Xu Jiuyun, Hao Zhuangyuan, Sun Xiaoting
College of Computer Science and Technology, China University of Petroleum (East China), Qingdao 266580, China.
Sensors (Basel). 2019 Jul 23;19(14):3231. doi: 10.3390/s19143231.
Mobile edge computing (MEC) has become more popular both in academia and industry. Currently, with the help of edge servers and cloud servers, it is one of the substantial technologies to overcome the latency between cloud server and wireless device, computation capability and storage shortage of wireless devices. In mobile edge computing, wireless devices take responsibility with input data. At the same time, edge servers and cloud servers take charge of computation and storage. However, until now, how to balance the power consumption of edge devices and time delay has not been well addressed in mobile edge computing. In this paper, we focus on strategies of the task offloading decision and the influence analysis of offloading decisions on different environments. Firstly, we propose a system model considering both energy consumption and time delay and formulate it into an optimization problem. Then, we employ two algorithms-Enumerating and Branch-and-Bound-to get the optimal or near-optimal decision for minimizing the system cost including the time delay and energy consumption. Furthermore, we compare the performance between two algorithms and draw the conclusion that the comprehensive performance of Branch-and-Bound algorithm is better than that of the other. Finally, we analyse the influence factors of optimal offloading decisions and the minimum cost in detail by changing key parameters.
移动边缘计算（MEC）在学术界和工业界都越来越受欢迎。目前，借助边缘服务器和云服务器，它是克服云服务器与无线设备之间延迟、无线设备计算能力和存储不足的重要技术之一。在移动边缘计算中，无线设备负责输入数据。同时，边缘服务器和云服务器负责计算和存储。然而，到目前为止，如何平衡边缘设备的功耗和时延在移动边缘计算中尚未得到很好的解决。在本文中，我们关注任务卸载决策策略以及卸载决策对不同环境的影响分析。首先，我们提出一个同时考虑能耗和时延的系统模型，并将其转化为一个优化问题。然后，我们采用两种算法——枚举法和分支定界法——来获得使包括时延和能耗在内的系统成本最小化的最优或近似最优决策。此外，我们比较了两种算法的性能，并得出分支定界算法的综合性能优于另一种算法的结论。最后，我们通过改变关键参数详细分析了最优卸载决策和最小成本的影响因素。
