A study on Modeling of Dependency between Configuration Parameters and Overall Energy Consumption in Wireless Sensor Network (WSN)

In this paper, we study a new approach to model the overall Energy Consumption (EC) in Wireless Sensor Networks (WSN). First, we extract parameters involving in the EC of WSNs. The dependency between configuration parameters and the average residual energy of a specific application is then investigated. Our approach has three key steps: profiling, parameter reduction, and modeling. In profiling, a sensor network simulator is re-run 800 times with different values of the configuration parameters in order to profile the average residual energy in nodes. In the parameter reduction, three statistical analyses (p-value, linear and non-linear correlation) are applied to the outcome of profiled experiments in order to separate the effective parameters on WSN residual energy. Finally, linear regression is used to model the relation between the chosen effective parameters and the residual energy. The evaluation based on running the simulator for another 200 times with different values of the effective parameters shows that the model can predict the residual energy of nodes in WSN with average error of less than 13%.