A Reinforcement Learning based Edge Cloud Collaboration
Hiroki Kobari, Zhaoyang Du, Celimuge Wu, Tsutomu Yoshinaga, Wugedele Bao · 2021
Recently, edge computing has attracted more and more attention. Compared with traditional cloud computing, edge computing can reduce communication delay. However, the processing capability of edge computing is not as good as cloud computing. The proposed method combines the advantage of the low communication delay of edge computing and the high processing capability of cloud computing. We use the Q-learning algorithm to balance network load between the edge server and the cloud server to reduce the average service time. Simulation results show that the proposed method suppresses the task failure rate while reducing the average service time.