A Multi-tier Cost Model for Effective User Scheduling in Fog Computing Networks

In this paper, we investigate a cost model and the resulted cost-minimization user scheduling problem in multi-tier fog computing networks. For a typical multi-tier fog computing network consisting of one fog control node (FCN), multiple fog access nodes (FANs) and user equipments (UEs), how to model the cost paid to FANs for motivating resources sharing, and how to effectively schedule UEs to minimize the overall cost for FCN, are still problems to be resolved. To solve these problems, a unified multi-tier cost model, including the service delay and a linear inverse demand dynamic payment scheme, is proposed, and a cost-minimization user scheduling problem is formulated. Further, the user scheduling problem is reformulated as a potential game and proven to possess a nash equilibrium (NE) solution. The corresponding distributed algorithm, namely Cost-Oriented User Scheduling (COUS) algorithm, is developed to achieve an NE of the user scheduling game. Analytical and simulation results show that the COUS can offer near-optimal performance in terms of the overall cost. Besides, the dynamic payment scheme can achieve a win-win outcome for both FANs and FCN, but an unfair workload distribution among FANs, compared with the fixed payment scheme.