From mobiles to clouds: Developing energy-aware offloading strategies for workflows
2012 ACM/IEEE 13th International Conference on Grid Computing, 2012•ieeexplore.ieee.org
Cloud computing and mobile computing are two of the most influential technologies that look
set to change the face of computing in the coming years. Combination of the two provides us
with an unprecedented opportunity to provide highly portable and yet content-rich and
computation-intensive services to the end user. In this paper we investigate the possibility of
using code/task offload techniques between mobile and cloud in order to reduce the energy
cost of workflows deployed on mobile devices. We first present a vision in which mobile …
set to change the face of computing in the coming years. Combination of the two provides us
with an unprecedented opportunity to provide highly portable and yet content-rich and
computation-intensive services to the end user. In this paper we investigate the possibility of
using code/task offload techniques between mobile and cloud in order to reduce the energy
cost of workflows deployed on mobile devices. We first present a vision in which mobile …
Cloud computing and mobile computing are two of the most influential technologies that look set to change the face of computing in the coming years. Combination of the two provides us with an unprecedented opportunity to provide highly portable and yet content-rich and computation-intensive services to the end user. In this paper we investigate the possibility of using code/task offload techniques between mobile and cloud in order to reduce the energy cost of workflows deployed on mobile devices. We first present a vision in which mobile devices are coordinated over a network, which is equipped with a layer of cloud-like infrastructures which we term cloudlets, whose computational resources can be leveraged by the mobile devices to host the execution of mission-critical mobile workflows in an energy-aware manner. We then build a model that encompasses various characteristics of the workflow's software and the network's hardware devices. With this model, we construct the objective functions that guide the offload decisions. We then present a heuristic algorithm that produces statistical and dynamic offload plans according to these objective functions and their variations both statically and dynamically. We conclude the paper with a series of simulation studies, the results of which give insight into the offload-ability of workflows of different characteristics. The results also illustrate how different hardware specifications can affect offload efficiency. These studies indicate that our offload algorithm can significantly improve the energy efficiency and execution speed of mobile workflows.
ieeexplore.ieee.org