Empirical Investigation of Offloading Decision Making in Industrial Edge Computing Scenarios

Edge Computing (EC) is a paradigm introduced to support the end devices with the execution of computation intensive tasks while maintaining their intended Quality of Service (QoS). To achieve this, one or more Edge Servers (ES) with powerful capabilities are placed in a close proximity to the edge devices to provide the assistance needed. In order to do that, EC utilizes the concept of application offloading, which is the idea of moving the computation intensive tasks to be computed on a more powerful server. The act of offloading is not always a beneficial choice due to the aspect of availability of involved resources and data communication between devices. Therefore, to achieve a successful offloading process, the offloading decision making needs to answer the four questions of when, what, where and how to offload. In this paper, we investigate the “When to offload?” question, which is concerned with whether the offloading process results in a positive gain in performance or not. To strengthen our conclusions, we use empirical observations in a real setup running a set of emulated applications.