English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In this paper, we consider privacy aspects of wireless federated learning(FL) with Over-the-Air (OtA) transmission of gradient updates from multipleusers/agents to an edge server. By exploiting the waveform superpositionproperty of multiple access channels, OtA FL enables the users to transmittheir updates simultaneously with linear processing techniques, which improvesresource efficiency. However, this setting is vulnerable to privacy leakagesince an adversary node can hear directly the uncoded message. Traditionalperturbation-based methods provide privacy protection while sacrificing thetraining accuracy due to the reduced signal-to-noise ratio. In this work, weaim at minimizing privacy leakage to the adversary and the degradation of modelaccuracy at the edge server at the same time. More explicitly, spatiallycorrelated perturbations are added to the gradient vectors at the users beforetransmission. Using the zero-sum property of the correlated perturbations, theside effect of the added perturbation on the aggregated gradients at the edgeserver can be minimized. In the meanwhile, the added perturbation will not becanceled out at the adversary, which prevents privacy leakage. Theoreticalanalysis of the perturbation covariance matrix, differential privacy, and modelconvergence is provided, based on which an optimization problem is formulatedto jointly design the covariance matrix and the power scaling factor to balancebetween privacy protection and convergence performance. Simulation resultsvalidate the correlated perturbation approach can provide strong defenseability while guaranteeing high learning accuracy.

Over-the-Air Federated Learning with Privacy Protection via Correlated Additive Perturbations