Reinforcement Learning with Cybersecurity Dataset for Internet of Things Multifactor Authentication

Read full text Summarise with VIRAT
Authors: Tiba S. Mahmood, Emad H. Al-Hemiary
Cited by: 0
References: 25

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

Authentication mechanism between different parties of the Internet of Things system is significant to prevent unauthorized access, especially in the Industrial environments, where multiple factors collaborate to achieve that goal.Most traditional multifactor authentication schemes are known to be static and require equivalent verification across all devices, reducing usability to a dynamic risk context.On the other hand, the current advances in Machine Learning techniques encourage their integration in many systems where security is required, but in an efficient way.The work in this paper implements a multifactor authentication mechanism powered by a reinforcement learning technique to adjust the authentication factors based on the contextual risk level.The learning and testing phases are carried out using a modified dataset named mTON-IoT with multifactor authentication features that have been derived from the publicly known cybersecurity dataset TON-IoT.The practical implementation of the proposed system involves using the Double Q-Learning technique within a Raspberry Pi network and securely exchanging data using Elliptic Curve Cryptography over the Message Queuing Telemetry Transport protocol.The experimental results show a high decision accuracy of 99.72%, confirming the trained model's ability to obtain an intelligent balance between security and usability in resource-constrained IoT.

Fields of Study & Subjects

User Authentication and Security Systems Advanced Authentication Protocols Security Access Control and Trust

Publication details

Journal / source
International journal of intelligent engineering and systems
Volume
19
Issue
4
Pages
1189–1210
Publication date
23 March 2026
Language
EN