Private Wireless Federated Learning with Anonymous Over-the-Air Computation

In conventional federated learning (FL), differential privacy (DP) guarantees\ncan be obtained by injecting additional noise to local model updates before\ntransmitting to the parameter server (PS). In the wireless FL scenario, we show\nthat the privacy of the system can be boosted by exploiting over-the-air\ncomputation (OAC) and anonymizing the transmitting devices. In OAC, devices\ntransmit their model updates simultaneously and in an uncoded fashion,\nresulting in a much more efficient use of the available spectrum. We further\nexploit OAC to provide anonymity for the transmitting devices. The proposed\napproach improves the performance of private wireless FL by reducing the amount\nof noise that must be injected.\n