Abstract
In the mobile edge computing (MEC) paradigm, the underlying business processes of mobile applications can be modeled as application graphs, and mobile users are allowed to offload mobile tasks in these graphs to nearby edge servers to speed up their mobile applications. Nevertheless, various challenges, especially the quality of such a mobile task offloading in edge computing environment, are yet to be properly tackled. Most studies and related offloading strategies based on the assumption that mobile users are fully stationary when they are offloading tasks to edge servers. However, this is not realistic in real-world where edge users are keeping moving, which has a great impact on the success of task offloading and further affects the response time of mobile applications. In this paper, we take the mobility of edge users into consideration and employ a deep learning method to predict the future trajectories of mobile users. Then we develop an online method to solve the offloading problem based on the quality-of-service (QoS) and the predicted user trajectories in real-time. Experiments based on real-world user trajectories and edge server dataset show that our proposed approach can achieve lower offloading failure count and shorter response time than traditional ones.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have