Over-the-Air Federated Learning with Massive MIMO Using Antenna Selection

Over-the-air federated learning (OTA-FL) has been recently proposed as an enabling technology for learning a shared model collaboratively in a wireless network in a privacy-preserving fashion. With the implementation of massive MIMO, the large antenna array provides a promising beamforming gain at the server and hence leads to a considerable suppression of error in the over-the-air aggregation step. This talk will present our recent work that studies OTA-FL in massive MIMO systems by considering a realistic scenario in which the edge server, despite its large antenna array, is restricted in the number of radio frequency (RF)-chains. For this setting, the beamforming for over-the-air model aggregation needs to be addressed jointly with antenna selection. This leads to an NP-hard problem due to the integral nature of the optimization. We tackle this problem via two different approaches, which will be briefly introduced in this talk. Our experimental results depict that the learning performance of the scenario with all the antennas being active at the parameter server (PS) can be closely tracked by selecting less than 20% of the antennas at the PS.