Abstract

Edge computing is commonly adapted to reduce network delay and fluctuation in the traditional cloud-terminal architecture. Many research teams have been dedicated to algorithms and optimizations in the cloud-edge data cache and computation offloading schemes, while there is a blank in practical implementation to facilitate such a paradigm. In this work, we proposed a component-based framework that facilitates dynamic partitioning of a software program. We have designed and implemented the very first test-bed for further optimizing the components distribution strategy among cloud, edge, and terminal devices. Experiments have revealed the characteristics of components execution in the proposed architecture, showing that the system can improve computing performance under the real-world unstable network environments.