Dynamic Service Placement Algorithm for Partitionable Applications in Mobile Edge Computing

Kun Lu, Jianyu Song, Linlin Yang, Guorui Xu, Mingchu Li

Abstract: Mobile edge computing (MEC) has become a new computing paradigm, which has caused new challenges, including how to dynamically place services to maintain user-perceived delays and determine the number of simultaneous executions of partitionable applications to optimize the quality of experience (QoE). What's more, the battery energy level of mobile devices and the operating cost of the service provider will also increase the difficulty of improving service performance. In order to solve the contradiction between the above factors and service performance, we study the performance optimization of mobile edge service placement for partitionable applications under the constraints of long-term cost budget and battery energy level. A centralized online service placement algorithm (COSPA) based on Lyapunov optimization is proposed, and the performance boundary of COSPA is theoretically analyzed. By stabilizing the average migration cost and the battery energy of the mobile device near a constant, the COSPA algorithm can obtain an asymptotically optimal solution. The experimental results based on the real dataset imply that the COSPA algorithm can obtain higher performance gains compared with the benchmarks and the Distributed Algorithm (DA).