Sensor networks are data-centric networks constrained with limited battery power and processing capabilities. One of the crucial challenges in sensor network is energy hole problem. In order to deal with the challenge, there exists several mechanisms, of which clustering is considered an energy-efficient solution. In general, clustering refers to the technique of grouping nodes on the basis of similarity in spatial arrangement. An appropriately clustered network helps in processing and aggregation of sensed data before routing the information to destined location. This paper proposes a soft computing based recursive approach for implementing density-based hierarchical clustering, for Gaussian distributed sensor network. Our proposed probabilistic approach creates hierarchical clusters recursively, which not only addresses the problem of energy hole, but also reduces transmission delay, thereby maintaining data freshness.

传感器网络是以数据为中心的网络，受限于电池电量和处理能力的限制。传感器网络中的关键挑战之一是能量空穴问题。为了应对这一挑战，存在多种机制，其中将集群视为一种节能解决方案。通常，聚类是指基于空间排列相似性对节点进行分组的技术。适当的群集网络有助于在将信息路由到目的地之前处理和聚合感测到的数据。本文为高斯分布式传感器网络提出了一种基于软计算的递归方法，以实现基于密度的层次聚类。我们提出的概率方法递归地创建了层次聚类，