A Low-Energy Data Gathering Technique for Mobile Sink-Based Wireless Sensor Networks Using Unsupervised Learning and TSP
Fahim Hasan, Selina Sharmin, Sajeeb Saha, Tanvir Ahammad · 2024
Adding mobile sinks (MSs) to wireless sensor networks (WSNs) has proven to increase network longevity and improve data delivery services. We accomplish this goal by reducing routing expenses and preventing the emergence of any hotspot zones inside the network. On the other hand, path determination and MS visiting point selection in WSNs are difficult. In this study, we use MS to create an effective data collection method known as the LEDG (Low-Energy Data Gathering) technique. Using an unsupervised learning-based Hierarchical Agglomerative Clustering (HAC) algorithm, we construct the clusters, and we also find the ideal number of clusters and cluster heads (CHs) using the Silhouette Coefficient (SC) method. At last, we find out the visiting sites of MS. By utilizing the ideal number of clusters, we can lower energy consumption and multi-hop communication in a cluster. Additionally, we suggest a low-complexity trajectory plan among the CHs for the MS utilizing the Geometric algorithm based TSP (Traveling Salesman Problem). We run simulations in Network Simulator-3 (NS-3), and the results show that our approach outperforms previous literature-based solutions.