Seismic-driven neural intelligence coupled with hybrid evolution strategy for accurate intrusion detection in cloud-IoT systems

Cloud-hosted Internet of Things (IoT) platforms are now the cornerstones of digital infrastructure in today's age, providing scalable, networked spaces for data sharing, processing, and automation. But their open and distributed nature has also made them attractive targets for almost any form o...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:The Journal of supercomputing Jg. 81; H. 16; S. 1557
Hauptverfasser: Saravanan, T., Pugalenthi, R.
Format: Journal Article
Sprache:Englisch
Veröffentlicht: New York Springer US 12.11.2025
Springer Nature B.V
Schlagworte:
Algorithms
Architecture
Artificial intelligence
Cloud computing
Collaboration
Compilers
Computer Science
Connectivity
Construction industry
Cybersecurity
Data processing
Deep learning
Digital transformation
Effectiveness
Intelligence
Internet of Things
Interpreters
Intrusion detection systems
Machine learning
Pattern recognition
Processor Architectures
Programming Languages
Real time
Security systems
Software services
Workloads
Deep learning
Cloud computing
Feature analysis and classification
Intrusion detection system (IDS)
Security
Internet of things (IoT)
ISSN:1573-0484, 0920-8542, 1573-0484
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Cloud-hosted Internet of Things (IoT) platforms are now the cornerstones of digital infrastructure in today's age, providing scalable, networked spaces for data sharing, processing, and automation. But their open and distributed nature has also made them attractive targets for almost any form of cyber-attack, indicating the pressing need for effective and intelligent intrusion detection mechanisms. Old intrusion detection systems are typically overloaded with massive false positives, poor flexibility when confronted with changing attack vectors, and poor scalability in facilitating varied network environments. In addition, most existing deep learning models and hybrid models are plagued with fixed feature engineering, poor parameter tuning, and poor contextual pattern recognition, all of which compromise their real-time detection quality and efficiency. In response to such threats, this paper presents a novel system comprised of the combination of the Seismic Pattern Integration with Neural Attention Engine (SPINE) for adaptive and efficient intrusion detection, and the Hummingbird Optimized Pied Kingfisher Evolution (HoPE) bio-inspired hybrid optimization algorithm designed to improve the learning effectiveness of the system. The SPINE model utilizes a seismic pattern-based feature to understand the neural attention, which allows it to capture very detailed spatial and temporal attack patterns. Furthermore, the HoPE algorithm plays a crucial role in hyperparameter adjustment by employing an evolutionary strategy that takes its inspiration from the incredible motor and sensor abilities of the hummingbird and pied kingfisher during their fast and agile evasive aerial hunting. In combination, this synergy creates a strong and self-sustaining detection mechanism that has the ability to learn sophisticated intrusion patterns in real-time. The SPINE + HoPE model topped all current state-of-the-art methods consistently with an all-time best performance of 99.81% accuracy, 99.70% precision, 99.58% recall, and 99.82% F1-score when tested against the BoT-IoT dataset. All these results demonstrate the effectiveness of the proposed method in carrying out accurate, efficient, and robust intrusion detection in cloud-based IoT networks and hence setting a new benchmark for cybersecurity intelligence.
Bibliographie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1573-0484
0920-8542
1573-0484
DOI:10.1007/s11227-025-07988-z