Bridge the Gap Between ADMM and Stackelberg Game: Incentive Mechanism Design for Big Data Networks

Abstract

Alternating direction method of multipliers (ADMM) has been well recognized as an efficient optimization approach due to its fast convergence speed and variable decomposition property. However, in big data networks, the agents may not feedback the variables as the centralized controller expects. In this paper, we model the problem as a Stackelberg game and design a Stackelberg game based ADMM to deal with the contradiction between the centralized objective of the controller and the individual objectives from the agents. The Stackelberg game based ADMM can converge linearly, which is not dependent on the number of agents. The case study verifies the fast convergence of our game-based incentive mechanism.