Machine Learning based Sink Node Positioning for Lifetime Extension of Multihop Wireless Sensor Network

Each sensor node in the multi-hop wireless sensor network participates in the data relay. In the existing sensor network, nodes near the sink nodes shorten their network lifetime due to energy hole problems cause by frequent data relays. To solve this problem, we propose a ring-based routing simultaneous wireless information and power transfer (RBR-SWIPT) technique which fuses a RBR algorithm with an energy harvesting technique. However, existing studies only assume which sensor nodes are uniformly distributed, otherwise we have confirmed the performance degradation. We also confirm that the lifetime of the network changes depending on the distribution of sensor nodes. In this paper, we propose an optimal sink node location search using density-based spatial clustering with noise (DBSCAN), which is a density-based clustering, and verify its performance. We also prove that it is possible to search for the optimal sink node location according to the distribution of sensor nodes through the proposed method.