Reverse Multidimensional Auction Based Vehicle Selection and Resource Allocation for IoV

Federated Learning (FL) is gaining popularity in the Internet of Vehicles (IoV), which has led to a rise in demand for high-quality communication and computation resources. To address the communication and computation issues, we formulate the vehicle selection and channel allocation problem in the IoV as a social welfare optimization problem. We propose an efficient reverse multi-dimensional auction-based vehicle selection and channel allocation scheme, called RAFS, to enhance the overall performance of FL in the IoV. Simulation results demonstrate the efficacy of RAFS in improving social welfare and constructing higher quality virtual global datasets.