Predicting The Users' Next Location From Wlan Mobility Data

Accurate prediction of user mobility allows the efficient use of resources in our ubiquitously connected environment. In this work we study the predictability of the users' next location, considering a campus scenario with highly mobile users. We utilize Markov predictors, and estimate the theoretical predictability limits. Based on the mobility traces of nearly 7400 wireless network users, we estimate that the maximum predictability of the users is on average 82%, and we find that the best Markov predictor is accurate 67% of the time. In addition, we show that moderate performance gains can be achieved by leveraging multi-location prediction.