A Scheduling Framework for Latency Optimization on 5G Mobile Edge Computing Infrastructures

Abstract

Minimizing the latency in communications is assuming a paramount importance in 5G architectures, mainly for mission-critical applications associated to the IoT environment. However, the need of interacting with cloud-based applications introduces unacceptable delays essentially due to wide area transport activities over the Internet. Many solutions are emerging for facing this problem ranging from cloudlets to fog and edge computing architectures, pushing flexible virtualization environments nearer to the end devices, ideally within the base stations. Unfortunately, most of these infrastructures are managed by cloud providers in a network- oblivious way so that in presence of multiple nodes operating at the edge level in the same area no latency optimization strategies are taken into consideration. Accordingly, we present a novel latency- aware edge node selection framework based on a multi-objective bin packing problem transposed in the mobile edge computing scenario where the selection criterion is driven by both latency optimization and load balancing, managed according to a time slotted scheme.