Energy-prediction scheduler for reconfigurable systems in energy-harvesting environment

Energy harvesting has been demonstrated to be a promising approach to mitigating energy constraints. Unlike battery-based energy, available system energy significantly varies for energy-harvesting systems. Partial dynamic reconfiguration is adopted as an effective approach for accelerating wireless sensor network (WSN) applications. Although a reconfigurable system can achieve better performance compared with software implementation, reconfiguration potentially requires a large amount of energy and time, particularly for cases where reconfiguration occurs frequently. To address this issue, a novel weather-aware scheduler based on the weather-conditioned moving average (WCMA) prediction algorithm is proposed in this study. To demonstrate the authors approach, a heterogeneous reconfigurable node is also proposed. The implementation of the proposed approach can improve reconfigurable system performance by up to 50% under energy-harvesting conditions. The novelty of this work is 2-fold. First, a prototype is adopted to demonstrate its efficiency for WSN. Second, a novel scheduler is proposed to manage hardware reconfiguration. In the scheduler, WCMA is used to predict future harvested energy.