Towards Intelligent Energy-Aware Self-Organised Cellular Networks (iSONs)

This thesis investigates the application of intelligent energy-aware resource management techniques for current and future wireless broadband deployments.

Energy-aware topology management is firstly studied aiming at dynamically managing the network topology by fine tuning the status of network entities (dormant / active) to scale the energy consumption with traffic demands. This is studied through an analytical model based on queueing theory and through simulation to help understand its operational capabilities under a range of traffic conditions. Advanced radio resource management is also investigated. This reduces the number of nodes engaged in the service whenever possible reducing the energy consumption at low and medium traffic loads while enhancing system capacity and QoS when the traffic load is high. As an enabling technology for self-awareness and adaptability, Reinforcement Learning (RL) is applied to manage network resources in an intelligent, self-aware, and adaptable manner. This is complemented with a range of novel cognitive learning and reasoning algorithms which are capable of translating past experience into valuable sets of information in order to optimise decisions taken as part of the radio resource and topology management functionalities. Dependencies between the proposed techniques are also addressed formulating an intelligent self-adaptable approach, which is capable of dynamically deactivating redundant nodes and redirecting traffic appropriately while enhancing system capacity and QoS.