End-to-end network slicing for edge computing optimization

Abstract

User-centric services proliferated by the smart devices is getting more demanding and characteristically diversified. Fall-risk assessment, augmented reality and similar services coexist in a shared heterogeneous setting. To meet the diversified and often conflicting requirements of the services, the physical network is decomposed into virtual slices. In order to address the optimal network slicing problem for various service types with different performance requirements, this study proposes a Mixed Integer Programming (MIP) model. This optimization model aims to satisfy the demands of the services and provide complete isolation among them through virtual resources reservation, including both networking and computation. Additionally, a heuristic implementation NESECS (NEtwork Slicing for Edge Computing Services) is proposed as an efficient solution for the cases where the optimization tools remain inadequate. The performance of the proposed solutions is evaluated with an extensive set of experiments. The obtained results indicate that the proposed optimization model is capable of providing optimal or near optimal solutions for small network instances, and NESECS algorithm can provide good solutions for larger instances by eliminating the time complexity.