Reliability Analysis For Deep Learning Secured Wireless Sensor Networks

Abdulrasool, Abeer Hussein; Hamza, Ekhlas Kadhim; Hasan, Ahmed Mudheher

doi:https://doi.org/10.33103/2617-3352.1518

	Reliability Analysis For Deep Learning Secured Wireless Sensor Networks
IRAQI JOURNAL OF COMPUTERS, COMMUNICATIONS, CONTROL AND SYSTEMS ENGINEERING
Volume 26, Issue 1, April 2026 PDF (4.81 M)
Document Type: Research Paper
DOI: https://doi.org/10.33103/2617-3352.1518
Authors
Abeer Hussein Abdulrasool^* ¹; Ekhlas Kadhim Hamza²; Ahmed Mudheher Hasan³
¹control and systems engineering
²Control and System Eng. Dept. /University of Technology-Iraq
³Collage of Artificial Intelligence Engineering, University of Technology, Iraq,
Abstract
With the advancement of software, wireless sensor networks (WSNs) face many problems and challenges, such as loss or limitation of node power resources, as well as intrusions and other cyberattacks. This study aims to improve the reliability of WSNs to mitigate the impact of cyber- attacks by applying eﬀective detection and prevention techniques to ensure their continuous operation, as well as to maintain the security, privacy, and integrity of information. Several eﬃcient approaches and smart techniques have been presented in literature to enhance the reliability and security of WSNs against various cyber-attacks. Recent studies focused on the deep learning (DL) algorithms as eﬃcient and robust solutions in detecting and eliminating the threats up on WSNs. In this study, a deep long short tern memory (LSTM) algorithm that utilizes both short-term and long-term memory and performs well with long-term dependencies was presented to handle intrusions in the sensor network. The model uses LSTM networks to manage these variances and provide security and reliability because mobile attacks might have unpredictable patterns. Every time the regulated parameters are iterated, each node communicates its position and degree of movement, along with other significant information. Important metrics included the ratio of active to dormant nodes, data received per node, communication time, energy consumption per iteration, and communication time between nodes. These measures help assess how stable the model is in diﬀerent scenarios. Such measurements allow us to predict future node behaviors and the network’s highly reliable operation in addition to focusing on stability and security. The proposed LSTM deep learning attack defense model has been applied to improve the reliability of the sensor network by 30%, extend the lifetime of active nodes by 22%, and reduce node energy consumption in the WSN by 55%.
Keywords
Reliability Analysis; Wireless Sensors Networks; Deep Learning Technique; Long Short Term Memory (LSTM) Algorithm; Life Time

Statistics Article View: 4 PDF Download: 13