Task offloading engine for heterogeneous mobile Clouds

Abstract

The limitations in computational resources and battery power of mobile devices led to the concept of offloading compute-intensive tasks to powerful devices. We have developed a framework to offload tasks from a mobile device to other nearby heterogeneous devices. It contains an offloading engine to selectively choose the target devices for the execution of the offloaded tasks to address optimal scheduling across devices with diverse capabilities. Our initial conducted runtime measurements show the feasibility of this concept. As preliminary results, we show that offloading compute intensive tasks from a device with less computational capability to a set of nearby more powerful devices can reduce the overall computational time by approximately 50%.