Dynamic energy management of energy harvesting wireless sensor nodes using fuzzy inference system with reinforcement learning

This paper considers a scenario that a wireless sensor node is powered by harvesting energy from ambient solar energy and required the sustainable operation. Fuzzy inference system with reinforcement learning is used in this study for the dynamic energy management of the node in the energy harvesting environment. By interacting with the given environment, the proposed method adjusts the duty cycle of the sensing task in the node according to the proposed method. The outcomes of these interactions are evaluated by fuzzy inference system in terms of reward that express how well the duty-cycle adjustments in satisfying the given requirement of energy neutrality. Experimental results show that the proposed method obtained better convergence and less root mean square deviation off energy neutral level in comparing with other methods.