Section: New Results

Green Networking and Smart Grids

Participants : Sara Alouf, Eitan Altman, Alberto Benegiamo, Alain Jean-Marie, Giovanni Neglia.

In [35] , Rodrigo A. Vaca Ramirez and John S. Thompson (Univ. of New England), in collaboration with Eitan altman and Victor Ramos Ramos (UAM - Univ. Autonoma Metropolitana Unidad Iztapalapa) consider a low complexity virtual Multiple-input Multiple-output (MIMO) coalition formation algorithm. The goal is to obtain improvements in energy efficiency by forming multi-antenna virtual arrays for information transmission in the uplink. Virtual arrays are formed by finding a stable match between single antenna devices such as mobile station (MS) and relay stations (RS) by using a game theoretic approach derived from the concept of the college admissions problem. They focus on enhancing the MS performance by forming virtual coalitions with the RSs. Thus, power savings are obtained through multi-antenna arrays by implementing the concepts of spatial diversity and spatial multiplexing for uplink transmission. They focus on optimizing the overall consumed power rather than the transmitted power of the network devices. Furthermore, it is shown analytically and by simulations that when overall consumed power is considered as an optimization metric, the energy efficiency of the single antennas devices is not always improved by forming a virtual MIMO array. Hence, single antenna devices may prefer to transmit on their own when channel conditions are favorable. In addition, the simulation results show that the proposed framework provides comparable energy savings and a lower implementation complexity when compared to a centralized exhaustive search approach that is coordinated from the Base Station.

Sara Alouf, Ioannis Dimitriou (now at Univ. Patras, Greece) and Alain Jean-Marie had worked on the modeling of wireless communication base stations with autonomous energy supply (solar, wind). They had proposed a versatile 5-dimensional Markov model of the device, and shown that the Quasi Birth-Death framework is adequate for solving the model. This work has been completed with a companion product-form model based on E. Gelenbe's modeling of energy networks with signals [48] .

In [16] , Eitan Altman in collaboration with Cengis Hasan, Manjesh Kumar Hanawal (IIT Mumbai), Shlomo Shamai (Technion), Jean-Marie Gorce (Inria project-team Socrate ), Rachid El-Azouzi (UAPV) and Laurent Roullet (Alcatel Lucent Bell Labs) study both the uplink and downlink energy efficiency based on the assumption that base stations are distributed according to an independent stationary Poisson point process. This type of modeling allows to make use of the property that the spatial distribution of the base stations after thinning (switching-off) is still a Poisson process. This implies that the probability of the SINR can be kept unchanged when switching-off base stations provided that one scales up the transmission power of the remaining base stations. The authors then solve the problem of optimally selecting the switch-off probabilities so as to minimize the energy consumptions while keeping unchanged the SINR probability distribution. They then study the trade-off in the uplink performance involved in switching-off base stations. These include energy consumption, the coverage and capacity, and the impact on amount of radiation absorbed by the transmitting user.

Energy demand and production need to be constantly matched in the power grid. The traditional paradigm to continuously adapt the production to the demand is challenged by the increasing penetration of more variable and less predictable energy sources, like solar photovoltaics and wind power. An alternative approach is the so called direct control of some inherently flexible electric loads to shape the demand. Direct control of deferrable loads presents analogies with flow admission control in telecommunication networks: a request for network resources (bandwidth or energy) can be delayed on the basis of the current network status in order to guarantee some performance metrics. In [53] G. Neglia, in collaboration with G. Di Bella (Telecom Italia, Italy), L. Giarré and I. Tinnirello (Univ. of Palermo, Italy) go beyond such an analogy, showing that usual teletraffic tools can be effectively used to control energy loads. In particular they propose a family of control schemes which can be easily tuned to achieve the desired trade-off among resources usage, control overhead and privacy leakage.

The massive introduction of Electric Vehicles (EVs) will make fleet managers spend a significant amount of money to buy electric energy. If energy price changes over time, accurate scheduling of recharging times may result in significant savings. In [29] C. Rottondi (IDSIA Dalle Molle Institute for Artificial Intelligence, Switzerland), G. Neglia and G. Verticale (Politecnico di Milano, Italy) evaluate the complexity of the optimal scheduling problem considering a scenario with a fleet manager having full knowledge of the customers’ traveling needs at the beginning of the scheduling horizon. They prove that the problem has polynomial complexity and provide complexity lower and upper bounds. Moreover, they propose an online sub-optimal scheduling heuristic that schedules the EVs’ recharge based on historical travelling data. They compare the performance of the optimal and sub-optimal methods to a benchmark online approach that does not rely on any prior knowledge of the customers’ requests, in order to evaluate whether the additional complexity required by the proposed strategies is worth the achieved economic advantages. Numerical results show up to of 35% cost savings with respect to the benchmark approach.