A framework for energy efficient NFV in 5G networks

Several critical benefits are encompassed by the concept of NFV when this concept is brought under the roof of 5G such as scalability, high level of flexibility, efficient utilisation of network resources, cost and power reduction, and on demand allocation of network resources. NFV could reduce the cost for installing and maintaining network equipment through consolidating the hardware resources. By deploying NFV, network resources could be shared between different users and several network functions in a facile and flexible way. Beside this the network resources could be rescaled and allocated to each function in the network. As a result, the NFV can be customised according the precise demands, so that all the network components and users could be handled and accommodated efficiently. In this paper we extend the virtualization framework that was introduced in our previous work to include a large range of virtual machine workloads with the presence of mobile core network virtual machine intra communication. In addition, we investigate a wide range of traffic reduction factors which are caused by base band virtual machines (BBUVM) and their effect on the power consumption. We used two general scenarios to group our finding, the first one is virtualization in both IP over WDM (core network) and GPON (access network) while the second one is only in IP over WDM network (core network). We illustrate that the virtualization in IP over WDM and GPON can achieve power saving around (16.5% – 19.5%) for all cases compared to the case where no NFV is deployed, while the virtualization in IP over WDM records around (13.5% – 16.5%).