View in EDS

FD3RN: Fused Weight–Based Hybrid Distributed Drift‐Enabled Deep Learning Model for Malicious Node Detection in Wireless Sensor Network.

Saved in:
Bibliographic Details
Title: FD3RN: Fused Weight–Based Hybrid Distributed Drift‐Enabled Deep Learning Model for Malicious Node Detection in Wireless Sensor Network.
Authors: Mohammed, Zeeshan Ahmed1 (AUTHOR) dr.zeeshan.mohamm@gmail.com, Shabana, Mahammad2 (AUTHOR), Rafi, Mohammed3 (AUTHOR), Masthan, Kaja4 (AUTHOR)
Source: International Journal of Communication Systems. 7/10/2026, Vol. 39 Issue 10, p1-23. 23p.
Subjects: Wireless sensor network security, Intrusion detection systems (Computer security), Deep learning, Mathematical optimization, Recurrent neural networks, Wireless sensor networks
Abstract: Wireless sensor networks (WSNs) consist of lightweight, low‐power processing units known as sensor nodes, which communicate wirelessly via radio transceivers to form an interconnected network. However, there is a possibility of the malfunctioning of such sensor nodes in the network, thereby resulting in security concerns. The earlier attack detection frameworks exhibit certain discrepancies in terms of reduced detection accuracy, increased false alarm rate, and higher execution timings. Therefore, the research proposes an efficient fused weight tuning–based hybrid Distributed Drift‐enabled Deep Recurrent neural Network (FD3RN) to perform malicious node detection. The drift mechanism enables the model to offer optimal performance with reduced false positives, irrespective of the dynamic data characteristics. Furthermore, the weight updation using the fused adam gradient descent (FAGD) optimization improves the robustness of the research with better adaptability. In addition, the usage of the gated attention strategy aids in the better prioritization of the essential features, thereby reducing the overfitting challenges. Moreover, the synthetic minority oversampling technique (SMOTE) mitigates the consequences of using the imbalanced data in terms of reduced computational complexity. The results underscore the advantage of using the proposed model in malicious node detection through attaining 98.12% accuracy, 98.86% precision, 97.39% recall, and 98.12% F1‐score for 90% training using the UNSW‐NB15 dataset. [ABSTRACT FROM AUTHOR]
Copyright of International Journal of Communication Systems is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
Description
Abstract:Wireless sensor networks (WSNs) consist of lightweight, low‐power processing units known as sensor nodes, which communicate wirelessly via radio transceivers to form an interconnected network. However, there is a possibility of the malfunctioning of such sensor nodes in the network, thereby resulting in security concerns. The earlier attack detection frameworks exhibit certain discrepancies in terms of reduced detection accuracy, increased false alarm rate, and higher execution timings. Therefore, the research proposes an efficient fused weight tuning–based hybrid Distributed Drift‐enabled Deep Recurrent neural Network (FD3RN) to perform malicious node detection. The drift mechanism enables the model to offer optimal performance with reduced false positives, irrespective of the dynamic data characteristics. Furthermore, the weight updation using the fused adam gradient descent (FAGD) optimization improves the robustness of the research with better adaptability. In addition, the usage of the gated attention strategy aids in the better prioritization of the essential features, thereby reducing the overfitting challenges. Moreover, the synthetic minority oversampling technique (SMOTE) mitigates the consequences of using the imbalanced data in terms of reduced computational complexity. The results underscore the advantage of using the proposed model in malicious node detection through attaining 98.12% accuracy, 98.86% precision, 97.39% recall, and 98.12% F1‐score for 90% training using the UNSW‐NB15 dataset. [ABSTRACT FROM AUTHOR]
ISSN:10745351
DOI:10.1002/dac.70517