1. Enhancing wireless sensor network performance through anomaly-based intrusion detection and artificial neural network: IEEE 802.15.4 protocol.
- Author
-
Sultan, Hasan Farag and Fadhil, Mohammad Natiq
- Subjects
- *
ARTIFICIAL neural networks , *INDUSTRIAL robots , *WIRELESS channels , *ENVIRONMENTAL monitoring , *WIRELESS sensor networks , *ZIGBEE , *INTRUSION detection systems (Computer security) - Abstract
Over the years, wireless sensor networks (WSNs) turned out to be a game-changing technology for numerous purposes like environmental monitoring, industrial automation, agriculture and many more. WSN comprises low-priced, small sensors that collect and send data to the central node through the wireless channel. WSNs can provide you with data that you can use to track and manage everything from temperature and humidity to light. In this paper we present the state-of-the-art research on wireless sensor networks covering the areas of sensor design, network architecture, routing protocols, and energy management. First, the challenges specifically related to the WSNs as it uses limited resources and the one of the most energy efficient protocols, Zigbee, are examined. We used the backpropagation-supervised multi-layer neural network to achieve high prediction accuracy, employing three theories of backpropagation: Three different approaches, scale conjugation, Bayesian regularization, and Levenberg Marquardt, which are the results of my study in this paper. These methods were assessed using accuracy as the criterion for each of the tasks, and the results were given out as percentages. The best score was gotten by Bayesian regularization, it scored 99.73%, while the scale conjugation scored 98.96%. The Levenberg Marquardt approach attained 98.10% score which slightly lower than that of 98.45% for the BFGS method. The output implies that Bayesian regularization provides the highest accuracy for this task. Moreover, conjugation scale and Levenberg-Marquardt can be also employed but with less efficiency. [ABSTRACT FROM AUTHOR] more...
- Published
- 2025
- Full Text
- View/download PDF