A Supervised Learning Approach for Reducing Latency during Context Switchover in 5G MEC

摘要

Multi-Access Edge Computing (MEC) is an important 5G paradigm, which allows application servers to be deployed in the ‘edge’ of the network, which drastically reduce latency and optimize network bandwidth. Context switchover from one edge application server (EAS) to another ensures service continuity for the UE during mobility. EAS trigger context switchover upon reception of notification about UE movement from Edge Enabler Server (EES). Sooner the notification to EAS better the performance of context switchover but the challenge is to arrive at optimum order of EAS in a very dense deployment. In this paper, we propose a novel method to determine the notification order of EAS to meet latency and QoS requirements of 5G. We make use of supervised learning models to sequence the EASs to be notified. The learning models simulated show that the complexity of ordering the EAS is reduced as much as 77% when compared to a linear sorting algorithm.