DAIS: deep artificial immune system for intrusion detection in IoT ecosystems
Vineeta Soni, Devershi Pallavi Bhatt, Narendra Singh Yadav, Siddhant Saxena · International Journal of Bio-Inspired Computation · 2024
IoT has risen rapidly over the past decade. Massive data flow in a dynamic, decentralised environment threatens data security. This study addresses machine learning issues in IoT intrusion detection. DAIS is a bio-inspired artificial immune system architecture. The DAIS technique replicates the innate immunity and self-adaptive immune processes, which secures the dynamic IoT environment from existing and novel 'zero-day' assaults. The proposed DAIS architecture outperforms existing data-centric intrusion detection approaches and achieves benchmark accuracy of 99.87% on the MQTTset dataset and 87.64% on the imbalanced KDD-CUP-99 dataset. This means the proposed architecture is more robust to real-world attack scenarios and provides an end-to-end pipeline to secure the dynamic and evolving IoT network ecosystem.