A power-efficient K-means clustering algorithm for Wireless Sensor Networks (WSN) is proposed. This algorithm aims to manage the consumption of energy by WS nodes and enhance the running time for WSN given space constraints. WS node cluster formation is structured as a sample space partition in k-means for the reason that the radio channel is unstable and the distribution of the nodes is coarse. After measuring the overall network energy consumption, the optimal Cluster Heads (CH’s) are evaluated on the basis of network size. The length of space from CH to node is evaluated and the membership weight is considered for the objective function. We propose an approach for making numerous node clusters using an improved K-means clustering algorithm called Optimal K-means (OK-means). A single hop communication mode is employed for intra-cluster communication whereas a multi-hop communication mode is used by the inter-cluster communication. The performance is evaluated using Ns-2 simulator. The outputs of these simulations show that the proposed algorithm achieves uniform distribution in spatial domain of CH. Which effectively balance the energy consumption. Further, extensive simulations have been carried out by varying node densities to demonstrate the full potential of OK-means.

提出了一种用于无线传感器网络（WSN）的高效节能的K均值聚类算法。该算法旨在管理WS节点的能源消耗，并在给定空间限制的情况下延长WSN的运行时间。WS节点群集的形成被构造为以k均值表示的样本空间分区，原因是无线电信道不稳定且节点的分布较粗糙。在测量了整个网络的能耗之后，将根据网络规模评估最佳的簇头（CH）。评估从CH到节点的空间长度，并考虑目标函数的隶属度。我们提出一种使用称为最优K均值（OK-means）的改进的K均值聚类算法来制作众多节点集群的方法。集群内通信采用单跳通信模式，集群间通信采用多跳通信模式。使用Ns-2模拟器评估性能。这些模拟的输出表明，该算法在CH的空间域中实现了均匀分布。从而有效地平衡了能耗。此外，已经通过改变节点密度进行了广泛的仿真，以证明OK均值的全部潜力。