International Journal of Engineering Research and Development e-ISSN: 2278-067X, p-ISSN: 2278-800X, www.ijerd.com Volume 13, Issue 3 (March 2017), PP.100-109 100 Energy Harvesting Using Adaptive Duty-Cycling Algorithm - Wireless Sensor Networks Sireesha Pendem 1 , Kancharla Suresh 2 1,2 Electonics and Communication Engineering Department, Swami Vivekananda Institute Of Technology ABSTRACT: With the wide spread use of wireless sensor network, the management of the energy resources has become a topic of reseaech. Wireless sensor nodes which harvest energy from the environment have become an to battery hooped up nodes. Requirements for economical use of the extracted energy led to development of algorithms that manage the node functions depending on the amount of collected energy. This article introduces a unique solution of adaptively setting the duty-cycle of a wireless sensor nodes so as to maximize its monitoring lifetime. The developed algorithms are particularly suited to energy harvesting wireless sensor networks situated in locations where energy is scarce or where harvested power exhibits ample diurnal or seasonal variation. The results described in this article shows that the proposed wireless sensor network architecture can represent a viable solution for monitoring indoor environments characterized by low illumination. The setup was tested and validated under various lighting conditions, using the adaptive techniques described in the paper. Keywords: wireless sensor networks; energy harvesting; power management; duty-cycling; supercapacitors; solar energy I. INTRODUCTION Wireless sensor networks (WSN) are traditionally powered using batteries. Although this method is acceptable for some applications, it is difficult to ensure maintenance in scenarios where nodes are placed in remote locations and the effort to replace their batteries becomes considerable. The main advantage of harvesting energy is the extended functional time of the node and lower costs due to eliminating the need of battery replacement. However, there are some issues to be addressed in design regarding harvested energy availability. A method to efficiently store the energy when it can actually be collected is needed. The unpredictability of energy levels has to be accounted for in the power management scheme of the sensor node. Although for solar there is a known cycle of day and night, the amount of energy can differ from one day to another, sunny and cloudy days, and this can affect the long term node operation. This article focuses on harvesting solar energy using small footprint photovoltaic panels and super capacitors as storage devices. The objective is to demonstrate that the hardware and firmware setup is a working solution for continuously monitoring indoor environment conditions with energy harvesting nodes. II. RELATED WORK In recent years several work pursued the design of efficient algorithms in energy harvesting sensor nodes. The article written by Kansal et al. [1] is one of the first works to study power management in energy harvesting WSN. It introduced Energy Neutral Operation (ENO) for energy harvesting nodes. This states that the energy consumption will be kept in balance with the gained energy over a defined time period, such that the node operates continuously. The proposed algorithm is splitting time in N equal slots per day. Inputs are the current harvested energy levels along with predicted harvesting values. Prediction on the amount of solar energy to be harvested is done using an Exponentially Weighted Moving-Average (EWMA) filter applied on the previously collected data. The duty-cycle is computed to compensate for the difference between the actual values and the ones predicted by the model. Vigorito et al. [2] propose a model-free approach to the duty-cycling problem by exploiting adaptive control theory techniques. A control algorithm is applied to the dynamic system, the harvesting node, with the objective of keeping the voltage within an interval centered on a target value. The equation to be solved is minimizing the quadratic cost function |output – target| 2 while keeping the ENO valid. After calculating the duty-cycle, a smoothing function is applied on the obtained value in order to lower the variance. It is considered that lowering the variance can be a requirement of certain WSN applications. In the work published by Cammarano et al. [3], a new prediction model named Pro-Energy is proposed, claiming an improvement of 60% over previous models such as EWMA or WCMA [4]. The disadvantage of EWMA was that the weight of the previous day data in estimating the energy intake for the current day was too large, leading to prediction errors when sunny and cloudy days were alternating.Hsu et al. [5], introduce a system model view of an energy harvesting node and a theoretical framework to calculate the optimal power