Full Program

Evaluating Smart Home Privacy: The Relationship Between Encrypted Sensor Data and Occupancy Prediction Through Machine Learning

Summary:

Smart meters and other IoT devices in smart homes have revolutionized residential energy management. However, this comes at significant privacy risks, as attackers can exploit leaked data to predict household occupancy. Therefore, we introduce a novel cryptographic scheme subdivided into two. Both methods involve a robust two-fold encryption process, i.e.; performing Fernet encryption and random shuffling vs Fernet encryption and strategy based operations on household data, respectively. To test the effectiveness of these two methods, we predict the occupancy using an AutoML classifier on encrypted and unencrypted data. Whereas for unencrypted data, attackers could have achieved >95% across all performance metrics, we found that the strategic method attains a remarkable F1-Score of 0.451 for predicting absence at 100% encryption level. The corresponding value for the random shuffling method is 0.004. These findings signify the importance of encryption as a privacy-preserving measure in smart homes and thereby prevent household thefts.

Author(s):

Sneha Mohanty    
University of Freiburg
Germany

Sneha Mohanty is working as a scientific co-worker in the chair of Computer Networks and Telematics, headed by Prof. Dr. Christian Schindelhauer, at the University of Freiburg, Germany. She is currently pursuing her PhD and her research focuses mainly on Cryptography and Indoor Localization.

https://cone.informatik.uni-freiburg.de/members/sneha-mohanty/

Demetrios Papadopoulos    
University of Freiburg
Germany

Demetrios N. Papadopoulos is working as a scientific co-worker in the chair of Control and Integration of Grids, headed by Prof. Dr. Anke Weidlich, at the University of Freiburg, Germany. He is currently pursuing his PhD at the chair and his research focuses on using Machine Learning in Energy Systems modelling and forecasting.

https://www.inatech.uni-freiburg.de/de/professuren/tev/mitarbeiter/dimitrios-papadopoulos

Christian Schindelhauer    
Georges-Köhler-Allee 051 79110 Freiburg
Germany

Prof. Dr. Christian Schindelhauer is currently heading the chair of Computer Networks and Telematics at the University of Freiburg in Germany. His research interests lie in Cryptography, Indoor Localization and Algorithm design.

https://cone.informatik.uni-freiburg.de/members/christian-schindelhauer