科研成果详情

Intelligent networking and big data analytics are two important pillars for the operation of systems. Edge computing is frequently used in smart IoT systems, particularly in those which cannot be served efficiently through cloud computing due to the limitations in bandwidth, latency and Internet connectivity. However, applications always generate a large amount of data, which are pre-programmed and predefined to run on the cloud or edge platform and can't be changed at run time. The applications may gain better performance if they synergistically run on the cloud and edge platform. In this study, a novel algorithm called Dynamic Switching Algorithm is proposed to ensure intelligent dynamics where all tasks are either offloaded on cloud or edge according to the system's real-time conditions. We further divide applications into four types based on their real-time requirements. Each type of application is set to a reasonable latency to make sure the system to have less processing time. The results demonstrate that our method outperforms two state-of-the-art methods, decreasing both the average delay and energy consumption of offloading by 8.17%~66.90% and 3.76%~78.60% respectively. The experimental evaluations show that the performance of the proposed method could effectively offload tasks in smart IoT systems.