Battery state-of-charge approximation for energy harvesting embedded systems

Batteries play an integral role in Wireless Sensor Networks as they provide the energy necessary to operate the individual sensor nodes. In order to extend the network's lifetime, and theoretically permit continuous operation even for systems with high-energy consumption, environmental energy harvesting has attracted much interest. It has been shown that the motes' utility can be improved significantly if run-time knowledge of remaining battery capacity is available. In this work, a light-weight and cost effective approach to approximating the battery state-of-charge (SOC) based on voltage measurements is presented. Despite commonly perceived as inferior to other approaches, a performance evaluation shows that SOC approximations with over 95% accuracy are possible. It is further shown that battery inefficiencies due to e.g., temperature and aging are taken into consideration despite not explicitly modeling these effects. The approach only requires system input voltage measurements, but benefits from optional current and temperature measurements.