Green Mobile Networks: from self-sustainability to enhanced interaction with the Smart Grid

Nowadays, the staggering increase of the mobile traffic is leading to the deployment of denser and denser cellular access networks, hence Mobile Operators are facing huge operational cost due to power supply. Therefore, several research efforts are devoted to make mobile networks more energy efficient, with the twofold objective of reducing costs and improving sustainability. To this aim, Resource on Demand (RoD) strategies are often implemented in Mobile Networks to reduce the energy consumption, by dynamically adapting the available radio resources to the varying user demand. In addition, renewable energy sources are widely adopted to power base stations (BSs), making the mobile network more independent from the electric grid. At the same time, the Smart Grid (SG) paradigm is deeply changing the energy market, envisioning an active interaction between the grid and its customers. Demand Response (DR) policies are extensively deployed by the utility operator, with the purpose of coping with the mismatches between electricity demand and supply. The SG operator may enforce its users to shift their demand from high peak to low peak periods, by providing monetary incentives, in order to leverage the energy demand profiles. In this scenario, Mobile Operators can play a central role, since they can significantly contribute to DR objectives by dynamically modulating their demand in accordance with the SG requests, thus obtaining important electricity cost reductions. The contribution of this thesis consists in investigating various critical issues raised by the introduction of photovoltaic (PV) panels to power the BSs and to enhance the interaction with the Smart Grid, with the main objectives of making the mobile access network more independent from the grid and reducing the energy bill. When PV panels are employed to power mobile networks, simple and reliable Renewable Energy (RE) production models are needed to facilitate the system design and dimensioning, also in view of the intermittent nature of solar energy production. A simple stochastic model is hence proposed, where RE production is represented by a shape function multiplied by a random variable, characterized by a location dependent mean value and a variance. Our model results representative of RE production in locations with low intra-day weather variability. Simulations reveal also the relevance of RE production variability: for fixed mean production, higher values of the variance imply a reduced BS self-sufficiency, and larger PV panels are hence required. Moreover, properly designed models are required to accurately represent the complex operation of a mobile access network powered by renewable energy sources and equipped with some storage to harvest energy for future usage, where electric loads vary with the traffic demand, and some interaction with the Smart Grid can be envisioned. In this work various stochastic models based on discrete time Markov chains are designed, each featuring different characteristics, which depend on the various aspects of the system operation they aim to examine. We also analyze the effects of quantization of the parameters defined in these models, i.e. time, weather, and energy storage, when they are applied for power system dimensioning. Proper settings allowing to build an accurate model are derived for time granularity, discretization of the weather conditions, and energy storage quantization. Clearly, the introduction of RE to power mobile networks entails a proper system dimensioning, in order to balance the solar energy intermittent production, the traffic demand variability and the need for service continuity. This study investigates via simulation the RE system dimensioning in a mobile access network, trading off energy self-sufficiency targets and cost and feasibility constraints. In addition, to overcome the computational complexity and long computational time of simulation or optimization methods typically used to dimension the system, a simple analytical formula is derived, based on a Markovian model, for properly sizing a renewable system in a green mobile network, based on the local RE production average profile and variability, in order to guarantee the satisfaction of a target maximum value of the storage depletion probability. Furthermore, in a green mobile network scenario, Mobile Operators are encouraged to deploy strategies allowing to further increase the energy efficiency and reduce costs. This study aims at analyzing the impact of RoD strategies on energy saving and cost reduction in green mobile networks. Up to almost 40% of energy can be saved when RoD is applied under proper configuration settings, with a higher impact observed in traffic scenarios in which there is a better match between communication service demand and RE production. While a feasible PV panel and storage dimensioning can be achieved only with high costs and large powering systems, by slightly relaxing the constraint on self-sustainability it is possible to significantly reduce the size of the required PV panels, up to more than 40%, along with a reduction in the corresponding capital and operational expenditures. Finally, the introduction of RE in mobile networks contributes to give mobile operators the opportunity of becoming prominent stakeholders in the Smart Grid environment. In relation to the integration of the green network in a DR framework, this study proposes different energy management policies aiming at enhancing the interaction of the mobile network with the SG, both in terms of energy bill reduction and increased capability of providing ancillary services. Besides combining the possible presence of a local RE system with the application of RoD strategies, the proposed energy management strategies envision the implementation of WiFi offloading (WO) techniques in order to better react to the SG requests. Indeed, some of the mobile traffic can be migrated to neighbor Access Points (APs), in order to accomplish the requests of decreasing the consumption from the grid. The scenario is investigated either through a Markovian model or via simulation. Our results show that these energy management policies are highly effective in reducing the operational cost by up to more than 100% under proper setting of operational parameters, even providing positive revenues. In addition, WO alone results more effective than RoD in enhancing the capability to provide ancillary services even in absence of RE, raising the probability of accomplishing requests of increasing the grid consumption up to almost 75% in our scenario, twice the value obtained under RoD. Our results confirm that a good (in terms of energy bill reduction) energy management strategy does not operate by reducing the total grid consumption, but by timely increasing or decreasing the grid consumption when required by the SG. This work shows that the introduction of RE sources is an effective and feasible solution to power mobile networks, and it opens the way to new interesting scenarios, where Mobile Network Operators can profitably interact with the Smart Grid to obtain mutual benefits, although this definitely requires the integration of suitable energy management strategies into the communication infrastructure management.