SA-IDS: A single attribute intrusion detection system for Slow DoS attacks in IoT networks

Reed, Andy; Dooley, Laurence and Kouadri Mostéfaoui, Soraya (2025). SA-IDS: A single attribute intrusion detection system for Slow DoS attacks in IoT networks. Internet of Things, 30, article no. 101512.

DOI: https://doi.org/10.1016/j.iot.2025.101512

Abstract

Internet of Things (IoT) technologies are expanding and pervade evermore application domains bringing a raft of positive user benefits. However, the matter of application layer security and the omnipresent danger of Denial of Service (DoS) attacks remains a significant risk to effective IoT performance. DoS is especially serious in IoT networks given the propensity for malicious nodes to mimic legitimate nodes encountering slow connectivity, a problem intensified in very stochastic traffic environments where higher node latencies create even stealthier Slow DoS conditions.

The contribution this paper presents is a flexible single attribute intrusion detection system (SA-IDS) for IoT networks, which employs a novel variable threshold range for just the delta time network attribute, to accurately detect Slow DoS attacks in highly stochastic traffic, while crucially still being able to reliably discriminate malicious from legitimate slow node activity. Experimental results in a live IoT network compellingly demonstrate the superior detection performance of SA-IDS under the stealthiest Slow DoS attack conditions, where genuine nodes with high latency are almost indistinguishable from malicious nodes, thus rendering existing Slow DoS detection methods ineffective that rely solely on static thresholds based on network traffic attribute analysis.

Viewing alternatives

Metrics

Item Actions

Export

You can export this page using these formats Digital Object Identifier - DOI