Enabling Green and Delay-Aware 5G Edge Through Shared Cloud-Native Network Functions

Abstract

Satisfying the resource requirements of Network Functions (NFs) in an environmentally friendly and cost-effective way is a challenging task for cloud providers. The placement and resource allocation procedures for these NFs should address the Quality of Service (QoS) constraints, which result in ‘underutilized resources’. Network slicing offers solutions to reduce costs through resource sharing; however, sharing NFs, which obviously contributes to achieving QoS in a cost-effective way, has not been adequately considered in the literature. Moreover, the existing research on shared NF placement so far mostly uses NP-Hard algorithms. In this paper, we propose a cloud-native NF (CNF) sharing model and a heuristic (D-CNFSH) that is delay-aware and cost-effective for 5G edge networks by extending our previous work, CNFSH (Yılmaz and Alagöz, 2023), which focuses on NF sharing in the 5G core without considering delay. In addition, we introduce a performance isolation mechanism that maintains the high availability (HA) of high-priority slices while exploiting shared NFs. We also propose a novel metric, the ratio of underutilized resources to allocated resources, clearly showing the effect of sharing. This ratio is around 30% with NoShare schemes, while less than 0.4% is achieved with D-CNFSH, indicating a green 5G implementation that serves more network slices with less resource consumption.