Scholarlink

<p>The rapid evolution of 6G communication and the utilization of the Terahertz (THz) spectrum have catalyzed the deployment of 3-dimensional ultra-dense Wireless Sensor Networks (WSNs). However, the extreme energy restrictions of nano/micro devices and the severe propagation losses inherent to THz frequencies introduce critical routing challenges. In such environments, traditional multi-hop data transmission toward a central gateway inevitably triggers the energy bottleneck, causing rapid node depletion, severe network partitioning and massive data loss. To overcome these physical layer constraints, this study proposes a novel Machine Learning-supported, energy-efficient routing and clustering algorithm specifically designed for 6G-enabled 3-dimensional energy-restricted ultra-dense WSNs. The primary novelty of this work lies in strictly integrating THz energy constraints with geometric spatial variance optimization. By utilizing an unsupervised 3-dimensional <i>k-Means</i> algorithm integrated with a quantitative Elbow method, the proposed architecture autonomously determines the mathematically optimal number and spatial coordinates of 6G-enabled routers based on the exact geometric density of the network. The performance of the proposed model is evaluated across scalable 3-dimensional deployments ranging from 1000 to 7500 nodes and benchmarked against Direct Routing, LEACH, DBSCAN and Energy-Aware Routing protocols. Extensive simulation results demonstrate that the proposed ML-based architecture effectively eliminates bottleneck congestion, achieving an exceptional average network coverage rate of 93.19%. By minimizing intra-cluster THz transmission distances to an average hop count of 1.43, the proposed algorithm provides unparalleled energy conservation, reducing the end-to-end energy cost of successful packet deliveries by over 83% compared to the Direct Routing baseline. This distributed routing optimization explicitly prevents the dramatic packet drops and energy waste observed in traditional unclustered topologies, thereby significantly extending the network’s operational lifetime and reliability.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine learning supported router localization and path construction algorithm for 6G-enabled ultra-dense WSNs

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月30日

查看全文
  • Omer Gulec,
  • Emre Sahin

摘要

The rapid evolution of 6G communication and the utilization of the Terahertz (THz) spectrum have catalyzed the deployment of 3-dimensional ultra-dense Wireless Sensor Networks (WSNs). However, the extreme energy restrictions of nano/micro devices and the severe propagation losses inherent to THz frequencies introduce critical routing challenges. In such environments, traditional multi-hop data transmission toward a central gateway inevitably triggers the energy bottleneck, causing rapid node depletion, severe network partitioning and massive data loss. To overcome these physical layer constraints, this study proposes a novel Machine Learning-supported, energy-efficient routing and clustering algorithm specifically designed for 6G-enabled 3-dimensional energy-restricted ultra-dense WSNs. The primary novelty of this work lies in strictly integrating THz energy constraints with geometric spatial variance optimization. By utilizing an unsupervised 3-dimensional k-Means algorithm integrated with a quantitative Elbow method, the proposed architecture autonomously determines the mathematically optimal number and spatial coordinates of 6G-enabled routers based on the exact geometric density of the network. The performance of the proposed model is evaluated across scalable 3-dimensional deployments ranging from 1000 to 7500 nodes and benchmarked against Direct Routing, LEACH, DBSCAN and Energy-Aware Routing protocols. Extensive simulation results demonstrate that the proposed ML-based architecture effectively eliminates bottleneck congestion, achieving an exceptional average network coverage rate of 93.19%. By minimizing intra-cluster THz transmission distances to an average hop count of 1.43, the proposed algorithm provides unparalleled energy conservation, reducing the end-to-end energy cost of successful packet deliveries by over 83% compared to the Direct Routing baseline. This distributed routing optimization explicitly prevents the dramatic packet drops and energy waste observed in traditional unclustered topologies, thereby significantly extending the network’s operational lifetime and reliability.