Abstract
The increased use of mobile devices has led to the creation of complex mobile applications that require more resources than are readily available on mobile devices. As resources such as processing power and storage are found on the cloud, resources of mobile devices can be increased by using cloud-based mobile augmentation. However, some resources, specifically battery life, and bandwidth cannot be augmented. To augment mobile device resources such as battery life, offloading can be used. This research discusses offloading methods and examines the approaches used in related research. It is found that most of the energy consumed when offloading is due to network communication, as opposed to computation when executing locally. When offloading to the cloud consumes less energy than local execution, the battery life of a mobile device can be conserved. Choosing between offloading and local execution is called an offloading decision. To make offloading decisions that conserve battery life, the decision-making process is explored. A challenge identified when making offloading decisions is accurately estimating the energy consumption of tasks when offloading and when executing locally. As the energy consumption profile of each device differs according to the capabilities of the device, this aspect is explored. The research conducted in this dissertation proposes the Switch framework. The Switch framework conserves the limited battery life on mobile devices by estimating the consumption of energy of a task and choosing the least expensive option. A software-based device-specific energy consumption profile is created for this purpose. Switch is evaluated using the Switch prototype, which has been designed according to the specifications of the framework. The prototype is evaluated by comparing the estimated energy consumption against the measured energy consumption. The evaluation of the framework suggests that Switch can successfully be used to conserve battery life on mobile devices by making intelligent offloading decisions.
M.Sc. (Information Technology)