The field of federated learning in wireless networks is moving towards addressing the challenges of straggler delays, communication overhead, and privacy preservation. Researchers are exploring innovative solutions such as pinching-antenna systems, differential privacy, and coherence-aware distributed learning to improve the efficiency and accuracy of federated learning in wireless networks. Notable papers in this area include:

• Pinching-antenna-enabled Federated Learning, which investigates the use of pinching-antenna systems to shorten the worst links and increase the on-time completion probability in asynchronous federated learning.
• Differential Privacy as a Perk, which demonstrates that differential privacy can be achieved without employing artificial noise in over-the-air federated learning.
• Fed-PELAD, which proposes a novel federated learning framework that incorporates personalized encoders and a LoRA-adapted shared decoder to reduce communication overhead and improve CSI feedback accuracy.
• Coherence-Aware Distributed Learning, which introduces a resource-reuse strategy based on product superposition to efficiently schedule both static and dynamic devices in practical wireless federated learning systems.
• Non-Convex Over-the-Air Heterogeneous Federated Learning, which develops novel OTA-FL SGD updates that allow a structured, time-invariant model bias while facilitating reduced variance updates.