A Framework to Define a Measurement-Based Model of MEC Nodes in the 5G System

Abstract

Multi-access edge computing (MEC) represents an emerging solution to address the issues related to reliability, availability, context awareness and low latency in modern and future mobile networks. Despite the great interest and effort in developing MEC solutions, only a few works are available in the literature to model the existing trade-offs between available resources and the related performance of MEC nodes. This paper aims to fill this gap by proposing a measurement-based approach to indirectly derive an effective model for any type of MEC node, capable of capturing the existing trade-offs among different performance parameters. Basically, the idea is to characterize the input-output relationship between offered computational and communication load and achievable performance. A network emulator is used to generate the data to define the model. The results provide proof-of-concept of idea by analyzing the different trade-offs and achievable performance of a sample MEC node under CPU and networking limitations.