Tri-Variate Copula Modeling for Spatially Correlated Observations in Wireless Sensor Networks

摘要

Correlated Observations arise in Wireless Sensor Networks (WSNs) comprising of crowded sensor nodes monitoring a common physical phenomenon. Correlation exists both in spatial and time domain, numerous models have addressed linear dependency in sensor observations. However, Copulas model both linear as well as non-linear dependency in spatial domain. In this paper we have proposed a fusion model for generalized case using Copulas and evaluated it for a tri-variate case. A 3D Copula model previously introduced is computed and analyzed based on Neyman-Pearson framework. Gaussian and Student-t Copulas demonstrate a superior performance for spatially correlated observations as compared to Chair-Varshney rule for independent observations.