Abstract—This paper focuses on the problem of data sensor networks (WSNs). The data time window is a landmark window, from the time WSN starts working up to the current time. The objective is to group sensory data generated by sensor nodes deployed in a two¬dimensional physical space by the similarity of sensory data in the multi¬dimensional sensory data space. To perform in¬network data clustering efficiently, we propose HilbertMap, a novel dimensionality reduction technique based on the Hilbert Curves, to map a multi¬dimensional data space to a two-dimensional physical space. Through this mapping, the communications for clustering mostly occur between geographically nearby sensor nodes. We have conducted simulation experiments on both real¬world and synthetic datasets. Our results show that HilbertMap improves the communication efficiency while maintaining a good clustering quality.

Index Terms—Hilbert mapping, sensor networks, data clustering.

Cite: Longjiang Guo, Meirui Ren, Jinbao Li, Yong Liu, and Chunyu Ai, "H-luster: A Novel Efficient Algorithm for Data Clustering in Sensor Networks ," Journal of Communications, vol. 6, no.2, pp.168-178, 2011. Doi: 10.4304/jcm.6.2.168-178