Energy Efficient Network Function Virtualisation in 5G Networks

Once the dust settled around 4G, 5G mobile networks become the buzz word in the world of communication systems. The recent surge of bandwidth-greedy applications and the proliferation of smart phones and other wireless connected devices has led to an enormous increase in mobile traffic. Therefore, 5G networks have to deal with a huge number of connected devices of different types and applications, including devices running life-critical applications, and facilitate access to mobile resources easily. Therefore given the increase in traffic and number of connected devices, intelligent and energy efficient architectures are needed to adequately and sustainably meet these requirements. In this thesis network function virtualisation is investigated as a promising paradigm that can contribute to energy consumption reduction in 5G networks. The work carried out in this thesis considers the energy efficiency mainly in terms of processing power consumption and network power consumption. Furthermore, it considers the energy consumption reduction that can be achieved by optimising the locations of virtual machines running the mobile 5G network functions. It also evaluates the consolidation and pooling of the mobile resources. A framework was introduced to virtualise the mobile core network functions and baseband processing functions. Mixed integer linear programming optimisation models and heuristics were developed minimise the total power consumption. The impact of virtualisation in the 5G front haul and back haul passive optical network was investigated by developing MILP models to optimise the location of virtual machines. A further consideration is caching the contents close to the user and its impact on the total power consumption. The impact of a number of factor on the power consumption were investigated such as the total number of active users, the backhaul to the fronthaul traffic ratio, reduction/expansion in the traffic due to baseband processing, and the communication between virtual machines. Finally, the integration of network function virtualisation and content caching were introduced and their impact on improving the energy efficiency was investigated