Energy Efficient Scheduling Manager for cloud-based mobile applications
Mobile cloud computing is the state-of-the-art mobile distributed computing paradigm which comprises of three heterogeneous domains: mobile computing, cloud computing, and wireless networks aiming to enhance the computational capabilities of resource-constrained mobile devices towards a rich user experience. For example, heavy computations can be offloaded to the cloud to reduce energy consumption for the mobile device. However, we discovered that, in some mobile cloud application cases, it is more energy inefficient to use cloud computing than the traditional computing conducted in the local device.
In our study, we chose a navigation application, Osmand, running on an Android mobile platform to do the empirical measurement because Osmand has both cloud-based and non-cloud-based versions. So, we were able to compare non-cloud-based Osmand and cloud-based Osmand in term of energy efficiency. In the empirical measurements, we found that non-cloud-based Osmand and cloud-based Osmand consume a similar amount of energy regarding to LCD and GPS activities. For non-cloud based Osmand, the majority of CPU energy consumption was used on calculating route results. In addition, we found that the complexity of maps affects the CPU energy consumption. On the other hand, the cloud-based Osmand consumes more CPU energy than non-cloud-based Osmand because the majority of CPU energy consumption was used for creating events and displaying on the mobile screen. Moreover, cloud-based Osmand needs a network connection to send a request to the cloud to calculate a route result. Thus, 3G communications is considered as an extra factor that causes energy consumption in cloud-based Osmand. We found that 3G communications makes cloud-based Osmand energy inefficient due to tail energy in 3G communication costing extra energy consumption.
Therefore, a prototype of an Energy Efficient Scheduling Manager (EESManager) was proposed and developed based on the awareness of tail energy and the complexity of maps. The results from the evaluation show that EESManager can improve the energy efficiency of cloud-based Osmand in two cases: 1) if the mobile device receives route results back from the cloud within the tail time, and 2) in the case when the mobile device does not receive the route results from the cloud within the tail time and the map scenario is simple.