run on low-cost hardware, thus starts malfunction with aging. In both the cases, the network gets compromised and does not perform as desired. Therefore, Anomaly Detection Systems (ADS) and Intrusion Detection Systems (IDS) are a critical aspect for almost all kinds of MANETs. In this Paper, an approach is developed to tabulate the network data on a regular basis. This tabulated data is examined through an automated procedure to ensure that the network is functioning in the normal way and is unaffected. This is done by designing a Support Vector Machine based Classifier. Initially, the data is manually classified into two classes, the first corresponding to the normal operation of the network and the second corresponding to the malicious operation of the network. The proposed work extends the work of T. Poongothai et. al [1]. By extending the feature set that forms the input to support vector machine. The feature set used previously involves the use data forwarding behavior at the network layer. However, the approach presented in this Paper extends the feature set by including, in addition to the packet forwarding features of the network layer, the node credit score based on the trust measure of the node. The attack / anomaly profile is created by simulation and it turns out that the proposed scheme outperforms the existing approach for anomaly / intrusion detection.

Other Details

Paper ID: IJSRDV5I60134
Published in: Volume : 5, Issue : 6
Publication Date: 01/09/2017
Page(s): 1999-2006

Article Preview

Download Article