MOCCAA - MObile Cloud Computing AdaptAble

The primary objective of Mobile Cloud Computing (MCC) is the support of resource-constrained mobile devices by means of Cloud or nearby computing resources. In general, this is achieved by outsourcing or offloading computational intensive tasks of mobile applications. This allows, for instance, to virtually extend the processing power of mobile devices and, indirectly, to save battery life. In case remote resources are not available or the communication link is broken, still a local execution is possible. The challenges and issues that arise in such a scenario are manifold. The essential ones, which are also addressed in this thesis, can be summarized with the following four questions. Which part of a mobile application should be offloaded to the Cloud to accelerate, for example, its execution or to relieve the mobile device with respect to the battery consumption? Therefore, those parts of an application have to be identified that have a high resource need, but whose offloading would only lead to a small communication overhead. The decision where to run which part of the application has to take into account the current latencies to the remote resources and the dynamics of their availability and usage. This requires a Resource Management System that is able to discover matching remote resources fast and efficiently without delaying the offloading procedure too much. Subsequently, we have to determine when or under which conditions which parts shall be offloaded. This decision-making requires a monitoring on client side and a prediction about the future resource consumption of the given application parts. The final question addresses the procedure of offloading itself and the applied communication protocol. It can be summarized with how to offload. In the last few years, several approaches for MCC have been developed and examined in the research community. Most of them deal with one or two of the aforementioned aspects, but do not consider the positive and negative effects for the other dimensions. The key point of this thesis is that a comprehensive solution for MCC must handle all core questions listed above in an integrated fashion. With our framework MOCCAA we did not only strive for just another partitioning and offloading technique that is additionally equipped with a standard Resource Management System and further modules. Instead, the single modules of MOCCAA shall cooperate and interact with each other in such a way that all dimensions can benefit and are improved a little bit further.