Tag
This paper presents an adaptive joint compression and synchronization mechanism for federated split learning to reduce communication overhead in IoT rainfall prediction, achieving significant traffic reduction without major loss in predictive quality.