DEV Community

Cover image for Computational Offloading in Mobile Cloud Computing
Thiago
Thiago

Posted on

Computational Offloading in Mobile Cloud Computing

Back in 2014, in this paper, we've got to see the definition of Mobile Cloud Computing (MCC). It says that MCC is the combination of Cloud Computing and Mobile Computing 🀯. The concept of Cloud Computing sounds like the most obvious solution to the hardware limitation of Mobile devices. I know what you're looking at your newest iPhone or Xiaomi and thinking that limitation might not be the right word, in fact, Smartphones nowadays are supercomputers in our pockets, but think for a moment:

What if the battery of your phone dies RIGHT NOW? It is a "hardware limitation".
What if you are on a bus, just passing through a tunnel, how flaky is your mobile connection going to be?

I could keep talking about "limitations" of our Smartphones but if you think, those two scenarios that I mentioned, would not be a problem for an old fashion Desktop for example, as it is always plugged in an electricity cable, always in a cable internet or even your house wifi, pretty good and stable.
But mobile devices are not only Smartphones, right? It can be anything, such as your Smartwatch or your Nintendo Switch - That's why I always carry a Power Bank 16000mah with me πŸ§žβ€β™‚οΈ -. So as you can see, MCC Computational Offloading is still a very relevant topic in 2020.

Context-Awareness

Context-Awareness is one of the most important topics in MCC. It has a significant impact on the computational offloading due to the uncertainty whether this would be beneficial or not. Context Awareness in MCC can be classified into two approaches: Context-Aware application partitioning and Context-Aware computational offloading.

Context-Aware application partitioning

If you already know what are the resource intensive features of your application and you will run it in the cloud, as simple as that. mCloud proposes an architecture for dynamic offloading, this framework composed of several components for decision making and context-monitor, was developed in Java using reflection for selecting the essential piece of code to be offloaded to the cloud, it basically wraps the source code and sends to the cloud, where it will be compiled and executed. The advantage of mCloud is that you will be able to offload smaller "parts" of your application but as you can imagine, just by doing that, sounds like heavy computing itself. Also, this approach can not be the best decision in cases where the mobile application uses hardware features like sensors or GPS.

Context-Aware computation offloading

The Context-Aware methodology prioritises energy efficiency, application execution and performance. Here, you won't have specific functionality that will always be offloaded to the Cloud, the decision to offload or not, will be based in some aspects:

Objective Awareness

The computational offloading has three principles that were already mentioned in this research: performance enhancement, energy efficiency and execution support. Usually, the application will need to focus on one of those principles according to the respective scenario. The mCloud architecture, for example, prioritises performance and energy efficiency.

Performance Awareness

In the majority of the cases, the computational offloading focus on performance enhancement, but sometimes, the computational offloading takes more time to offload the task than running it locally, so in these cases, the offloading is not executed, thus not to compromise the performance of the application.

Energy Awareness

Every time that a mobile device runs an intensive computational task locally, a significant amount of energy is consumed, it is a problem due to the energy limitation of the devices.
Many authors use energy awareness as the main factor to decide when to offload or not the computational task to the cloud, however, to offload the task itself, the mobile also consumes a considerable amount of energy during the request and result of this process. The mCloud framework takes into account the consumption of energy during the decision-making phase; an algorithm creates an estimative of the amount of energy that the particular task will spend. This information affects the final decision to offload or not to the cloud.

Resource Awareness

Resource Awareness requires the application, to be aware of the mobile device, it means that if the device does not have sufficient resources to run a computational process, plus the awareness of the cloud environment, so when the application has an intensive computational task to run. The cloud awareness verifies the conditions to offload; these two factors will be checked during the phase of the offloading decision-making.
The aim of this research is performance enhancement as the major factor to decide
if the offloading will be beneficial or not for each scenario. Also, context awareness is another important element during the decision-making phase. So it is possible to say that the experiment done for this research is a combination of performance awareness and resource awareness.

Conclusion

This is a summary of a topic that I believe that it is still relevant, mainly if we look at the mobile games that became such a big market and so many other examples.

Thanks for reading! 😁

References:
https://ieeexplore.ieee.org/document/6553297
https://ieeexplore.ieee.org/document/7158968
https://www.researchgate.net/publication/287971934_mCloud_A_Context-Aware_Offloading_Framework_for_Heterogeneous_Mobile_Cloud

Top comments (0)