Department of Computer Science, College of Computer Science & Information Technology, Jazan University, Jazan 45142, Saudi Arabia.
Bule Hora University, Ministry of Education, Oromia, Ethiopia.
Comput Intell Neurosci. 2023 Jul 26;2023:4758852. doi: 10.1155/2023/4758852. eCollection 2023.
With no requirement for an established network infrastructure, wireless sensor networks (WSNs) are well suited for applications that call for quick network deployment. Military training and emergency rescue operations are two prominent uses of WSNs. The individual network nodes must carry out routing and intrusion detection because there is no predetermined routing or intrusion detection in a wireless network. WSNs can only manage a certain volume of data, and doing so requires a significant amount of energy to process, transmit, and receive. Since sensors have a modest energy source and a constrained bandwidth, they cannot transmit all of their data to a base station for processing and analysis. Therefore, machine learning (ML) techniques are needed for WSNs to facilitate data transmission. Other current solutions have drawbacks as well, such as being less reliable, more susceptible to environmental changes, converging more slowly, and having shorter network lifetimes. This study addressed problems with wireless sensor networks and devised an efficient clustering and routing algorithm based on machine learning. Results from simulations demonstrate that the proposed system beats previous state-of-the-art models on a variety of metrics, including accuracy, specificity, and sensitivity (0.93, 0.93, and 0.92 respectively).
无线传感器网络（WSN）无需建立网络基础设施，非常适合需要快速网络部署的应用。军事训练和紧急救援行动是 WSN 的两个突出应用。由于无线网络中没有预定的路由或入侵检测，因此单个网络节点必须执行路由和入侵检测。WSN 只能管理一定数量的数据，而处理、传输和接收这些数据需要大量的能量。由于传感器的能源有限，带宽也有限，因此不能将所有数据都传输到基站进行处理和分析。因此，WSN 需要机器学习（ML）技术来促进数据传输。其他当前的解决方案也存在缺点，例如可靠性较低、对环境变化更敏感、收敛速度较慢以及网络寿命较短。本研究针对无线传感器网络的问题，设计了一种基于机器学习的高效聚类和路由算法。模拟结果表明，所提出的系统在各种指标上都优于以前的最先进模型，包括准确性、特异性和敏感性（分别为 0.93、0.93 和 0.92）。
