Have you been upstairs? On the accuracy of registrations of ascended and descended floors in iPhones

Abstract

The strong integration of smartphones in everyday life offers many new investigative opportunities. In particular, digital traces from smartphones can now increasingly be used to infer information about actions performed by their users in the physical world. For instance, the iPhone Health App is known to contain a large number of timestamped digital traces related to activities in the physical world, such as number of steps taken, distances travelled and floors ascended. In this study, we experimentally investigate the accuracy of registration of number of floors in iPhones. For this, seven test subjects ascended and descended floors of different heights with five different iPhones, where number of floors, walking speed and carrying location were varied.

Analysis of data shows that the iPhone Health App predominantly records information on floors when walking upstairs and virtually never when walking downstairs. iPhones contain other timestamped traces from which information on both ascended and descended floors in specific periods can be derived. The number of registered floors is primarily determined by the height difference travelled. From our experiments and information in the Health App, it follows that a height difference of approximately 3 m corresponds to the registration of one floor. For the height differences studied, the number of floors registered by the iPhones match the number of 3 m blocks in the height difference in 70–80 % of the cases. Other factors, such as walking speed and carrying location of the phones only have a minor effect on the accuracy of registered information. Additional experimentation shows that no floors are registered by the iPhones when ascending or descending using an elevator. When ascending or descending using an escalator, floors are only registered by the iPhones when the subjects walked on the escalator, not when standing still on the escalator.