Migration-Aware Optimized Resource Allocation in B5G Edge Networks

Abstract

The fifth-generation and beyond (B5G) communication systems are evolving for computation-intensive and communication-sensitive applications with diverse Quality-of-Service (QoS) requirements on processing, bandwidth, latency, and reliability. This work focuses on an ultra-dense edge network with Multi-access Edge Computing (MEC) facilities, serving agents that execute their tasks by touring the cells. Specifically, we propose a novel methodology for optimally and flexibly managing task offloading in the context of heterogeneous computing and communication services required by real-time robotic applications. Differing from many related work, the proposed approach takes the number of admitted service migrations and the QoS upper and lower bounds as binding constraints. We model the QoS evolution based on the agent positions, the MEC servers serving the agents, the QoS requirements, the communication capabilities in the edge network, and the computing capabilities of the servers. The model is formalized as a mixed-integer linear program (MILP) to obtain an optimal schedule for the service migrations and communication and computation bandwidth allocation. Experimental results show that the approach outperforms baseline approaches and can scale to large deployments.