In this paper we provide a distributed implementation of the k-means clustering algorithm, assuming that each node in a wireless sensor network is provided with a vector representing an observation (e.g., sensorial data, position or both). The proposed algorithm, by means of one-hop communication only, partitions the nodes into groups that have small in-group and large out-group distances. Since the partitions may not have a direct relation with the topological clusters, an algorithm for the computation of the centroid of a sparse subset of the nodes is provided. Depending on the nature the observation of each node, it is possible to obtain a position clustering, a measure clustering or a hybrid clustering.
View on arXiv