Research On Resource Management In Space-Air Integrated Networks

The space-air integrated network(SAIN)is a key issue in the 6G communication technologies,and also is the basic information network infrastructure in the "13th Five-Year Plan" of China.SAIN helps to implement global seamless coverage and provide information service all over the world.SAIN includes satellites in the space and the multiple aircrafts in the air such as high altitude platforms(HAPs),airships,unmanned air vehicles(UAVs),etc.,and the ground station,control center,data processing center,and the terrestrial users.SAIN has wide applications for military and civil usage,emergency rescue,navigation,etc.However,the structure of SAIN has multiple layers,heterogenous elements,multiple resources and complex demands.Satellites in the space segment of SAIN move periodically,and the aircrafts in the air also have different moving speed.All these resources have different serving abilities.In addition,the boosting users have various demands.Hence,how to design the SAIN framework and leverage multiple heterogenous resources to provide users with various services,and achieve efficient resource utilization are key challenges and problems in SAIN.Therefore,in this paper,we focus on dealing with the framework and resource management for SAIN,including the software defined satellite network based resource allocation structure and corresponding strategies,cooperation of HAPs and low earth orbit(LEO)satellites for data collection,and cooperating LEO satellites and UAVs for data transmission and UAV trajectory.In detail,1.To improve the resource flexibility and efficiency of satellite networks,the software defined network(SDN)and network function virtualization(NFV)technologies are utilized and we design the resource allocation strategies based on the software defined satellite networks(SDSNs).Specifically,we have firstly designed the software defined time evolving graph(SDTEG)to depict the multiple resources as well as the multiple tasks.SDTEG can accurately represent the periodic and dynamic SDSN topologies.Then,we have proposed the SDTEG based virtual network function(VNF)orchestration to complete the resource allocation for SDSN.Due to the NP-hard solution in the large-scale networks,we utilize the special block-angular structure of the formulation and leveraged column generation.To guarantee the optimal solution,we design the branch-and-price algorithm,which can obtain the optimal solution more efficient than exhaustive searching,and thus the solution from branch-and-price algorithm can serve as the benchmark for other sub-optimal algorithms.Furthermore,to accelerate the solution in large-scale networks,we also propose an approximation algorithm for the master problem,and leverage the beam search for branching to obtain the upper bound and lower bound more faster.Simulation results validate the effectiveness of the proposed schemes,and the performance from different resource ability are inspected,which can serve as inspirations for LEO satellite network virtualization design.2.We consider joint HAP and LEO satellite networks in SAIN to provide services for the remote area users,in which HAPs serve as the stable base stations in the air and LEO satellites help with the collected data transmission.Since the resource availability of HAP based access and LEO satellite based transmission is limited,and the LEO satellite network is high-dynamic,it is more intractable for the resource management for users' data transmission.To cope with these challenges,firstly we propose the spaceair network model including HAPs and LEO satellites to serve the remote area users.Due to the resource limitation and multiple demands,to satisfy more users,we design the restricted three-sided matching with size and cyclic preference lists algorithm to deal with the matching among users,HAPs and LEO satellites in one quasi-static time slot.Moreover,to alleviate the complexity from high dynamic topology of LEO satellite networks,we propose the two-layer matching algorithm,including the GaleShapley based many-to-one matching between users and HAPs,and the random path to pairwise-stable matching based many-to-many matching algorithm between HAPs and LEO satellites.Extensive simulations verify the proposed algorithms and the applicability for different scenarios,and we can choose the appropriate algorithm for different scenarios to efficiently manage the space-air resources.3.We propose the model of LEO satellite network assisted and UAV-based remote data collection and transmission in SAIN,and optimize the energy-efficient UAV trajectory.Two types of data are considered in the model:delay-tolerant data and delay-sensitive data,and we design the system model by combining two types of data transmission.In detail,the delay-tolerant IoRT data leverages the UAV carry-store mode to earth:the data are collected,stored and carried by a UAV,and fly with the UAV to the ground destination and then,use the optical cable to the procession center.Due to the large delay of UAV carry-store mode,it is not applicable for the delay-sensitive data,which should be transmitted back to ground data processing center in a limited time.Since LEO satellite networks can provide low delay transmission,we consider leveraging LEO satellite networks to relay the delay-sensitive data collected by UAVs.Specifically,the data collected by UAVs is transmitted to LEO satellites,and then utilize the satellite-to-satellite(S2S)links and satellite-ground links to ground stations,and use the high-speed optical cable to the data processing center.In addition,considering the limited payloads of UAV,at the same time of satisfying IoRT users' demands,we focus on minimizing the total energy cost of UAV.Moreover,to alleviate the problem tractability,we design the Dantzig-Wolfe decomposition and column generation based algorithm.Further,we present a heuristic algorithm for the subproblem to copy with the complexity of large-scale networks.Finally,simulation results verify the proposed algorithms,and the insight of combining UAV and LEO satellite networks for two types of data transmission is analyzed.