A Comparative Study on IOT Security Using Machine Learning Techniques

Title

A Comparative Study on IOT Security Using Machine Learning Techniques

Creator

Joseph, Eldo Ephraim; Sharma, Vandana; Shoran, Preety

Description

This increased reliance on networks has made the security of IoT devices a highly burning issue. Among the sea of threats, the threats associated with DDoS attacks are at a high level since they do damage to the functionality of devices and unavailability of networks. Detection and mitigation of DDoS attacks will demand IoT environments based on powerful classification algorithms. This paper evaluates the performances of three prominent algorithms: Decision Tree, Random Forest, and Histogram-Based Gradient Boosting for the classification of DDoS attack traffic within IoT networks. An IoT-23 dataset comprising a subset of attacks, including DDoS, is used herein for the purpose of achieving high classification accuracy to ensure a reliable evaluation of attacks. The results clearly show that all three algorithms are pretty good in terms of detection performance, and Histogram-Based Gradient Boosting is the best in terms of generalization accuracy. These results open new perspectives for the implementation of machine learning, generally, and Histogram-Based Gradient Boosting, specifically, directed to improving security in IoT networks against DDoS attacks, which is an extremely promising result when working within the light of some insights for future research and development within this critical area of security. The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2026.

Source

Lecture Notes in Networks and Systems;Volume;1496 LNNS;pp.439-450

Date

01-01-2026

Publisher

Springer Science and Business Media Deutschland GmbH

Subject

Artificial intelligence; Cybersecurity; Internet of Things (IoT); Machine learning

Coverage

Joseph E.E., Christ University, Bengaluru, India; Sharma V., Christ University, Bengaluru, India; Shoran P., Christ University, Bengaluru, India

Rights

Restricted Access; Hardcopy may be available in the library

Relation

ISSN: 23673370; ISBN: 978-981968106-8;

Format

online

Language

English

Type

Conference paper

Identifier

https://doi.org/10.1007/978-981-96-8107-5_32

https://www.scopus.com/pages/publications/105028314790?origin=resultslist