Markovian models for harvested energy in wireless communications

Energy harvesting provides a perpetual but unreliable energy source for powering communication devices without access to a fixed power supply. The harvested energy can be treated as a stochastic process due to the random nature of the energy source, such as solar energy and piezoelectric (PZ) energy. Typically, a stationary Markovian (SM) model for the harvested energy is used in the literature for analysis. In this paper, we propose a generalized Markovian (GM) model that captures the non-stationarity of Markovian model, by introducing an additional scenario parameter. Given the scenario parameter (that typically changes slowly), the harvested energy (that typically changes quickly) then follows a stationary Markovian model. Based on the empirical measurements, we use the Bayesian information criterion to quantify whether the generalized model can better model the harvested energy. We conclude that the PZ energy can be better modeled by the GM model, while it is sufficient to model the solar energy by a SM model.

I. Introduction

Energy harvesting is a viable option as an perpetual energy source for powering communication devices in many practical scenarios. The devices may, for example, be located in remote locations without access to power supply and where it is inconvenient, dangerous, or even impossible to change the batteries. Examples of energy that can be harvested include solar energy, piezoelectric (PZ) energy and thermal energy.