SM@RT offloader: supporting adaptive computation offloading for Android applications

Computation offloading is a promising technique to help improve the performance as well as reduce the battery consumption of a smartphone app (short for application) by executing some parts of the app on a remote server. However, to make a favorable offloading decision, one has to consider lots of factors including the program structure of the app, the hardware configuration of the smartphone, the current resources usages (e.g., battery, CPU, and memory), and the volatile network conditions. It is almost impossible to implement a program logically built-in in the app to support adaptive offloading along with the ever-changing app contexts. In this paper, we present a middleware, SM@RT Offloader, to support adaptive computation offloading for Android mobile apps. Our experiment shows the feasibility and effectiveness of SM@RT Offloader.