The federated learning (FL) in edge computing presents a robust solution to privacy concerns in modern data-driven environments. Traditional centralized data processing methods struggle with large-scale data transmission, stringent privacy regulations, and the need for real-time decision-making. FL allows for collaborative model training without the need for centralized data sharing. Existing surveys on FL in edge environment are often narrow in scope, focusing solely on individual techniques like blockchain or differential privacy. There is a lack of comprehensive summaries and analyses that integrate these diverse approaches. This paper analyzes various methodologies and improvements proposed by other researchers, including the adoption of advanced techniques such as differential privacy and blockchain, as well as the application of these combined technologies in vehicular networks, healthcare, and industrial domains. Additionally, we discuss the current limitations and challenges of existing solutions, offering insights into potential areas for future research.