Cost-Effective Resource Allocation for Multitier Mobile Edge Computing in 5G Mobile Networks

Abstract

Mobile edge computing (MEC) is currently one of the key technologies that can facilitate the evolution of the future digitized economy. MEC can provide ubiquitous computational capabilities through the multitier deployment of servers to ensure lower latencies and tighter integration with 5G, the Internet of Things, blockchains and artificial intelligence. In this paper, we propose a new approach to optimizing hardware resource allocation for edge nodes in a multitier MEC hierarchy. In addition to a centralized unit, we consider active antenna units and distributed units equipped with edge nodes of different computational capacities. A parametric Bayesian optimizer is implemented for hardware resource allocation to increase the overall computational capacity of a 5G-based MEC system. Simulation results show that for given budget constraints, the proposed solution outperforms pseudorandom resource allocation in terms of the proportion of computational tasks completed. The achievable gains are in the range of 20-40 %, depending on the task complexity and selected budget threshold.