Relay-Assisted Federated Edge Learning: Performance Analysis and System Optimization

In this paper, we study a relay-assisted federated edge learning (FEEL) network under latency and bandwidth constraints. network, N users collaboratively train global model assisted by xmlns:xlink="http://www.w3.org/1999/xlink">M intermediate relays one server. We firstly propose partial aggregation spectrum resource multiplexing at the in order to improve communication of FEEL system. Furthermore, derive analytical asymptotic expressions system outage probability convergence rate. For purpose improving performance, further optimize maximizing number who participate each round learning, through allocation wireless among relays. Specifically, two (BA) schemes have been proposed, assuming either instantaneous or statistical channel state information (CSI). Simulations show advantages proposed BA over other benchmarks, regarding accuracy rate considered network.