Design of asynchronous federated learning incentive mechanism

Federated learning has been widely paid attention as a new type of distributed machine that can protect data privacy and ensure security. Asynchronous federation learning, variant traditional effectively improve model training efficiency. The introduction incentive mechanism help asynchronous to utility effectively. A federated is constructed using the Stackelberg game, which optimizes central server owner utilities, respectively, game. Based on this, we derive equilibrium solution whole finally analyze feasibility by arithmetic examples obtain optimal effect.