ACM Transactions on Sensor Networks (TOSN)
| Hi-index | 0.00 |
Nowadays, many sensor network applications are not only interested in the raw data of a single sensor node, but also the overview distribution features of network-wide sensory data. Data-centric clustering can provide an overview of the sensory data distribution. However most data-centric clustering researches are based on snapshot data rather than evolving data, which is not feasible as data evolves. In this paper, we propose a novel clustering method in wireless sensor networks called MCC: Model-based Continuous Clustering. In MCC, we do the clustering in a continuous way to make every node belong to the most suitable cluster at every epoch which guarantees that the clustering structure can represent the up-to-date data distribution to the most extent. Moreover, we build a model for every sensor node based on its recent data to capture the data evolving trend. MCC is based on these models. The extensive experiments on real-datasets show the effectiveness and efficiency of MCC