Comparative Analysis of ML Classifiers for IoT Botnet Detection Using UNSW-NB15

View at publisher Summarise with VIRAT
Authors: Preeti Kailas Suryawanshi, Sahil Jagtap
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

With the widespread deployment of IoT devices, their security is now a significant issue with the increasing botnet attacks taking advantage of system vulnerabilities. The conventional security mechanisms are unable to detect sophisticated intrusions, and hence there is a requirement for intelligent solutions. This study presents a ML-IDS for IoT botnet capture using the UNSW-NB15 dataset. The proposed methodology includes data preprocessing steps involving Exploratory Data Analysis (EDA), label encoding, feature selection, and SMOTE for class balancing. Different classifiers such as Decision Tree, Random Forest, Logistic Regression, KNN, XGBoost, and MLP are compared for binary and multiclass threat diagnosis Finally, Decision Tree classifier was superior in diagnosis recall, F1-score, accuracy and precision compared to others and is hence a perfect choice for real-time deployment in resource-constrained IoT networks. The results justify the need to select interpretable and computationally efficient models for IoT security. The model also shows good generalizability for diverse attack types in multiclass applications. The research also addresses the limitation of dealing with imbalanced datasets and optimal feature selection. The proposed IDS presents a lightweight and scalable approach to detecting diverse botnet intrusions. This research contributes to improving secure IoT deployments with realistic, AI-based defences.

Fields of Study & Subjects

Network Security and Intrusion Detection Internet Traffic Analysis and Secure E-voting Network Packet Processing and Optimization

Publication details

Journal / source
CRC Press eBooks
Pages
265–270
Publication date
25 March 2026
Language
EN
Document type
Book chapter