Green Federated Learning via Carbon-Aware Client and Time Slot Scheduling

Key Points

• The aim is to enhance federated learning by implementing carbon-aware scheduling strategies.
• Developed a carbon-aware scheduling algorithm for clients in federated learning.
• Analyzed the impact of time slot management on carbon emissions during training sessions.
• Implemented simulations to assess the efficiency of the new strategies.
• Reduced carbon emissions associated with the training process.
• Improved overall performance of federated learning systems.
• Showed favorable client engagement during scheduled time slots.