A Hierarchical Game Framework for Resource Management in Fog Computing

Abstract

Supporting real-time and mobile data services, fog computing has been considered as a promising technology to overcome long and unpredicted delay in cloud computing. However, as resources in FNs are owned by independent users or infrastructure providers, the ADSSs cannot connect and access data services from the FNs directly, but can only request data service from the DSOs in the cloud. Accordingly, in fog computing, the DSOs are required to communicate with FNs and allocate resources from the FNs to the ADSSs. The DSOs provide virtualized data services to the ADSSs, and the FNs, motivated by the DSOs, provide data services in the physical network. Nevertheless, with fog computing added as the intermediate layer between the cloud and users, there are challenges such as the resource allocation in the virtualized network between the DSOs and ADSSs, the asymmetric information problem between DSOs and ADSSs, and the resource matching from the FNs to the ADSSs in the physical network. In this article, we propose a three-layer hierarchical game framework to solve the challenges in fog computing networks. In the proposed framework, we apply the Stackelberg sub-game for the interaction between DSOs and ADSSs, moral hazard modeling for the interaction between DSOs and FNs, and the student project allocation matching sub-game for the interaction between FNs and ADSSs. The purpose is to obtain stable and optimal utilities for each DSO, FN, and ADSS in a distributed fashion.