Accessing private user information through smart devices has become an increasing concern recently. Even though smart device operating systems require user permission for perceived sensitive information, side channel attacks using inference can result in serious security breaches. In this work, we present an online learning method that infers location from raw pressure sensor data. The online method mitigates the error arising from changing weather conditions, without the need to obtain any data from weather stations. We demonstrate how a long short-term memory (LSTM) neural network trained by sequences of locations and their corresponding pressure patterns is capable of inferring user location. Experimental results showing the accuracy of the proposed system are presented.
|Original language||English (US)|
|Title of host publication||2020 9th Mediterranean Conference on Embedded Computing (MECO)|
|Publisher||Institute of Electrical and Electronics Engineers (IEEE)|
|State||Published - Jun 2020|
Bibliographical noteKAUST Repository Item: Exported on 2021-04-13
Acknowledgements: We would like to express our deep gratitude towards UCI Anteater Express shuttle service management for their help in collecting the data needed to validate the proposed methodology.