Optimized Self-Attention Pyramidal Convolutional Neural Network for Intrusion Detection Framework in IoT
Padma Yenuga, Satya Narayana Reddy Beeram, Sitanaboina S L Parvathi, Gautham Reddy, Venugopal Boppana, Sunitha Davuluri, Repudi Ramesh, Meda Srikanth, B. Vara Prasad Rao, Ravi Kumar Munaganuri, Narasimha Rao · Journal of Advances in Information Technology · 2025
This research finds an important place in intrusion detection within the landscape of IoT when it puts forward the optimized Intrusion Detection System (IDS) solution.This is enabled by using the Self-Attention Pyramidal Convolutional Neural Network (SAPCNN) that is powered by Hybrid Ebola and Bald Eagle Search Optimization Algorithm, thereby enhancing classification accuracy.The methodology of this technique is basically a preprocessing tool called Dynamic Context-Sensitive Filtering (DCSF) aimed at removing data redundancy as well as filling missing values, followed by the mechanism of Pelican Optimization Algorithm (POA)-based feature selection.Performance evaluations on the Canadian Institute for Cybersecurity Intrusion Detection System 2017 Dataset (CICIDS2017) dataset reveal that the proposed IDS can accurately detect Distributed Denial of Service (DDoS) attacks with a precision of 98.9% and reduce the computational time by 31.7% as compared to baseline models.These results therefore indicate that the model can successfully handle complex Internet of Things (IoT) intrusion scenarios with great precision and efficiency.