Fusion model for statistically dependent observations in wireless sensor networks
Sunayana Jadhav, Rohin D. Daruwala · 2016
High density of sensors in Wireless Sensor Networks (WSN) causes sensor observations to be highly correlated in space and time domain. Measurements received at sensors can be characterized by multivariate distribution. Fusion of Statistically Dependent observations in WSN has become a major concern. An approach addressing fusion of dependent data in a WSN environment is done for two sensor design case. This paper extends fusion rule for three and four sensors using Bahadur-Lazarsfeld's polynomial expansion. Modeling includes evaluation of fusion rule and corresponding weighting factors. Joint sensor probabilities are further obtained from weighting factors. Fusion model illustrates that statistically dependent observations are computationally expensive as compared to those of independent ones.