Analysis of Latency-Aware Network Slicing in 5G Packet xHaul Networks

Abstract

Packet-switched xHaul networks are a scalable solution enabling convergent transport of diverse types of radio data flows, such as fronthaul/midhaul/backhaul (FH/MH/BH) flows, between remote sites and a central site (hub) in 5G radio access networks (RANs). Such networks can be realized using the cost-efficient Ethernet technology, which enhanced with time-sensitive networking (TSN) features allows for prioritized transmission of latency-sensitive fronthaul flows. Provisioning of multiple types of 5G services of different service requirements in a shared network, commonly referred to as network slicing, requires adequate handling of transported data flows in order to satisfy particular service/slice requirements. In this work, we investigate two traffic prioritization policies, namely, flow-aware (FA) and latency-aware (LA), in a packet-switched xHaul network supporting slices of different latency requirements. We evaluate the effectiveness of the policies in a network-planning case study, where virtualized radio processing resources allocated at the processing pool (PP) facilities, for two slices related to enhanced mobile broadband (eMBB) and ultra-reliable low latency communications (URLLC) services, are subject to optimization. Using numerical experiments, we analyze PP cost savings from applying the LA policy (vs. FA) in various network scenarios. The savings in active PPs reach up to 40%-60% in ring scenarios and 30% in a mesh network, whereas the gains in overall PP cost are up to 20% for the cost values assumed in the analysis.