Chang, Y.Yuan, X.Li, B.Niyato, D.Al-Dhahir, Naofal2019-06-282019-06-282018-09-181558-25588https://hdl.handle.net/10735.1/6632Full text access from Treasures at UT Dallas is restricted to current UTD affiliates.Energy efficiency is a key performance metric for ultra-dense wireless sensor networks. In this letter, an unsupervised learning approach for topology control is proposed to prolong the lifetime of ultra-dense wireless sensor networks by balancing energy consumption. By encoding sensors as genes according to the network clusters, the proposed genetic-based algorithm learns an optimum chromosome to construct a close-to-optimum network topology using unsupervised learning in probability. Moreover, it schedules some of the cluster members to sleep to conserve the node energy using geographically adaptive fidelity. Simulation results demonstrate the superior performance of the proposed algorithm by improving energy efficiency in comparison with state-of-the-art algorithms at an acceptable computational complexity.en©2018 IEEECytologyAlgorithmsEnergy consumptionElectric network topologyDetectorsWireless sensor networksBioinformaticsElectric power-plants--Efficiency--Computer programsBioenergeticsTopologyEnergy consumptionarticleChang, Y., X. Yuan, B. Li, D. Niyato, et al. 2018. "A joint unsupervised learning and genetic algorithm approach for topology control in energy-efficient ultra-dense wireless sensor networks." IEEE Communications Letters 22(11): 2370-2373, doi:10.1109/LCOMM.2018.28708862211