Distributed Interference-Aware Traffic Offloading and Power Control in Ultra-Dense Networks: Mean Field Game With Dominating Player

Both complicated interference and frequent handoffs exist in 5G ultra-dense small cell networks (UDNs), which significantly degrade the system performance. Therefore, a novel power control and simple traffic offloading scheme is proposed to handle the interference and handoffs. Owing to the severe interference coupling, this problem is formulated as a dynamic stochastic game (DSG) between small cells, which can characterize the stochastic property of channel states. Nevertheless, it is hard to solve the DSG system when the number of players exceeds 2. Therefore, to solve the DSG system in UDNs, we use the mean field approximation theory and convert the DSG into a mean field game (MFG) in which the game equilibrium is analyzed by solving two tractable partial differential equations, from the global and individual perspectives. In contrast to the conventional MFG works, there usually exists a dominating player, which introduces higher interference due to the location or larger transmit power in UDNs, for assumed generic player. Consequently, we formulate the framework of MFG with a dominating player and lots of generic players. Within the formulated framework, first we design the combined cost function considering both the aggregate interference and signal-to-interference-plus-noise ratio performances for both the generic player and its dominator. Then, we derive the corresponding Fokker-Planck-Kolmogorov and Hamilton-Jacobi-Bellman equations. Next, we propose a novel interference-aware traffic offloading and power control policy. Finally, we present numerical results showing the energy and spectrum efficiency performances of the proposed policy compared with the policy without traffic offloading.