Dynamic Placement And Joint Resource Allocation For UAV-mounted Base Station

Nowadays,mobile networks are experiencing explosive growth.Influenced by people's daily activities,the spatio-temporal unevenness of mobile network traffic demand is becoming more apparent.The uneven spatio-temporal traffic demand poses a significant challenge to the deployment of mobile networks base stations.Restricted by the static deployment of ground base stations(GBS),it is challenging to meet the uneven mobile traffic demand under the traditional mobile networks.Considering the spatio-temporal uneven characteristics of the traffic demand,this paper focuses on the unmannded aerial vehicle(UAV)-based dynamic base station deployment methods in UAV-assisted mobile networks.The UAV-assisted mobile network selects areas to be assisted based on the results of traffic demand analysis based on the results of traffic demand analysis and prediction.The type of UBS can be determinded by the nature of the traffic demand.Different from traditional GBS,UBS requires wireless backhaul links to connecte the ground core network,and the deployment location is flexible and controllable.The deployment location is related to the quality of fronthaul and backhaul links,and affects the wireless resource allocation.This paper studies the spatio-temporal correlation characteristics and prediction of mobile network traffic demand,and then study the deployment of two types of UBS,including hovering position,user allocation,and resource management optimization of rotary-wing UBSs,as well as the dynamic wireless resource allocation and feasible area restracted trajectory energy consumption optimization of fixed-wing UBS.The main innovations are as follows:(1)Research on mobile traffic demand prediction model.In order to explore the spatio-temporal two-dimensional correlation characteristics of mobile network traffic,this paper proposes the basic spatio-temporal traffic demand prediction models and spatio-temporal traffic demand multi-task learning architecture,which can solve the problem of traffic demand prediction for correlated cells.The spatio-temporal traffic demand multi-task learning architecture describes the relationship between the traffic demand of the local cell and correlated cells,which improves the prediction accuracy of the traffic demand.The experiments on real-world data demonstrate that the spatio-temporal two-dimensional mobile network traffic demand prediction model proposed in this paper can significantly improve the accuracy of the traffic demand prediction.Notably,the normalized mean square error is reduced from 0.057 to 0.039.The prediction results provide effective support for UBS deployment planning in UAV-assisted mobile network.(2)Research on wireless resource allocation and deployment placement optimization of multiple rotary-wing UBSs.As the wireless backhaul links of UBSs are sensitive to the propagation environment,this paper designs the joint resource allocation of fronthaul and backhaul links for both in-band and out-band wireless backhaul modes with the consideration of user fairness,designs the user association method of multiple dynamic base stations with load balancing,and designs the placement optimization algorithm of UBS on this basis.The joint optimization of UBS placement and wireless resource allocation improves mobile network performances and user experiences.The simulation results show that the joint fronthaul and backhaul optimization method proposed in this paper improves the average user rate and the fairness index of user rate by 49%and 47%compared with the traditional methods.(3)Research on wireless coverage method for fixed-wing UBS with minimum rate requirement.In order to provide wireless services with minimum rate requirement from constantly moving fixed-wing UBS,the dynamic resource allocation schemes are designed for both in-band and out-band backhaul modes,and the feasible flgiht area of UBS is determinded under the minimum rate requirement.Moreover,to reduce the flight energy consumption and make full use of the feasible flight area,a trajectory optimization algorithm for fixed-wing UBS is designed.The simulation results show that the UBS trajectory designed by the proposed method can reduce the energy consumption by up to 12%compared with the simple circular trajectory.