A system that is used to offload computing from a mobile device to the cloud is presented in this paper. Presumably, this system can dramatically reduce battery usage and also increase computing performance, since cloud servers have a lot more processing power than mobile devices. The idea is that if a system can offload some parts of the software to the cloud, this will increase performance and reduce battery usage. Some applications like face recognition can really benefit from this approach, because you only have one input and one output and very intense computing in that case. When the application sends data to the cloud to be computed, there are gains in both computing and battery consumption.

For many applications, even for experts, it is difficult to determine which parts of the software should be offloaded. When we think about web applications, the offloading approach cannot provide any advantages. JavaScript and animations especially make dividing applications much more difficult. As previously mentioned, trying to offload computation to the cloud may fail when it is not done carefully, and may also decrease battery life.

This idea has been tried before [1,2]. Even if the results seem promising in the beginning, this approach has several drawbacks [3]. Previous studies suggest that when offloading is done with 3G/4G connections, battery consumption increases due to data transfer requirements. The authors provide a clear explanation of the formulation of optimal safe-point scheduling under temporal disconnections and deadlines, which makes the paper different from most on code offloading projects. However, in the evaluation part, even if the improvements indicate that it is an effective algorithm, there is no clear description or indication that the safe points are really optimized. Also, the figures are not very readable, so it is difficult to understand what they are showing without reading the relevant parts of the paper that describe what they show. Overall, the results seem very promising; compared to basic offloading techniques, this approach is more effective.