Development of Hybrid XTF-Modified Grey Wolf Optimization and Energy Aware Support Vector Machine (XTF-GWO-SVM) For Intrusion Detection in MANETs

Publication Details

Publication Date

10/09/2026

Volume / Issue

Vol 1, Issue 4 (2026)

Article No.

061

Journal

British Journal of Contemporary Research

DOI

https://doi.org/10.67693/BJCR-P5GDNGFJ

Received

31 Aug 2026

Views

4

Downloads

0

Abstract

Mobile ad-hoc network, one decentralized, infrastructure-less wireless system which provides communication services for military operation, disaster recovery operation and mobile computing. These networks offer an abundance of feasibility for mobile ad-hoc networking, but meanwhile there are more issues such as dynamic topology, the sharing of the wireless medium, and the limited battery energy. The common intrusion detection frameworks based on SVM classifiers often experience several problems like stationary decision boundaries, parameter sensitivity and huge energy consumption, which are unsuitable for the resource restricted mobile ad-hoc network nodes. This paper describes a hybrid X shaped Transfer Function (XTF) Modified Grey Wolf Optimization (GWO) and Energy Aware Support Vector Machine (SVM) model named XTF-GWO-SVM. It co-optimizes the feature subset selection and the hyper parameters of SVM classifiers and imposes a penalty function for energy expensive feature subset in the objective function of the optimization process. Moreover, an Information Gain weighted by energy residual energy is used in initial population seeding, a transfer function in X shape is used to realize the binary feature mapping, and a multi objective fitness function is introduced to balance the detection performance and processing cost. Then the model was implemented in MATLAB R2025a and compared with the conventional GWO-SVM method on the basis of the ITU-ML5G-PS-006 (2025) data, in four kinds of injected attack ratio (25%, 50%, 75%, 100%), the detection accuracies obtained for the developed method are 88.42%, 91.15%, 91.15% and 96.82% respectively, while for the conventional method are 82.10%, 84.45%, 87.30% and 89.15%. And meanwhile both of the precision, recall and F1-score had the corresponding improvement compared to the traditional method for every injected ratio. This research shows that combining energy awareness in classification with an advanced binary optimizer results in a scalable and energy conscious mobile ad hoc network intrusion detection framework feasible for real time deployment.

License

This article is published under the Creative Commons Attribution 4.0 International License . Free to read, share, and adapt with attribution.