Mean-Field Approximation based Optimization in Ultra-Dense Small Cell Networks

• 1、Key Laboratory of ICD&CAD, Beijing University of Posts&Telecommunications, Beijing 100876

Abstract：Ultra-dense network will play a pivotal role in 5G. The large scale of networks makes the conventional resource allocation methods inadequate to tackle due to their high time complexity. In this paper, a novel approach for joint power and subcarrier allocation for optimizing throughput in ultra-dense small cell networks (UDSNs) is proposed adopting mean-field approximation (MFA) method based on mean-field (MF) theory. The problem is modeled as a dynamic stochastic game (DSG) between small-cell base stations (SBSs) due to their severe coupling in interference. Assuming a large homogeneous UDN deployment, this game is solved through mean-field approximation (MFA) method in which the mean-field equilibrium (MFE), i.e., system\'s optimal control policy, is derived by an iteration between systems mean state and control policy of SBSs. Simulation results show that the proposed approach achieves up to 24.7% gains in throughput compared to a baseline method which uses an adaptive transmission policy.

Keywords： Communication and Information System Ultra-Dense Small Cell Network mean-field dynamic stochastic game