Recent developments, such as a fast-moving Internet of Things (IoT) ecosystem and increasing apprehensions about data privacy, bandwidth restrictions, and latencies in centralised machine learning systems, have spurred interest in Federated Learning (FL) across edge networks. Federated Learning provides a way of edge-based AI model training without transporting the original data to central servers. The analytics framework that provides this paradigm shift benefits in terms of privacy preservation, less ingestion latency for communication, and real-time decisions for services, thus turning it to a highly prized model for healthcare applications, autonomous vehicles, energetic infrastructures on smart cities, and industrial Internet of Things. This chapter focuses on the architecture, working of Federated learning, and its advantages over the conventional deep learning network architectures in the edge networks. The first chapter comprises the main principles of FL and its differences from the classic machine learning. Then, it delves into the technical aspects of deploying FL on heterogeneous edge devices, such as model aggregation strategies, communication protocols, and resource constraints. Introduces various privacy-preserving techniques that could be adopted for FL. In addition, the chapter is complemented with real world use case scenarios that demonstrate how FL can be successfully deployed in edge networks. It also addresses fundamental issues such as system scalability, data and device heterogeneity, trust management, and fault tolerance. We analyze the potential solutions and recent advances in adaptive FL algorithms and hierarchical edge-cloud collaboration. Overall, this chapter highlights the revolutionary capabilities of Federated Learning in edge environments and suggests categories for further advancements toward resilient, secure and effective edge-intelligent systems. With its unique dual-focus and ability to transcend the lines between distributed AI and edge computing, FL is set to reshape the future of intelligent data processing and insights generation at the edge.