Optimal Deployment of Wireless Sensor Networks Based on Reverse Node Selection
Wei Wang, Jiaxin Wu, Rongfang Du, Naiyu Cui · 2024
There are a large number of redundant nodes in wireless sensor networks, which causes a serious waste of network resources and affects network performance. This paper proposes an optimal deployment algorithm for wireless sensor networks based on reverse node selection and graph attention mechanism. By assessing the similarity between the newly generated graph topology of the sensor network and that of the unoptimized one, the algorithm adeptly identifies redundant sensor nodes and puts them to sleep. This meticulous process ensures the optimal deployment of wireless sensor network. In the graph similarity comparison stage, a graph attention mechanism is introduced to learn the attention weights of nodes and compute the graph embedding, aiming to obtain a more accurate expression of graph similarity. The experiment is conducted on China's Ground-Based Basic Meteorological Observation Sensor Network (CGBMOSN). The results demonstrate that the algorithm can capture up to 95 percent of the complete sensor network information with only 60 percent of nodes.