| In emergency communication scenarios such as hot spots and natural disasters,UAV base stations are often required to provide temporary services for ground users.Because the channel conditions of ground users in each area differs significantly,the direction of the transmitting beam of the UAV base station will affect the system performance.At the same time,the number of users and data traffic in emergency communication scenarios are generally huge,the occupied communication resources will surge rapidly at the moment.Considering the above two problems,this paper optimizes multi-domain resources such as beam,power and users for 3D MIMONOMA UAV base station communication system while considering the mobility of UAV base station.The specific research content of this paper has the following two points.First,drone base stations use 3D beamforming and NOMA technology to serve emergency communication scenarios,considering downlink transmission when users are gathered,this paper presents an optimization problem aiming at maximizing the minimum user rate,and discusses the power allocation scheme and 3D beamforming factor design of NOMA from two aspects of known complete CSI and imperfect CSI.Aiming at perfect CSI,a power-beam scheme of ground base station is applied to UAV base station.Aiming at imperfect CSI,this paper proposes an estimating-search algorithm based on neural network.Through training,a model capable of estimating the optimal beam azimuth is obtained,and then the beamforming factor is obtained by combining with the exhaustive search algorithm to search other variables.Simulation results verify that the performance of the proposed algorithm is significantly improved than that of existing strategies such as random beam algorithm.Finally,the rate obtained by the above scheme is a function of the location of the UAV base station.Accordingly,this paper searches the best location of the UAV base station based on the PSO algorithm,and the simulation verifies that the search results of the PSO are consistent with the ergodic optimization results.Secondly,this paper studies the resource allocation scheme of multi-beam 3D MIMO-NOMA UAV using beam superposition when users are scattered.This paper first uses clustering method to group users,and then proposes a phased multi-beam resource allocation scheme.After inter-group power allocation,the multiple-beam system is resolved into several equivalent single-beam systems by analyzing the characteristics of the multi-beam system,so as to extend the perfect CSI power allocation and beamforming scheme in the point above to the multi-beam system.Simulation results show that the phased resource allocation scheme in this paper has obvious performance improvement compared with the existing multi-beam system based on beam superposition. |
