Computation offloading for multimedia workflows with deadline constraints in cloudlet-based mobile cloud

摘要

In recent years, mobile applications have become more demanding on device performance, especially for the multimedia applications. As the battery capacity and computing performance of mobile devices gradually fails to meet the needs of mobile applications for device performance, mobile cloud computing (MCC) is adopted to handle multimedia workflows. Offloading mobile applications to remote cloud with rich computing resources for execution helps improve the execution efficiency of multimedia applications on mobile devices. However, because the remote cloud is too far away from users, MCC has a high network latency, which influences the experience of multimedia applications that require high real-time performance. As a small mobile cloud computing center on the edge of Internet, cloudlet provides the powerful computing resources for surrounding mobile devices and supports resource-intensive and interactive mobile applications. However, in the cloudlet-based cloud computing environment, it is still a challenge to offload mobile applications to optimize the energy consumption of mobile devices while meeting the deadline of each mobile application. Given the challenge, an offloading method based on non-dominated sorting differential evolution (NSDE) is designed. Firstly, the multimedia application is modeled as a constrained multi-objective optimization problem by the workflow technology. Then, we use the NSDE algorithm to optimize this multi-objective optimization problem, and minimize the energy consumption with the constraints of meeting the deadline of each multimedia workflow. Finally, through a large number of experimental comparisons and analysis, the validity and superiority of our proposed method are verified.