Federated Meta-Learning Enhanced Acoustic Radio Cooperative Framework for Ocean of Things

Sixth-generation wireless communication (6G) will be an integrated architecture of "space, air, ground and sea". One the most difficult part this is underwater information acquisition which need to transmitt cross interface between water air.In senario, ocean things (OoT) play important role, because it can serve as a hub connecting Internet (IoT) (IoUT). OoT device not only collect data through methods, but also utilize radio frequence over air. For communications, acoustic communications (UWA COMMs) effective way for devices exchange information, always tormented by doppler shift synchronization errors. In paper, in order overcome UWA tough conditions, deep neural networks based receiver chirp communication, called C-DNN, proposed. Moreover, improve performance DL-model solve problem model generalization, we proposed novel federated meta learning (FML) enhanced cooperative (ARC) framework, dubbed ARC/FML, do transfer. Particularly, tractable expressions are derived convergence rate FML setting, accounting effects from both scheduling ratio, local epoch amount on single node.From our analysis simulation results, shown that, C-DNN provide better BER lower complexity than classical matched filter (MF) scenario. The ARC/FML framework has good under variety channels (FL). summary, promising scheme across