Abstract
In hyperdense heterogeneous networks (HetNets), efficient intercell interference management is an important issue. In this paper, using recent advances in mean-field game (MFG) theory, we propose a novel game-theoretic approach for interference management in HetNets. The intercell interference management issue in HetNets is formulated as two nested problems: an overlay problem at the macrocell base station (MBS) level and an underlay problem at the small-cell base station (SBS) level. In the overlay problem, the MBS selects the optimal action first, which satisfies its associated users with a minimum amount of cross-layer interference, taking into account the reaction of the SBSs. The underlay problem is then formulated as a noncooperative game among the SBSs. The mean-field theory is exploited to help decouple a complex large-scale optimization problem into a family of localized optimization problems. Thus, each SBS can implement its policy by using only its local information and some macroscopic information. In addition to the achieved tradeoff between spectral efficiency (SE) and energy efficiency (EE), our approach can substantially reduce the communication overhead and convergence time of interference management in hyperdense HetNets. Simulation results are presented to show the effectiveness of the proposed scheme.