Joint Optimal Pricing and Task Scheduling in Mobile Cloud Computing Systems
Joint Optimal Pricing and Task Scheduling in Mobile Cloud Computing Systems
复制标题
DOI：
10.1109/twc.2017.2707084
复制
发表时间：
2017-05
影响因子：
10.4
通讯作者：
Hamed Shah-Mansouri;V. Wong;R. Schober
中科院分区：
文献类型：
--
作者：
Hamed Shah-Mansouri;V. Wong;R. Schober
不断发展的移动云计算（MCC）范式使移动用户能够将其计算任务卸载到云服务器上。本文研究了MCC系统中的以下问题：1)哪些任务应该卸载到云服务器上？2)云服务的最优价格是多少？我们通过制定两个层次的优化问题来共同解决这些问题。在移动用户端，我们制定了一个考虑云服务能耗、延迟和价格的效用最大化问题，并获得延迟敏感和延迟容忍应用的最优调度。在云服务提供商（CSP）方面，我们通过制定一个一般非凸的利润最大化问题来确定最优定价策略。我们进一步提出了一种利用凸化和原始对偶方法来减轻非凸性的算法。通过数值研究，研究了移动用户的行为和CSP的定价策略。我们的研究结果表明，与文献中提出的调度程序相比，所提出的调度程序有效地平衡了能耗和延迟之间的权衡。此外，我们表明，与静态和动态定价策略相比，采用所提出的定价算法，CSP可以将其利润提高25%。
The evolving mobile cloud computing (MCC) paradigm enables mobile users to offload their computing tasks to cloud servers. In this paper, we study the following problems in MCC systems: 1) which tasks should be offloaded to cloud servers? 2) and what is the optimal price of cloud services? We jointly address these issues by formulating two levels of optimization problems. On the mobile users side, we formulate a utility maximization problem that takes the energy consumption, delay, and price of cloud services into account and obtain the optimal scheduling for both delay-sensitive and delay-tolerant applications. On the cloud service provider (CSP) side, we determine the optimal pricing strategy by formulating a profit maximization problem, which is non-convex in general. We further propose an algorithm using convexification and primal-dual methods to mitigate the non-convexity. Through numerical studies, we investigate the mobile users’ behavior and the CSP’s pricing strategy. Our results reveal that the proposed scheduler effectively balances the tradeoff between the energy consumption and delay in comparison with different schedulers proposed in the literature. Furthermore, we show that with the proposed pricing algorithm, the CSP can improve its profit by up to 25% compared with static and dynamic pricing strategies.