Um Mecanismo de offloading de dados com tomada de decisão

Abstract

According to IBGE (Brazilian Institute of Geography and Statistics), 92.1% of home access to the Internet was made by mobile phones. Due to the physical limitations, the processing power of these devices and the life of the batteries have not matched the growing demand of mobile applications. In the mobile cloud computing paradigm, offloading techniques are used to augment computation and power capacities of mobile devices as well as to reduce the execution time of tasks. In this dissertation, we propose a data offloading mechanism that selects and migrates files to a local infrastructure (cloudlet), assisting computation offloading frameworks to reduce the amount of data sent over the network. The mechanism uses the application methods execution history, as well as the network condition, to create decision trees that help deciding when and which files used by these methods should be transferred. The experiments results indicate that our mechanism reduces the processing offloading time by up to 19.5%.