ASIO-CNN: An adaptive swarm intelligence framework for IoT intrusion detection
2026, vol.18 , no.4, pp. 127-136
Article [2026-04-12]
The fast growth of Internet of Things (IoT) networks has put more risk of advanced cyber-attacks, and smart and flexible intrusion detection systems are needed. This paper presents a framework named NASIO-CNN as a Nature-Inspired Adaptive Swarm Immune Optimization framework that can automatically optimize hyperparameters of a Convolutional Neural Network (CNN) to identify IoT cyber-attacks. The proposed algorithm is a hybrid of immune clonal selection, adaptive mutation, swarm learning, and elite memory retention to achieve effective exploration and constant optimization. Statistical validation is done on the BoT-IoT, TON-IoT, UNSW-NB15 and CICIDS2017 datasets to test the framework. The experimental performance is shown to be better with 99.12% accuracy, 99.00% F1-score, 0.991 MCC and 0.998 ROC-AUC, and decreasing false positives. NASIO-CNN is an effective, scalable, and flexible framework that can be used to provide intelligent IoT intrusion detection and autonomous cyber defence.
Internet of Things, intrusion detection, convolutional neural network, nature-inspired optimization, hyperparameter tuning
https://doi.org/10.59035/ZYCL8339
Rahul R Papalkar, Sanjay Kumar Singh. ASIO-CNN: An adaptive swarm intelligence framework for IoT intrusion detection. International Journal on Information Technologies and Security, vol.18 , no.4, 2026, pp. 127-136. https://doi.org/10.59035/ZYCL8339