Federated Learning (FL) has emerged as a distributed machine learning paradigm that enables collaborative model training while preserving data privacy. Unlike traditional centralized learning frameworks which require collecting raw data at a single server, FL allows multiple clients to train models locally and share only model updates for global aggregation. In this review we examine twelve peer-reviewed surveys and research papers published between 2017 and 2025 that analyze federated learning architectures, communication mechanisms, privacy-preserving techniques, security threats, types of FL and real-world deployment scenarios. Drawing substantially from the comprehensive IEEE Access survey by Aledhari et al., our analysis shows that FL still faces major technical challenges including non-IID data distributions, high communication costs, scalability constraints and adversarial threats. We also highlight emerging research directions such as lightweight optimization, fairness-aware aggregation, blockchain-based trust mechanisms and personalized FL. This review consolidates existing work, presents a full 12-paper literature summary table, and outlines key open problems to guide future research on federated learning systems.
Rathod et al. (2026) studied this question.