Des utilisateurs tolérants au délai - une solution pour l'efficacité énergétique de bout en bout
Samantha Gamboa · HAL (Le Centre pour la Communication Scientifique Directe) · 2015
Cellular networks have been traditionally designed to keep the network infrastructure always operational. This in order to ensure ubiquitous service availability and enough capacity to serve the peak of usage of the customers. Recently, the concern about the energy efficiency of this paradigm has increased, and a different approach has concentrated the research efforts of industry and academy. In this new paradigm, the infrastructure is dynamically adapted to the temporal and spatial traffic variations, reducing the energy wastage. The majority of these studies make the adaptation of the infrastructure unnoticeable to the users. However, with the appropriate interactivity and incentives, some users may be willing to offer their cooperation to the network. In this thesis we consider the user cooperation in the design and control of energy efficiency techniques. We consider a specific type of cooperation in which the users are able to offset the start of their services for a bounded and known-in-advance delay. Based on proactive interaction with the users, the network may ask them to delay the start of their services if an energy efficiency technique is applied in the area where they are located (e.g. a base station is switched off). Thus, a portion of the traffic is shifted and the network can optimize the resource utilization in order to consume less energy. First, we present an overview and classification of the literature covering the main domains of the thesis, namely energy efficiency in cellular networks and user demand shaping. We describe as well the most recent cellular network architecture - LTE. Then, we propose two different strategies to control the network resources depending on the cooperation of the users and their delay tolerance, and we evaluate the impact of such cooperation schemes in different energy efficiency techniques. Afterwards, we propose a theoretical framework for the analytical evaluation of the proposed strategies. We obtained the theoretical bounds of the attainable energy savings when employing different energy efficiency techniques, and we investigated the trade-off between the waiting time bounds proposed to the users and the energy gains. We observed that increased delay tolerance leads to more energy gains, and that the gains have an upper bound determined by the system serving capacity. We also noted that delaying opportunistically the user services depending on the system conditions is more beneficial than systematically delaying all of them. Finally, we evaluated the strategies under more realistic conditions using system level simulations. We corroborated the theoretical trends and we observed that the attainable gains are limited by the duration of the network reconfiguration process.