Loading…

An efficient botnet detection with the enhanced support vector neural network

•Develop an Enhanced Support Vector Neural Network (ESVNN) for botnet detection.•Employ artificial flora algorithm for the parameter tuning of ESVNN model.•Test and classify the bot and normal traffic flows at the detection phase.•Validate the performance of the ESVNN model on CTU-13 dataset. As the...

Full description

Saved in:
Bibliographic Details
Published in:Measurement : journal of the International Measurement Confederation 2021-05, Vol.176, p.109140, Article 109140
Main Authors: Jagadeesan, S., Amutha, B.
Format: Article
Language:English
Subjects:
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•Develop an Enhanced Support Vector Neural Network (ESVNN) for botnet detection.•Employ artificial flora algorithm for the parameter tuning of ESVNN model.•Test and classify the bot and normal traffic flows at the detection phase.•Validate the performance of the ESVNN model on CTU-13 dataset. As the botnet makes the way for many illegal activities, it is considered as the most critical threats to cybersecurity. Although many detection models have been presented by the researchers, they couldn’t detect the botnets in an early stage. So to overcome this issue, an enhanced support vector neural network (ESVNN) is presented as the detection model in this paper. For enhancing the classification accuracy, the suitable features of traffic flows are selected from the dataset. By observing the constant response packets, the features such as response packet ratio of the bot, length of the initial packet, packet ratio and small packets are extracted. These extracted features are used as input features for the proposed ESVNN classifier or prediction model. In ESVNN, Artificial Flora (AF) algorithm is presented for enhancing the performance of SVNN. The AF is an intelligent algorithm which is inspired from the reproduction and the migration characteristics of flora. Simulation results depict thatthe novel botnet detection model achieves better accuracy and F-measure than the existing prediction models. The presented model has reached to a higher precision of 0.8709, recall of 0.8636, accuracy of 0.8684, and F-score of 0.8669.
ISSN:0263-2241
1873-412X
DOI:10.1016/j.measurement.2021.109140