Adaptive Filter for Energy Predictor in Energy Harvesting Wireless Sensor Networks

To design an autonomous Wireless Sensor Network (WSN), the harvested energy from environmental sources has been considered as a potential solution for long-term operations. A power manager embedded in the energy harvesting WSNs adapts the power consumption and computation loads according to the harvested energy to obtain a theoretically infinite lifetime. In order to design an effective power manager, it is of prime interest to benefit from an accurate energy predictor to estimate the energy that can be harvested in the near future. In this paper, a low complexity energy predictor using adaptive filter in proposed. Our predictor has a low memory requirement as it is only based on a previous historical harvested energy to estimate the future energy. Simulation results show that our energy predictor using adaptive filter can be applied for both solar and wind energy with an average error less than 15%.