SDG-CDA: Stackelberg Differential Games and Combinatorial Double Auctions-Based Pricing Mechanism in Cloud–Edge Environment

In this paper, we propose an innovative pricing mechanism for cloud-edge collaborative computing resources that combines Stackelberg differential games with combinatorial double auction. The scenario of trading heterogeneous computing resources between cloud data centers, edge servers and users is modeled as a two-stage game. The Stackelberg game equilibrium is solved by Hamilton-Jacobi-Bellman (HJB) equation to optimize the resource allocation and pricing between data centers and edge servers. Markov game with multi-agent reinforcement learning is used to ensure the optimal bidding strategy of users, while differential privacy mechanism is introduced to protect participants’ sensitive information. Experimental results show that the edge server utility is improved by at least 50% and the user utility by 30% compared to the baseline algorithm. The mechanism accelerates the convergence of the game process while protecting the privacy of auction participants, providing a novel and efficient solution to the resource allocation challenge in dynamic computing environments.