Adaptive edge-based behavioural anomaly detection and response system for smart home IoT Networks
School of Computing, MIT ADT University, Loni Kalbhor, Pune.
Research Article
Open Access Research Journal of Science and Technology, 2026, 17(01), 089-103.
Article DOI: 10.53022/oarjst.2026.17.1.0065
Publication history:
Received on 17 May 2026; revised on 24 June 2026; accepted on 26 June 2026
Abstract:
The quick expansion of Internet of Things (IoT) devices in smart home settings has made people much more vulnerable to online threats, especially large-scale IoT botnets and zero-day attacks. Traditional intrusion detection systems based on signatures are useless against new and emerging threats [1]. While many solutions rely on centralized processing, lack device-specific modeling, or do not incorporate adaptive enforcement mechanisms, recent approaches make use of behavioral anomaly detection, such as deep learning-based models for IoT traffic analysis [2]. In this paper, we propose an adaptive edge-based IoT security framework that uses unsupervised anomaly detection to learn device-specific behavioral baselines. The suggested system models typical device activity and instantly identifies deviations using the Isolation Forest algorithm [3]. AEIS uses a hybrid anomaly detection approach combining Isolation Forest (unsupervised) for detecting unknown attacks and Random Forest (supervised) for classifying known attack patterns. The framework ensures privacy preservation and lower latency by only analyzing metadata while operating at the network edge. A closed-loop, self-healing security architecture is made possible by a graduated policy engine that applies proportional restriction, isolation, and automated recovery. In comparison to static threshold-based methods, experimental evaluation using a hybrid dataset that combines publicly available IoT botnet traffic and real network metadata [2] shows increased detection accuracy and decreased false positive rates. For smart home and small-scale IoT deployments, the suggested framework provides a scalable, lightweight, and privacy-conscious solution [4].
Keywords:
Internet of Things (IoT); Smart Home Security; Anomaly Detection; Edge Computing; Isolation Forest; Behavioral Profiling; Intrusion Detection System (IDS); Device-Specific Modeling; Self-Healing Security; Network Metadata Analysis.
Full text article in PDF:
Copyright information:
Copyright © 2026 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution Liscense 4.0