An Incentive Mechanism for Federated Learning in Wireless Cellular network: An Auction Approach

Federated Learning (FL) is a distributed learning framework that can deal\nwith the distributed issue in machine learning and still guarantee high\nlearning performance. However, it is impractical that all users will sacrifice\ntheir resources to join the FL algorithm. This motivates us to study the\nincentive mechanism design for FL. In this paper, we consider a FL system that\ninvolves one base station (BS) and multiple mobile users. The mobile users use\ntheir own data to train the local machine learning model, and then send the\ntrained models to the BS, which generates the initial model, collects local\nmodels and constructs the global model. Then, we formulate the incentive\nmechanism between the BS and mobile users as an auction game where the BS is an\nauctioneer and the mobile users are the sellers. In the proposed game, each\nmobile user submits its bids according to the minimal energy cost that the\nmobile users experiences in participating in FL. To decide winners in the\nauction and maximize social welfare, we propose the primal-dual greedy auction\nmechanism. The proposed mechanism can guarantee three economic properties,\nnamely, truthfulness, individual rationality and efficiency. Finally, numerical\nresults are shown to demonstrate the performance effectiveness of our proposed\nmechanism.\n