Enhancing the energy efficiency of radio base stations

This thesis is concerned with the energy efficiency of cellular networks. It studies the dominant power consumer in future cellular networks, the Long Term Evolution (LTE) radio Base Station (BS), and proposes mechanisms that enhance the BS energy efficiency by reducing its power consumption under target rate constraints. These mechanisms trade spare capacity for power saving. First, the thesis describes how much power individual components of a BS consume and what parameters affect this consumption based on third party experimental data. These individual models are joined into a component power model for an entire BS. The component model is an essential step in analysis but is too complex for many applications. It is therefore abstracted into a much simpler parameterized model to reduce its complexity. The parameterized model is further simplified into an affine model which can be applied in power minimization. Second, Power Control (PC) and Discontinuous Transmission (DTX) are identified as promising power-saving Radio Resource Management (RRM) mechanisms and applied to multi-user downlink transmission. PC reduces the power consumption of the Power Amplifier (PA) and is found to be most effective at high traffic loads. DTX mostly reduces the power consumption of the Baseband (BB) unit while interrupting transmission and is better applied in low traffic loads. Joint optimization of these two techniques is found to enable additional power-saving at medium traffic loads and to be a convex problem which can be solved efficiently. The convex problem is extended to provide a comprehensive power-saving Orthogonal Frequency Division Multiple Access (OFDMA) frame resource scheduler. The proposed scheduler is shown to reduce power consumption by 25-40% in computer simulations, depending on the traffic load. Finally, the thesis investigates the influence of interference on power consumption in a network of multiple power-saving BSs. It discusses three popular alternative distributed uncoordinated methods which align DTX mode between neighbouring BSs. To address drawbacks of these three, a fourth memory-based DTX alignment method is proposed. It decreases power consumption by up to 40% and retransmission probability by around 20%, depending on the traffic load