Details

Abstrakt in Englisch

The increasing demand for wireless services and ubiquitous access comes to the mobile communications industry at the cost of a sizeable carbon footprint as well as a significant energy bill. A considerable amount of the total energy of network operators is used to run the radio access network, i. e., the base stations. This thesis develops theoretical foundations to understand, assess, and minimize energy consumption in mobile access networks. We first introduce a queueing-theoretic framework to analyze specifically the load-dependent dynamic part of the network energy. Among others, low-complexity approximation techniques for the so-called coupled processor queue (which is analytically intractable) are developed. Such queueing-theoretic models are fundamental for a numerically efficient system analysis in general, since they allow to compute a variety of performance indicators (e. g., power) for a variety of deployments and traffic conditions without resorting to elaborate dynamic system level simulations. We further characterize the relationship between network energy consumption per unit area and site densities, level of sectorization, as well as carrier frequencies for given coverage targets. Furthermore, we compare the average power requirements for two relevant strategies to expand existing networks: On the one hand, the deployment of low power nodes, called small cells, alongside conventional macro base stations, and, on the other hand, the installation of additional sectors at existing macro sites. The former strategy is heralded by network vendors and operators as a means to greatly increase network capacity and, at the same time, to yield significant energy savings compared to the latter. Our numerical results indicate, that small cell deployments lead to appreciable energy savings only in case of very high overall traffic demand or in case of very large differences between average and hotspot traffic intensity. In many cases, conventional deployments yield comparable energy consumption. In a last part, we extend existing approaches to optimize user association policies, i. e., the rules that identify locations in the network with individual cells. Such techniques are of practial importance to adapt cell areas (in particular small cells) to non-uniform traffic distributions, which improves the quality of service perceived by the users. Application of optimized user association policies is shown to have only small effects on the base station power consumption. We argue, however, that these techniques are key to sustain quality of service in scenarios where cells are deliberately switched off to save energy.