Mobility Prediction-Based Service Scheduling Optimization Algorithm in Cloudlets

Abstract

Cloudlet is an emerging technology in mobile cloud computing. However users may be far away from cloudlets due to the mobility of mobile users, which leads to a poor network connectivity, thus, user experience will be poor. While a user moves across multiple cloudlets areas, issues of service scheduling between cloudlets to better support user experience become important. In this paper, we consider a latency-sensitive and stateful service scheduling problem in cloudlets. We propose a novel cloudlet service model and formulate the problem with the goal of finding the optimal service running sequence which minimizes the average service response time during the whole running process of the service for a user. To solve this problem, we propose an algorithm called Mobility Prediction-based Markov Decision Process (MPMDP). The proposed algorithm takes user’s mobility prediction into account, and makes an decision based on Markov Decision Process to decide on which cloudlet the service should run for a user each time. Finally, we evaluate the effectiveness of the proposed MPMDP algorithm by simulations with real-world users’ traces. The simulation result shows our algorithm achieves a lower average response time compared with previous schemes.