Development of a novel context prediction algorithm and analysis of context prediction schemes

Más información sobre el libro

Context-awareness encompasses various sub-disciplines, each presenting intriguing research opportunities. While most studies focus on present or past context, this thesis explores the inference of future context. It delves into the context prediction task, identifying its opportunities and challenges, and establishes a definition for this task. Different context prediction schemes are categorized based on the extent of pre-processing or context abstraction applied to the data. The thesis evaluates these schemes for prediction accuracy and offers guidelines for application designers regarding their optimal use in specific situations. Analytical findings are validated through simulations across multiple application domains. Additionally, a novel architecture for context prediction is introduced, suitable for various levels of context abstraction, facilitating the distribution of components across interconnected devices in mobile ubiquitous computing environments. The thesis reviews existing context prediction algorithms, discussing their strengths and weaknesses, and presents a new algorithm tailored to the identified requirements of UbiComp environments. Simulation results for three exemplary algorithms are provided, showcasing their effectiveness across different application domains.