AI-Aided Downlink Interference Control in Dense Interference-Aware Drone Small Cells Networks

Recently, drone small cells (DSCs) has been brought into significant focus, which is the one key enabler for potentially facilitating terrestrial wireless communication systems. Meanwhile, in ultra-dense unmanned networks, artificial intelligence (AI) has been a useful and efficient tool for control and management of the multi-agents. This paper investigates a downlink interference control problem in ultra-dense unmanned networks with AI-aided approach, that each DSC can adjust its altitude to increase the data-rate. This problem is formulated as a mean field game (MFG) framework, an AI-aided method to make decisions. In this framework, each DSC controls its velocity to minimize the cost over a period, where the cost function is composed by the data-rate and height adjusting consumption. Meanwhile, in this model, we adopt the mean-field approximation (MFA) approach to derive the interference introduced from a large number of DSCs. Besides, the control strategy is described and explained by using the related Hamilton-Jacobi-Bellman (HJB) and Fokker-Planck-Kolmogorov (FPK) equations, respectively. Thus, a finite difference algorithm is proposed to solve the coupled partial differential equations, which can obtain the optimal altitude control strategy. The algorithm outputs show the optimal behaviors of DSCs in different environment scenarios. In additon, the simulation results verify that the proposed control strategy has better average signal to interference plus noise ratio (SINR) compared with the baseline method.