International Journal of Electrical and Computer Engineering (IJECE) Vol. 14, No. 5, October 2024, pp. 5543~5553 ISSN: 2088-8708, DOI: 10.11591/ijece.v14i5.pp5543-5553  5543 Journal homepage: http://ijece.iaescore.com Conflict-driven learning scheme for multi-agent based intrusion detection in internet of things Durga Bhavani Attluri, Srivani Prabhakara Department of Computer Science and Engineering, BMS Institute of Technology and Management, Bangalore, Visvesvaraya Technological University, Belagavi, India Article Info ABSTRACT Article history: Received Mar 5, 2024 Revised Jun 24, 2024 Accepted Jul 2, 2024 This paper introduces an effective intrusion detection system (IDS) for the internet of things (IoT) that employs a conflict-driven learning model within a multi-agent architecture to enhance network security. A double deep Q-network (DDQN) reinforcement learning algorithm is implemented in the proposed IDS with two specialized agents, the defender and the challenger. These agents engaged in an antagonistic adaptation process that dynamically refined their strategies through continual interaction within a custom-made environment designed using OpenAI Gym. The defender agent aims to identify and mitigate threats by matching the actions of the challenger agent, which is designed to simulate potential attacks in the environment. The study introduces a binary reward mechanism to encourage both agents to explore and exploit different actions and discover new strategies as a response to adversarial actions. The results showcase the effectiveness of the proposed IDS in terms of higher detection rate the comparative analysis also validates the effectiveness of the proposed IDS scheme with an accuracy of approximately 96%, outperforming similar existing approaches. Keywords: Internet of things Intrusion detection system Multi-agent system Reinforcement learning Security This is an open access article under the CC BY-SA license. Corresponding Author: Durga Bhavani Attluri Department of Computer Science and Engineering, BMS Institute of Technology and Management Bangalore, Karnataka, India Email: durga842004@bmsit.in 1. INTRODUCTION The increasing connection of physical and digital worlds through internet of things (IoT) devices has unleashed many benefits across industries [1]. However, increasing interconnected devices has created a complex cybersecurity landscape. This has resulted in unique vulnerabilities, which malicious actors often exploit. These attacks can compromise critical aspects, such as privacy, integrity, and availability [2]. Traditional security measures need help adapting to the dynamic nature of the IoT and attackers' evolving tactics [3]. Resource limitations on many devices necessitate robust and lightweight security solutions [4]. intrusion detection systems (IDSs) are crucial for identifying and mitigating cyber threats in the IoT, but their dependence on known attack signatures limits their effectiveness against unknown and evolving threats [5], [6]. This limitation highlights the need to develop adaptive IDSs that detect known attacks and adapt to uncover unknown security threats and attacks. Recent advancements in machine learning (ML) techniques and their integration into IDSs have shown promising results in the context of network security [7], [8]. However, IDS developed using supervised learning models highly depend on labelled data quality and are ineffective against evolving cyber threats. Unsupervised learning offers an alternative approach to developed IDS but suffers from high false positives and vulnerabilities to adversarial attacks [9], [10]. On the other hand, reinforcement learning (RL), an alternative ML approach, enables IDSs to adapt to evolving threats by continuously learning and exploring