LuxTrace - indoor positioning using building illumination Julian Randall* and Oliver Amft* and Juergen Bohn+ and Martin Burri* * Wearable Computing Lab., + Institute for Pervasive Computing, ETH Zurich, Switzerland ETH Zurich, Switzerland http://www.wearable.ethz.ch http://www.vs.inf.ethz.ch Abstract Tracking location is challenging due to the numerous constraints of practical systems including, but not limited to, global cost, device volume and weight, scalability and accuracy; these constraints are typically more severe for systems that should be wearable and used indoors. We investigate the use of wearable solar cells to track changing light conditions (a concept that we named LuxTrace) as a source of user displacement and activity data. We evaluate constraints of this approach and present results from an experimental validation of displacement and activity estimation. The results indicate that a distance estimation accuracy of 21 cm (80% quantile) can be achieved. A simple method to combine LuxTrace with complementary absolute location estimation methods is also presented. We apply carpet-like distributed RFID tags to demonstrate online learning of new lighting environments.