A Deep-Learning-Based Optimal Auction for Vehicular Edge Computing Resource Allocation

The vehicular edge computing technology extends the Internet of Vehicles(IoV) from cloud computing to edge computing, enabling IoV to support in-vehicle applications such as autonomous driving, high-definition video, and navigation planning with low latency and low bandwidth consumption costs. Due to the high deployment cost and maintenance cost of edge computing nodes, to improve the revenue of service providers and encourage edge computing service providers to deploy computing nodes, it is necessary to design an incentive mechanism for edge computing service providers. Auctions are an effective incentive design solution. This paper designs an optimal auction mechanism to maximize the revenue of edge computing service providers, which ensures the two important attributes of individual rationality and incentive compatibility and ensures the feasibility of allocation and efficient use of resources. Specifically, we designed a system model for pricing and allocating edge computing service providers in the Internet of Vehicles environment, and transformed the optimal auction problem of resources under the Internet of Vehicles into a mathematical programming model of the optimal auction with constraints. And designed a matching algorithm, allocation algorithm, and price calculation algorithm based on a neural network. Finally, we experiment and analyze the algorithm. The simulation results show that the proposed scheme is superior to the VCG algorithm in terms of revenue and resource utilization.