Implementation of a consumption and available energy estimator for WSN simulator

Abstract

Networked Embedded System (NES) is a fast growing application for embedded systems, in particular, for Wireless Sensor Networks (WSN). WSNs pose a set of challenges that should be address, such as the limited available energy. It is very important at the early stages of a WSN project to estimate the consumption and available energy of each sensor node and the whole network. Many WSN simulators have been developed in the recent past, but some simulators were not conceived for hardware modeling and they do not simulate the same application code used in the real sensor node, which are very important in order to estimate the consumption and available energy. Besides, WSN simulators have implemented only very simple and linear battery models and a realistic battery model have not been implemented yet. Therefore, it is very important to find out differences and similarities of WSN simulators and select the right one for a given application. Some taxonomies were proposed, but they addressed mostly the network and software domains only. This work proposes a new taxonomy for WSN simulators that is based on modeling four domains that WSN simulators must address, namely, the sensor node, the network, the environment and the energy. Based on the proposed taxonomy, the COOJA simulator was selected to implement a new energy consumption estimator and a realistic and non-linear battery model for alkaline batteries that incorporates the relaxation effect, discharge rate and capacity retention. This work also extends the COOJA simulator providing a new version that estimates the consumption and available energy of sensor nodes and the whole WSN. The proposed extended version of Cooja presents an average error less than 4.5% for the final remaining capacity and sensor node lifetime estimative.