“In the realm of mobile edge computing (MEC), the integration with federated learning (FL) has been a significant advancement, addressing the privacy and efficiency concerns of processing data on end-user devices (EDs). However, the energy constraints of battery-powered EDs pose a significant challenge for FL tasks. This paper offers an extensive survey on energy-efficient strategies for FL in MEC, examining system models, energy consumption, and strategies from learning-based, resource allocation, to client selection perspectives. It provides a detailed analysis, experimental results, and potential future research directions, paving the way for more sustainable and efficient FL implementations in MEC environments.”