Securing the vast network of interconnected devices has become a critical challenge with the proliferation of the Internet of Things (IoT) and edge computing. Intrusion detection plays an essential role in identifying and mitigating potential security threats. Federated learning (FL) emerges as a promising paradigm, enabling collaborative model training without centralizing sensitive data. This systematic review provides a comprehensive overview of the state-of-the-art federated learning for intrusion detection on edge devices. The increasing integration of IoT devices and edge computing into our daily lives requires innovative approaches to protect sensitive information. Traditional intrusion detection systems struggle with the scale and heterogeneity of edge devices, making it imperative to rethink and adapt existing security paradigms. This review synthesizes the insights of several key studies that explore the application of federated learning for intrusion detection in diverse domains, such as healthcare, vehicular networks, and smart grids. The methodologies, challenges, and performance metrics associated with federated learning in these contexts are examined. This review aims to guide future research and innovations in developing resilient and privacy-preserving security mechanisms for edge computing environments by addressing critical challenges like communication overhead, data heterogeneity, and privacy concerns.