Carbon-Aware Intrusion Detection: A Comparative Study of Supervised and Unsupervised DRL for Sustainable IoT Edge Gateways

2025-11-22

AI Summary

To address the poor adaptability, heavy reliance on labeled data, and neglect of energy efficiency and carbon emissions in traditional intrusion detection systems (IDS) deployed on IoT edge gateways, this paper proposes two lightweight deep reinforcement learning (DRL)-driven IDS models: AutoDRL-IDS (supervised) and DeepEdgeIDS (unsupervised). Our key contribution is the first introduction of a carbon-aware multi-objective reward function for edge IDS, jointly optimizing detection accuracy, energy consumption, and carbon efficiency. We design hybrid architectures—Autoencoder-DRL for unsupervised zero-day attack detection and LSTM-DRL for supervised high-accuracy classification. Experimental results on real-world edge datasets show that DeepEdgeIDS achieves 98% accuracy with robust zero-day detection capability, while AutoDRL-IDS attains 94%. Both models significantly improve energy efficiency and carbon efficiency, demonstrating strong suitability for resource-constrained edge devices.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Application Domains: Internet of Things, Sensor Networks & Smart Cities
• Search and Optimization: Learning to Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Sustainability and carbon-aware systems for Web, mobile, and WoT
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

The rapid expansion of the Internet of Things (IoT) has intensified cybersecurity challenges, particularly in mitigating Distributed Denial-of-Service (DDoS) attacks at the network edge. Traditional Intrusion Detection Systems (IDSs) face significant limitations, including poor adaptability to evolving and zero-day attacks, reliance on static signatures and labeled datasets, and inefficiency on resource-constrained edge gateways. Moreover, most existing DRL-based IDS studies overlook sustainability factors such as energy efficiency and carbon impact. To address these challenges, this paper proposes two novel Deep Reinforcement Learning (DRL)-based IDS: DeepEdgeIDS, an unsupervised Autoencoder-DRL hybrid, and AutoDRL-IDS, a supervised LSTM-DRL model. Both DRL-based IDS are validated through theoretical analysis and experimental evaluation on edge gateways. Results demonstrate that AutoDRL-IDS achieves 94% detection accuracy using labeled data, while DeepEdgeIDS attains 98% accuracy and adaptability without labels. Distinctly, this study introduces a carbon-aware, multi-objective reward function optimized for sustainable and real-time IDS operations in dynamic IoT networks.

Problem

• Addressing cybersecurity challenges in IoT edge networks against DDoS attacks
• Overcoming limitations of traditional IDS on resource-constrained edge gateways
• Developing carbon-aware intrusion detection systems for sustainable IoT operations

Innovation

• Unsupervised Autoencoder-DRL hybrid for intrusion detection
• Supervised LSTM-DRL model achieving high detection accuracy
• Carbon-aware multi-objective reward function for sustainable operations