A probability estimation-based feature reduction and Bayesian rough set approach for intrusion detection in mobile ad-hoc network

Mahendra Prasad*, Sachin Tripathi, Keshav Dahal

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

1 Citation (Scopus)

Abstract

A mobile ad-hoc network is a small and temporary network. This network has a different working principle and structure than wired networks. A source node transfers data to the destination node through intermediate nodes. Due to mobility of node, this network is more vulnerable to routing attacks. Many security mechanisms protect the network from intrusions, such as cryptography based, lightweight, and heavyweight techniques. But, these are not powerful enough mechanisms for mobile ad-hoc networks to mitigate routing attacks. Therefore, we have proposed an enhanced intrusion detection system for the mobile ad-hoc network that handles routing attacks. This method mainly generates 11 sub-datasets and also evaluates their quality using a fuzzy logic system. We suggest a probabilistic approach for feature ranking. The next process removes ineffective features from training and test sets. We have
applied a Bayesian rough set classifier that classifies the behavior of mobile nodes using incoming packets. The Bayes classifier is applied for ambiguous and unknown samples. Experimental results show that the average detection accuracy is 94.37% for blackhole attack and 99% for wormhole attack. The proposed method performs better than existing intrusion detection methods.
Original languageEnglish
JournalApplied Intelligence (APIN)
Early online date16 Jul 2022
DOIs
Publication statusE-pub ahead of print - 16 Jul 2022

Keywords

  • rough set theory
  • feature ranking
  • dataset quality evaluation
  • mobile ad-hoc network
  • blackhole attack
  • wormhole attack
  • Bayes classifier

Fingerprint

Dive into the research topics of 'A probability estimation-based feature reduction and Bayesian rough set approach for intrusion detection in mobile ad-hoc network'. Together they form a unique fingerprint.

Cite this