Wireless Federated Learning with Local Differential Privacy
This addresses privacy and efficiency challenges in distributed machine learning for wireless networks, offering a novel integration with incremental improvements in privacy scaling.
The paper tackles federated learning over a wireless channel with local differential privacy constraints, showing that wireless superposition enables bandwidth-efficient gradient aggregation with strong privacy guarantees, where privacy leakage per user scales as O(1/√K) compared to constant leakage in orthogonal transmission.
In this paper, we study the problem of federated learning (FL) over a wireless channel, modeled by a Gaussian multiple access channel (MAC), subject to local differential privacy (LDP) constraints. We show that the superposition nature of the wireless channel provides a dual benefit of bandwidth efficient gradient aggregation, in conjunction with strong LDP guarantees for the users. We propose a private wireless gradient aggregation scheme, which shows that when aggregating gradients from $K$ users, the privacy leakage per user scales as $\mathcal{O}\big(\frac{1}{\sqrt{K}} \big)$ compared to orthogonal transmission in which the privacy leakage scales as a constant. We also present analysis for the convergence rate of the proposed private FL aggregation algorithm and study the tradeoffs between wireless resources, convergence, and privacy.