Renewable energy management in cellular networks: An online strategy based on ARIMA forecasting and a Markov chain model

In this paper, we propose an online energy management strategy to minimize the operational expenses incurred by cellular base stations powered by both renewable and conventional energy. Our proposed strategy uses an Auto Regressive Integrated Moving Average (ARIMA) time series model for inter-day forecasting, a Markov chain model for intra-day predictions, and linear programming techniques for optimizing the decision variables on a real-time basis. To the best of our knowledge, the potential of these techniques has not been sufficiently explored in the literature. We assume that the base station is equipped with a rechargeable battery and a solar panel. Moreover, we consider real-time electricity pricing, and the application of a net-metering policy, whereby consumers are allowed to add excess renewable energy to the grid and obtain kilowatt credits in return. To evaluate the performance of the proposed algorithm, we devise an offline strategy which assumes non-causal knowledge of renewable energy generation and hence provides an upper bound in performance. Finally, we present numerical results obtained using real meteorological data, practical solar panel specifications, and factual energy tariffs. Through simulations we benchmark the proposed algorithm against the genie-aided strategy, and show its robustness by considering random energy rates.