Global State Context Prediction Techniques Applied to a Smart Office Building

Ubiquitous systems use context information to adapt appliance behavior to human needs. Even more convenience is reached if the appliance foresees the user’s desires and acts proactively. This paper introduces context prediction techniques based on previous behavior patterns, in order to anticipate a person’s next movement. We focus on two-level predictors with global first-level histories and two-state predictors respectively frequency analysis in the second level and compare these predictors with the Prediction by Partial Matching (PPM) method. We evaluate the predictors by some movement sequences of real persons within an office building reaching up to 69% accuracy in next location prediction without pre-training and, respectively, up to 96% with pre-training.