Context-aware decision engine for mobile cloud offloading

Abstract

We consider the decision engine of mobile cloud offloading systems, which decides whether to offload a given method to the cloud servers. We design, implement, and evaluate a context-aware decision algorithm, called CADA, to optimize the performance of the mobile devices with various optimization criteria, including short response time and low energy consumption. The CADA algorithm can work with various mobile cloud offloading systems proposed in the literature. We have implemented the CADA algorithm and integrated it with ThinkAir, and conducted experiments with real Android users. The evaluation results reveal that: (i) the importance of a context-aware decision engine, (ii) the high prediction accuracy of the CADA algorithm, (iii) the performance improvement (in both response time and energy consumption) achieved by the CADA algorithm, and (iv) the negligible overhead incurred by the CADA algorithm and other software components in our system. In addition to the CADA algorithm, we also develop a contextaware energy model for mobile devices, which is of interests in its own right.