Prediction of Context Time Series

Context awareness is a key feature of modern computing applications, allowing application behaviour to be adapted to context. The power and quality of context-aware applications can significantly be increased by not only considering past and present contexts for adaptation, but by also predicting and reacting to future contexts. This paper deals with fundamental challenges of context prediction. After an introduction to context prediction and a clarification of the basic terms associated with it, the paper proposes solutions concerning prediction algorithms and describes an architecture for context-aware computing, including context prediction, we have developed. The findings described in the paper are illustrated and verified by simulations using different types of context data and different prediction algorithms.