Research Article
Does Cycling Reveal Insights About You? Investigation of User and Environmental Characteristics During Cycling
@INPROCEEDINGS{10.1007/978-3-031-34776-4_10, author={Luca Hern\^{a}ndez Acosta and Sebastian Rahe and Delphine Reinhardt}, title={Does Cycling Reveal Insights About You? Investigation of User and Environmental Characteristics During Cycling}, proceedings={Mobile and Ubiquitous Systems: Computing, Networking and Services. 19th EAI International Conference, MobiQuitous 2022, Pittsburgh, PA, USA, November 14-17, 2022, Proceedings}, proceedings_a={MOBIQUITOUS}, year={2023}, month={6}, keywords={Behavior Analysis Activity Recognition Bike Identification User Recognition}, doi={10.1007/978-3-031-34776-4_10} }- Luca Hernández Acosta
Sebastian Rahe
Delphine Reinhardt
Year: 2023
Does Cycling Reveal Insights About You? Investigation of User and Environmental Characteristics During Cycling
MOBIQUITOUS
Springer
DOI: 10.1007/978-3-031-34776-4_10
Abstract
Smartwatches are increasingly being used as fitness and health trackers. To provide such a service, these devices have to collect and process movement data gathered by built-in accelerometers and gyroscopes. Based on these data, existing studies leveraging smartphones have shown that it is possible to distinguish users when they (1) walk, (2) perform different hand gestures, or (3) pick up their phone from the table. However, to the best of our knowledge, the case of cycling has not been addressed yet. The goal of this paper is to close this gap by investigating whether it is possible to infer information about users wearing a smartwatch coupled with their smartphone when cycling, their bike type, seat height, gear, and the terrain. In addition, we explore whether it is possible to distinguish individual users based on their movement patterns that may lead to their (re)identification. To this end, we conducted a user study with 17 participants, equipped with a smartphone and a smartwatch, who had to ride along a bike road for two km. Among others, our results show that it is possible to infer the four characteristics bike type, gear, seat height, and terrain with accuracies of 93.05%, 92.23%, 95.76%, 94.24% respectively and distinguish participants with a probability of 99.01%.
