An Energy-efficient Routing Protocol for Wireless Sensor Networks through Estimating Single Linkage Hierarchical Clustering

Hierarchical clustering (HC) algorithms are widely used in wireless sensor network (WSN) routing protocols, but few of these applications take into account noise in the data despite its ubiquity in situations where noise is evidently present, such as in wireless communications. We propose an energy-efficient HC routing protocol relying on an estimator of SLHC that takes into account the statistical model of noise in the data generation process: We construct a pseudo-likelihood model taking SLHC as a parameter to be estimated from noisy input data, and use Monte Carlo Expectation Maximization algorithm to generate a maximum likelihood estimator (MLE), according to which the routing protocol is determined. The simulations validate the strong performance of our method in terms of energy efficiency, especially for high noise.