PiEdge: An Edge-Driven PaaS Model for Network Slicing Automation

A recently proposed direction that is pursued by standardization organizations for 5G Network Function Virtualization (NFV) technologies at the core of next-generation networks (5G and beyond) is the adoption of “Cloud-native” design principles. These principles are catalyzed by Platform as a Service (PaaS) solutions that allow re-usability of common or dedicated Virtual Network Functions, in order to fulfill the diverse vertical requirements for network slices. PaaS solutions can also be extended to Multi-access Edge Computing environments (i.e. at the edge of the cellular network) to take advantage of predefined edge-based network functions and services and to decouple their lifecycle management from centralized, delay-introducing operations. Currently, core-to-edge coordination for end-to-end network slicing, from 5G Core Network to the edge, takes place at a very low abstraction level, leading to significant overheads that pose scalability and latency threats, especially crucial for mission-critical applications. To tackle this challenge, we describe a methodology for automating the interaction between the NFV MANO and edge platforms that encompass PaaS functions at the edge. The methodology offers benefits for 1) eliminating configuration, integration and management effort for end-to-end network slicing, and 2) the exposure of edge resources to the NFV MANO as usable platform services. We showcase the automation and management overhead benefits of our solution through experiments on an edge-based Internet of Things (IoT) use case, deployed at the 5G-VINNI Spanish 5G Facility site.