Privacy-Preserving Wireless Federated Learning Exploiting Inherent Hardware Impairments
Paper in proceeding, 2021

We consider a wireless federated learning system where multiple data holder edge devices collaborate to train a global model via sharing their parameter updates with an honest-but-curious parameter server. We demonstrate that the inherent hardware-induced distortion perturbing the model updates of the edge devices can be exploited as a privacy-preserving mechanism. In particular, we model the distortion as power-dependent additive Gaussian noise and present a power allocation strategy that provides privacy guarantees within the framework of differential privacy. We conduct numerical experiments to evaluate the performance of the proposed power allocation scheme under different levels of hardware impairments.

Author

Sina Rezaei Aghdam

Chalmers, Electrical Engineering, Communication, Antennas and Optical Networks

Ehsan Amid

Google Inc.

Marija Furdek Prekratic

Chalmers, Electrical Engineering, Communication, Antennas and Optical Networks

Alexandre Graell I Amat

Chalmers, Electrical Engineering, Communication, Antennas and Optical Networks

IEEE International Workshop on Computer Aided Modeling and Design of Communication Links and Networks, CAMAD

23784873 (eISSN)

9617764
978-1-6654-1779-2 (ISBN)

2021 IEEE 26th International Workshop on Computer Aided Modeling and Design of Communication Links and Networks (CAMAD)
Porto, Portugal,

Enhanced Security and Privacy for Wireless Federated Learning (SP4WFL)

Chalmers AI Research Centre (CHAIR) (2020-013 CHAIR CO), 2020-08-01 -- 2021-01-31.

Subject Categories

Computer Engineering

Telecommunications

Other Electrical Engineering, Electronic Engineering, Information Engineering

Areas of Advance

Information and Communication Technology

DOI

10.1109/CAMAD52502.2021.9617764

ISBN

9781665417792

More information

Latest update

4/21/2023