An Online Algorithm for Virtualized Network Function Placement in Mobile Edge Industrial Internet of Things

Abstract

Mobile edge Industrial Internet of Things (MEIIoT) is composed of Industrial Internet of Things (IIoT) and mobile edge computing, which is currently a new type of IIoT. MEIIoT has the characteristics of large scale and strong dynamics (e.g., network topology or number of IIoT devices would change from time to time). The placement of virtualized network functions (VNFs) in MEIIoT refers to placing multiple network functions (e.g., motion analyzer and video processor) on edge nodes in a form of software instances, so that IIoT devices can flexibly obtain services of these VNFs. However, an edge node can only be placed a small number of VNFs, because of its limited storage and computing resources. Therefore, if an IIoT device requires multiple VNFs, it needs to transmit its data to access several edge nodes, which would cause high delay. How to optimally place all the VNFs on edge nodes in MEIIoT, so that the whole access delay for all IIoT devices that requiring VNFs is minimized, is a challenging problem. In this article, we design an online placement algorithm. First, we decompose a long-term VNFs optimization problem into a series of one-shot optimization problems. Second, we formulate these one-shot problems into integer nonlinear programming problems, and prove that they are NP-hard. To overcome this hardness, we then propose a heuristic algorithm. Finally, we carried out extensive experiments with real-world datasets to validate the efficacy of our proposed solution.