Adaptive Task Scheduling in Cloud-Edge System for Edge Intelligence Application
Zeng Zeng, Weiwei Miao, Shihao Li, Xiaoyun Liao, Mingxuan Zhang, Rui Zhang, Changzhi Teng · 2021
The popularization of intelligent applications has brought convenience to human lives, but at the same time, it has also caused the explosion of big data. The traditional intelligent applications transmit big data to the cloud for computing, which puts pressure on network bandwidth, generates extensive transmission delays, and reduces the quality of services (QoS). The edge computing extends the computing power of the cloud to the edge of the network, which reduces the transmission delay. However, for resource-intensive intelligent applications, the edge resources are highly insufficient. Therefore, to fully use the edge resources and minimize the response delay of intelligent applications, we propose an adaptive task scheduling in a cloud- edge system. First, we propose coarse-grained and fine-grained services deployment algorithms to solve where and how to deploy, respectively. Then, we model the intelligent tasks as the directed acyclic graph (DAG) and propose an adaptive task scheduling algorithm based on the deployment of the services. Finally, we conduct a large number of simulation experiments and analyze the performance of the algorithms from three aspects: response time, task execution success rate, and QoS. The results show that our algorithms have better performance than the comparison algorithms.