Latency Evaluation of a Virtualized MME

TL;DR

This paper presents a queue model to evaluate latency in virtualized LTE Mobility Management Entities, aiding in optimal scaling and deployment for future mobile networks.

Contribution

It introduces a queue model for virtualized MME latency estimation and a traffic model for future mobile applications, supporting deployment decisions.

Findings

The model accurately predicts service time under various workloads.

Simulation results guide optimal number of virtual instances.

The traffic model forecasts control workload for future applications.

Abstract

Network Virtualization is one of the key technologies for developing the future mobile networks. However, the performance of virtual mobile entities may not be sufficient for delivering the service required for future networks in terms of throughput or service time. In addition, to take advantage of the virtualization capabilities, a criterion to decide when to scale out the number of instances is a must. In this paper we propose an LTE virtualized Mobility Management Entity queue model to evaluate its service time for a given signaling workload. The estimation of this latency can serve to decide how many processing instances should be deployed to provide a target service. Additionally, we provide a compound data traffic model for the future mobile applications, and we predict theoretically the control workload that it will generate. Finally, we evaluate the virtualized Mobility Management Entity overall delay by simulation, providing insights for selecting the number of virtual instances for a given number of users.