Skip to Main content Skip to Navigation
New interface

Secure identification for the Internet of Things

Marzieh Gheisari 1, 2 
Abstract : This thesis addresses the problem of authentication of low-power devices in the Internet of Things by introducing new functionalities: group membership verification and identification. The procedure verifies if a given IoT device is a member of a group without revealing the identity of that member. Similarly, group membership identification states which group the device belongs to without knowing the identity. We propose a protocol through the joint use of two mechanisms: quantizing templates into discrete embeddings, making reconstruction difficult, and aggregating several templates into one group representation, impeding identification. First, we consider two independent procedures, one for embedding, the other for aggregating. Then, we replace those deterministic functions with functions whose parameters are learned through optimization. Finally, rather than considering group assignments that are predetermined, group assignments are also learned together with representations of the groups. Our experiments show that learning yields an excellent trade-off between security/privacy and verification/identification performances. We also investigate the impact of the sparsity level of the features representing group members on both security and verification performances. It shows it is possible to trade compactness and sparsity for better security or better verification performance.
Complete list of metadata
Contributor : Marzieh Gheisari Connect in order to contact the contributor
Submitted on : Wednesday, November 24, 2021 - 10:13:11 AM
Last modification on : Friday, August 5, 2022 - 2:54:52 PM
Long-term archiving on: : Friday, February 25, 2022 - 6:27:30 PM


Files produced by the author(s)


  • HAL Id : tel-03445710, version 1


Marzieh Gheisari. Secure identification for the Internet of Things. Signal and Image Processing. Inria Rennes - Bretagne Atlantique, 2021. English. ⟨NNT : ⟩. ⟨tel-03445710⟩



Record views


Files downloads