Stackelberg Game-Based Pricing and Offloading in Mobile Edge Computing

Owing to the limited computing resources at the MEC server, reasonable strategies of resources pricing and task offloading are necessary to be designed. In this letter, a scenario of multi-users and single MEC server with limited computing resources is studied, where each end-user has a divisible task which could be offloaded portion to the MEC server for execution, both local and offloaded portions of a task cannot be instantiated for execution at the same time, and a stackelberg game based pricing and offloading is investigated. The game is established based on the relationship between the MEC server resources pricing and task offloading quantities, and the existence of the stackelberg equilibrium is proved. Differential evolution algorithm is then used to find the best strategies of the resources pricing for the MEC server and task offloading quantities for end-users. Simulation results show that the investigations can improve the profit of the MEC server and the utility of end-users, which result in the win-win situation.

I. Introduction

Mobile Edge Computing (MEC) can achieve lower serving latency by deploying the MEC service close to end-users [1]. However, due to the limited computing resources and strong competitive relation, how to develop a reasonable pricing strategy for the MEC server and how to determine optimal task offloading quantities among end-users have been challenging issues [2]–[4]. The issue of MEC server resources allocation and pricing is usually analyzed form the perspective of revenue and profit management. In [5], two dynamic pricing schemes were designed and analyzed to maximize the MEC server profit. By considering the limitations in some existing solutions, three dynamic pricing strategies for resources allocation were thoroughly discussed to give MEC service providers guidance on achieving the best profit [6]. The issue of determining reasonable task offloading quantity for the end-user is usually formulated as a NP-hard problem. By considering a multi-users MEC-enabled system, joint offloading decision and resources allocation was investigated in [7] to maximize the number of offloaded tasks for end-users while maintaining the MEC server resources at an acceptable level. In the single-MEC and multi-MEC systems with the dynamic communication environment, a joint optimization problem of task offloading and resource allocation was formulated to minimize the energy consumption of each end-user subject to the latency requirement and the limited resources [8].