Graph4Edge: a graph-based computation offloading strategy for Mobile-Edge workflow applications

Mobile Edge Computing (MEC) expands the capability of mobile devices so that users can run complicated and computation-intensive applications such as workflow and machine learning tasks. Computation offloading is the key technology for MEC and has attracted a lot of research efforts in recent years. However, most of the existing studies employed optimisation algorithms such as GA and PSO which have significant computation overhead. Meanwhile, the computation tasks are often assumed to be independent of each other, which is not applicable to workflow applications with strong task dependencies. To address these issues, we propose Graph4Edge which is a graph-based computation offloading strategy for mobile-edge workflow applications. In this paper, firstly, we formulate the computation offloading problem in MEC using a DAG (Directed Acyclic Graph) based model; secondly, we propose the shortest-path-based algorithm to find the optimal computation offloading plan; finally, preliminary experiments with real-world workflow traces are conducted to evaluate the performance of our proposed strategy. Given the promosing results demonstrated in this paper, we have also presented some important research directions for our future work.