Mobile-to-mobile opportunistic task splitting and offloading

With the advent of wearable computing and the resulting growth in mobile application market, we investigate mobile opportunistic cloud computing where mobile devices leverage nearby computational resources in order to save execution time and consumed energy. Our goal is to enable generic computation offloading to heterogeneous devices forming a mobile-to-mobile opportunistic computing platform. In this paper, we adopt (1) an analytical approach and (2) an experimental approach to highlight the gain given by mobile-to-mobile opportunistic offloading compared to local execution. We also investigate multiple offloading strategies with regards to both computation time and energy consumption. We propose an auto-splitting and offloading algorithms that computes the optimal chunks sizes that could be offloaded remotely to neighboring mobile device. We show that our splitting and offloading algorithm succeeds in picking the optimal chunk sizes and distribution with up to 99.7% efficiency. In addition, the offloader device saves up to 80% energy while offloading the task remotely. For instance if the offloader device is running out of battery, offloading is the ultimate solution to increase its lifetime.