The cooperation of multiple Unmanned Aerial Vehicles (UAVs) has become a promising sce-nario in Space-Air-Ground Integrated Networks (SA-GINs) recently due to their widespread applications,where wireless communication is a basic necessity and is normally categorized into control and non-payload communication (CNPC) as well as payload communication.In this paper,we attempt to tackle two challenges of UAV communication respectively on establishing reliable CNPC links against the high mobility of UAVs as well as changeable communi-cation conditions,and on offering dynamic resource optimization for Quality-of-Service (QoS) guaranteed payload communication with variable link connectiv-ity.Firstly,we propose the concept of air control-ling center (ACC),a virtual application equipped on the infrastructure in SAGINs,which can collect global information for estimating UAV trajectory and com-munication channels.We then introduce the knap-sack problem for modelling resource optimization of UAV communication in order to provide optimal ac-cess points for both CNPC and payload communica-tion.Meanwhile,using the air controlling informa-tion,predictive decision algorithm and handover strat-egy are introduced for the reliable connection with multiple access points.Simulation results demonstrate that our proposal ensures an approximate always-on reliable accessing of communication links and out-performs the existing methods against high mobility,sparse distribution,and physical obstacles.