Capacity planning for 5G packet-based front-haul

Packet switched transport technologies have been suggested to be used in 5G front-haul networks. Front-haul bandwidth and packet latency are two major challenges to be addressed in these networks. This paper focuses on capacity planning for packet-based front-haul technologies considering the latency requirements of the 5G front-haul flows. More specifically, we answer three main questions for Constant Bit Rate (CBR) and Variable Bit Rate (VBR) packet front-haul networks: First, is statistical multiplexing gain always achievable in packet front-haul networks? Second, what is the minimum bandwidth required to accommodate a set of front-haul flows on a front-haul link? Third, what is the proper bandwidth allocation and the scheduling algorithm among the packets of different antennas to guarantee the latency requirements of the flows? It is shown that for CBR traffic no statistical multiplexing gain is achievable and the Earliest Deadline First (EDF) is the optimal packet scheduler. For VBR traffic, statistical multiplexing gain is achievable only when the latency requirement of at least one flow is greater than the packet inter-arrival time. The optimal bandwidth allocation is also determined for a Weighted Fair Queueing (WFQ) scheduler. Finally, simulation is used to validate the theoretical results for both CBR and VBR front-haul traffic.