Game-Based Task Offloading in Cloud-Edge-End Network

Abstract

Mobile edge computing (MEC) is a key enabler for the next generation network by leveraging the cloud function to network edge and reducing costs required to transmit data to the cloud. In this paper, we consider a cloud-edge-end network, in which users subscribe services to the cloud, and the cloud delegates the services to edge servers since they can support time-sensitive applications locally. All the cloud, edge servers, and users intend to maximize their utilities. To model the dynamic interactions among cloud-edge-end, we propose a hierarchical dynamic game framework. Specifically, in the lower-level game, the offloading decisions of users are modeled as an evolutionary game. In the upper-level game, considering the users' dynamic selections, we model the pricing strategy of the cloud and resource allocations of the edge servers as a Stackelberg differential game. The equilibrium and the optimal strategies of the players in the game are derived. The simulation results validate the effectiveness of our designed reward operation mechanism.