Revenue-Oriented Optimal Service Offloading Based on Fog-Cloud Collaboration in SD-WAN Enabled Manufacturing Networks

The software-defined wide area network (SD-WAN) is considered one of the most promising paradigms for next generation manufacturing networks. However, SD-WAN users usually suffer from significant delays due to remotely deployed cloud centers. The requirements of delay-sensitive business services make optimal resource allocation methods very important. In this paper, we propose a revenue-oriented service offloading method to improve the efficiency of SD-WAN enabled manufacturing networks through fog-cloud collaboration. To maximize the service revenue, we formulate a coupled combinatorial optimization model to allocate computation and communication resources jointly between the fog node and the cloud. To solve this problem, we propose a service offloading method based on the counterfactual regret minimization (CFR) principle according to the dynamic workload state of the fog nodes. This method reduces the time complexity of problem-solving from exponential to polynomial, and achieves good performance that is very close to the optimal solution in terms of service efficiency. The outstanding contribution of this paper is to unify the multi-objective problem to the revenue scale for optimization to improve the overall service revenue of the SD-WAN. The simulation results show that our method outperforms benchmark methods in terms of both effectiveness and efficiency.