Energy Prediction for Energy Management Method of Sensor Node Powered by Temperature Difference Between Air and Shallow Underground Soil

We present an energy prediction model for the energy management of wireless sensor networks powered by temperature differences between near-surface air and shallow underground soil (TDASU). Energy harvesting (EH) using TDASU differs from other EH technologies as follows: (i) TDASU occurs throughout the day. (ii) TDASU has half-day periodicity in addition to the one-day periodicity common for solar energy. Our energy prediction model, denoted the TDASU-characterized moving average (TCMA), considers these characteristics of TDASU and refines the conventional prediction model in terms of half- and one-day forecasting. To assess the performance of our proposed solution, we used real TDASU data collected from fields in Japan and India. A comparative performance evaluation of TCMA and the conventional prediction method has shown that the former can reduce prediction errors by up to 25%.