Revenue-Oriented Service Offloading through Fog-Cloud Collaboration in SD-WAN

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