The field of federated learning is moving towards addressing the challenges of data heterogeneity, communication efficiency, and privacy preservation. Researchers are exploring innovative solutions to balance the personalization of models for local clients with generalization for the global model. One notable direction is the development of hierarchical knowledge structuring frameworks, which enable the formulation of sample logits into a multi-granularity codebook to represent logits from personalized per-sample insights to globalized per-class knowledge. Another significant trend is the proposal of asynchronous federated learning algorithms, which aggregate models in real-time, improving training speed while mitigating the issue of client model version inconsistency. Additionally, there is a growing interest in federated learning over wireless networks, such as 5G, WiFi, and Ethernet, with a focus on measurements and evaluation to improve communication efficiency. Noteworthy papers in this area include Hierarchical Knowledge Structuring for Effective Federated Learning in Heterogeneous Environments, which proposes a novel framework for effective federated learning, and Corrected with the Latest Version: Make Robust Asynchronous Federated Learning Possible, which introduces an asynchronous federated learning version correction algorithm based on knowledge distillation. Furthermore, FedSAUC: A Similarity-Aware Update Control for Communication-Efficient Federated Learning in Edge Computing is also worth mentioning, as it proposes an update control for federated learning by considering the similarity of users' behaviors. Overall, these advances aim to enhance the efficiency, privacy, and accuracy of federated learning models, paving the way for their widespread adoption in real-world applications.