• LTE-Advanced
• LTE
• Energy efficiency
• cellular networks
• resource allocation

In this work we describe and analyze resource scheduling solutions for energy efficiency in LTE and LTE-Advanced networks.
We start by presenting SimuLTE, a simulator for LTE and LTE-Advanced networks, based on OMNeT++ and developed in C++. It includes all the relevant protocol layers of the LTE stack, a realistic channel model and offers scheduling capabilities.
We then we start analyzing energy efficient solutions at various levels of the LTE network architecture. As a first step we focus on the user level: we describe the Discontinuous Reception (DRX) mechanism, that allows the central base station of an LTE cell to configure User Equipments for periodic wake/sleep cycles, so as to save energy. We investigate how to configure the parameters that defines DRX behavior and we explore the trade-off between energy saving and per-user QoS. We perform extensive simulations using realistic traffic models (i.e. VoIP, VoD, HTTP, YouTube-like) and factorial analysis for statistical soundness. We show that the above trade-off can be fine-tuned and that using less aggressive DRX settings prevent QoS degradation as the load increases.
The second step is performed at cell level: we propose a dynamic algorithm for interference coordination of base stations in heterogeneous networks. In such networks each cell has at least one macro node (i.e. a high power base station) and may have micro nodes (i.e. low power base stations). The LTE standard allows the macro node to define subframes as Almost Blank Subframes (ABS), where it may only transmit at low power, thus significantly reducing the interference suffered by micro nodes. ABS are arranged in patterns, configured by the macro node and notified to the micro. We propose an algorithm for dynamic selection of the ABS pattern, based on the instantaneous traffic request of both macro and micro nodes, and focused on energy efficiency. Our experimental results shows that this solution leads to a significant reduction of consumed energy at low loads, and better QoS performances at higher loads.
Finally we consider a whole network, i.e. a large set of adjacent heterogeneous cells controlled by the same operator. We present a framework to decide the activation and deactivation of micro nodes, based on load and energy efficiency considerations. Our framework relies on historical data (i.e. per-cell load curves) to select a set of candidate switch-on/switch-off instants, assuming a limited number of transitions is allowed in a day. The actual switching instants are detected online, by considering the traffic requests in each cell. We perform simulations with real-life traffic curves taken operator data. Our results show that this solution allows the network to cope with peak-hour traffic requests, and to reconfigure to a minimum coverage configuration whenever possible, thus saving energy (between 10 and 25% in our scenarios).