Enhancement of User Quality of Experience (QoE) for Service Migration in Context Aware Environment

Context awareness equips systems with intelligence and enables them to adapt according the environment. Accurateness of adaption depends on the quality of context. The context aware migration process enables an application to move between devices without user intervention. Contexts such as network condition, device resources utilization and user mobility may be outdated, and have potential to invoke an unnecessary migration which impacts the satisfaction level of user. Firstly, the work presented in this thesis propose a service migration framework (SMF) for designing application with migration capabilities. The elements which constitute the migration process were identified i.e. serialization, transfer and de-serialization of states. The time consumed by each element of the migration process, assisted in modelling loss of user experience as a numerical value, named as QoE loss score value. The effect of outdated context information on the migration process was analysed using QoE loss score value, and simulation results showed a linear relationship between QoE loss score value and outdated information. Secondly, the migration process concept proposed in SMF was used to transfer load across cores of a device. The aim was to enhance the user experience by improving responsiveness of the application that was affected due to unbalanced utilization of resources. Thirdly, the study investigated impacts of outdated context information on automatic migration trigger. As a use case, load migration decision for massively multiplayer online game server (MMOG) was simulated. Due to communication delay and load transfer delay, the shared context is often outdated, and may trigger unnecessary migration. A prediction based approach was presented, to use estimated future server state as additional context to balance load. The prediction algorithm showed better performance than the normal approach for various server load. Fourthly, the study proposed a structured approach for load selection mechanism, to transfer load between MMOG servers. The main aim of the proposed selection approach, was to enhance individual user
responsiveness. In summary, the thesis identified core elements of migration process. The mapping of loss of user experience as QoE loss score value, provides a performance metric for measuring performance of migratory application. Furthermore, the impact of out dated information in dynamic environment was reduced, by improving the quality of context through a prediction algorithm.