Minimizing Delay in NFV 5G Networks by Means of Flexible Split Selection and Scheduling

It is well known that network function virtualization will be a key enabler to meet the stringent requirements of 5G networks. However, fully centralized approaches, such as Cloud Radio Access Network (C-RAN), might not be feasible, considering the particular needs of the fronthaul links and the large cost of implementing such architectural shift. In this sense, flexible functional split brings a practical solution, which trades off performance and practicability. In spite of the growing interest in flexible functional split, little attention has been paid to the interaction of split selection and scheduling. In this paper, we analyze joint strategies that minimize traffic delay. We compare the global optimum solution with partial optimizations, that can be more suitable in practical implementations, using different scenarios. According to our results, fixed scheduling behaves alike the global optimum in heterogeneous RAN scenarios. Furthermore, we observe that it is usually better to optimize the split degree for fixed scheduling setups, than deciding a scheduling policy for a particular split configuration.