Base Station Operation For Energy Efficiency In Cellular Networks

With the increasing mobile terminals and the new mobile applications, the wireless traffic demand is growing dramatically in the cellular networks. Wireless operators need to improve the system capacity to support the new applications. It will, at the same time, introduce high energy consumption. It was expected that the traffic demand would have been increasing within decades. However, it is not acceptable if the energy consumption in cellular networks increases linearly with the traffic demands due to the system sustainability.In this thesis, we first study the multi-cell base station sleep mechanism. We try to utilize the sleep potential by a few information exchange between macro base stations in order to reduce the energy consumption. We formulate it as a combinatorial optimization problem to minimize the total energy consumption while guaranteeing users’ QoS. Based on the matching algorithm in game theory and a voting-based algorithm,we propose a decentralized sleep mechanism to minimize the total energy consumption and also to guarantee the user blocking ratio. Simulation results verify the proposed sleep mechanism by comparing it with other sleep schemes.Second, we analyze the base station sleep problem for heterogeneous users. In specific, we consider two types of users: rate guaranteed users have the outage probability requirement; best effort users want as high throughput as possible. In order to both minimize the total energy consumption and maximize the throughput of best effort users while guaranteeing the outage probability of rate guaranteed users, we formulate a joint optimization problem of base station density and the bandwidth ratio between the two types of uses. Using the stochastic geometry model, we characterize the properties of the optimization problem and then propose an iterative algorithm. Simulation results show that the iterative algorithm has good performance. We also analyze the impact of the system parameters. It shows that there is a tradeoff between the total energy consumption and the throughput of the best effort users, which can be adaptively controlled according to the current traffic in the system.