Mobile cloud computing helps to overcome the challenges of mobile computing by allowing mobile devices to migrate computation-intensive and data-intensive tasks to high-performance and scalable computation resources. However, emerging data-intensive applications pose challenges for mobile cloud computing platforms because of high latency, cost and data location issues. To address the challenges of data-intensive applications on mobile cloud platforms, we propose an application offloading optimisation model that schedules application tasks on an integrated computation environment named Mobile Edge Cloud Computing. The optimisation model is formulated as a mixed integer linear programming model, which considers both monetary cost and device energy as optimisation objectives. Moreover, the allocation process considers parameters related to data size and location, data communication costs, context information and network status. To evaluate the performance of the proposed offloading algorithm, we conducted real experiments on the implemented system with a variety of scenarios, such as different deadline and multi-user parameters. Our results demonstrate the ability of the proposed algorithm to generate an optimised resource allocation plan in response to dramatic fluctuations in application data size and network bandwidth. The proposed technique reduced the execution cost of data-intensive applications by an average of 46% and 76% in comparison with particle swarm optimisation (PSO) and full execution on a mobile device only, respectively. In addition, our new technique reduced mobile energy consumption by 35% and 84%, compared to PSO and full execution on a mobile device only, respectively.

通过允许移动设备将计算密集型和数据密集型任务迁移到高性能和可扩展的计算资源，移动云计算有助于克服移动计算的挑战。然而，由于高延迟，成本和数据位置问题，新兴的数据密集型应用程序给移动云计算平台带来了挑战。为了解决移动云平台上数据密集型应用程序的挑战，我们提出了一种应用程序卸载优化模型，该模型可在名为“移动边缘云计算”的集成计算环境中调度应用程序任务。优化模型被公式化为混合整数线性规划模型，该模型同时考虑了货币成本和设备能量作为优化目标。此外，分配过程考虑与数据大小和位置，数据通信成本，上下文信息和网络状态有关的参数。为了评估所提出的卸载算法的性能，我们在已实现的系统上针对各种场景（例如不同的截止日期和多用户参数）进行了实际实验。我们的结果证明了提出的算法能够响应应用程序数据大小和网络带宽的剧烈波动而生成优化的资源分配计划。与粒子群优化（PSO）和仅在移动设备上完全执行相比，所提出的技术分别将数据密集型应用程序的执行成本平均降低了46％和76％。此外，我们的新技术将移动能耗分别降低了35％和84％，