This thesis examines how sleep modes can be used in next generation mobile broadband systems (BuNGee) to substantially improve energy efficiency. Such systems employ a high degree of overlapping small cells in order to deliver very high throughput density for 5G networks in dense areas. It is shown how by limited exchange of information between neighbouring base stations it is possible to maintain quality of service (QoS), over a range of traffic loads, while enabling inactive base stations to sleep. Dynamic distributed topology management schemes are used here to switch off the small cells at low traffic load levels, while the remaining local traffic can be forwarded by their adjacent cells.
A novel analytical model is generated using multi-dimensional Markov processes and is used to predict the theoretical system performance and potential energy reduction when a set of parameters are varied. The parameters discussed include the traffic load thresholds to switch off/on a base station. This new model provides an understanding of how to obtain the maximum energy reduction while guaranteeing QoS by choosing suitable parameter values in such a network.
Performance of distributed energy efficient topology management schemes with a sleep mechanism are compared against the system without topology management. Results show the schemes deliver a significant energy reduction in energy consumption in the network, which is 35%-70% depending on the strategies used.
The corresponding simulation models are used to verify the analytical model. It is shown how traffic load based thresholds (used to switch on/off base stations) measured on adjacent base stations have a higher impact than the threshold on the base station itself. The latter threshold has very limited influence on the system energy efficiency.
An energy efficient topology management scheme employing combined sleep modes with handover for a BuNGee system is investigated as a way of providing further improvements to energy efficiency. Performance is examined using both analytical and simulation based models. A key aspect of this scheme is that a base station can redeploy its traffic load to its neighbours and then switch off itself when the local traffic is at a low/medium level.
