A Methodology for Intelligent Offloading Decision in Mobile Cloud Computing Environments

Abstract

Advances in mobile computing have allowed billions of people to use mobile devices to perform their everyday activities. On the other hand, the constant use of mobile applications ends up impacting the limited processing, memory, storage and, mainly, battery of these devices. To minimize these limitations, Mobile Cloud Computing (MCC) seeks to transfer the execution of mobile device tasks and applications to be processed on clouds or servers with higher computing powers. However, computational discharge in the cloud, known as offloading, will not always have a positive impact, and this impact depends on factors such as network quality. In this context, deciding when to perform this computational offloading is a task with high relevance and complexity. Thus, this work proposes a solution to support the use of artificial neural networks (ANN) to decide where the execution of tasks will occur most efficiently, local or remote. The decision is made from sensitive context information, including details of the mobile device and the task to be performed. The evaluation of this work showed that the proposed solution has a positive performance impact on MCC environments.