随着星链等大规模非静止轨道（NGSO）卫星星座的快速部署，NGSO卫星网络研究成为国际科技前沿热点。NGSO卫星网络用户间干扰可能限制系统容量，多种类型混合业务的动态变化则会影响卫星有限带宽的使用效率，亟需相关理论分析和设计。论文从星座系统模型、业务模型和动态资源管理方法的研究入手，针对NGSO卫星网络上行系统容量及动态资源管理方法开展研究，主要创新工作如下：首先，提出了一种NGSO卫星网络上行容量分析方法，研究了星座内、星座间干扰限制下的遍历容量，给出了NGSO卫星网络上行容量与系统主要参数的关系。通过建立用户到卫星的上行通信模型，利用随机几何理论分析上行系统中星座内部及星座之间的干扰情况，推导得到上行遍历容量理论表达式，给出遍历容量与系统主要参数的关系，并通过仿真得到验证，为NGSO卫星网络设计提供理论参考。其次，提出了一种基于卫星运行规律和业务信息预测的随机接入资源管理方法，设计了面向随机接入的动态资源管理方法，可较好地预测业务量，并有效降低系统接入开销。首先，设计了适用于NGSO卫星网络的业务预测及资源管理架构。在此架构下，提出了冗余优化的重复时隙ALOHA协议，通过预测业务需求，优化随机接入发包数目，在保证吞吐量不变的基础上平均提升了24%的能量效率。之后，设计了自适应更新间隔的接入控制及资源分配方法，以同时提升接入效率和降低信令开销为目标，优化资源管理方案及其更新间隔，在保证高接入效率的同时可节省97%的信令开销。最后，提出了一种混合业务下的联合资源管理方法，通过独立预测实时和尽力而为业务，为混合业务自适应分配卫星接入和传输资源，有效提升网络效用。引入网络效用概念衡量混合业务的上行资源效率，同时，以最大化总网络效用为目标建立混合整数优化问题，通过预测不同业务需求联合优化多种业务的接入控制参数、随机接入及业务传输资源分配方案。结果表明，在不同混合业务需求情况下，所提方法相较于传统方法可平均提升57%的网络效用，实现对上行资源的高效优化。
With the rapid deployment of large-scale non-geostationary orbit (NGSO) satellite constellations such as Starlink, the research of NGSO satellite networks has become a hotspot internationally. In NGSO satellite networks, intereference from other users may limit the system capacity and the dynamic changes of mixed traffic may affect the efficiency of limited resouces, and relevant analysis and design are urgently needed. Therefore, this thesis starts with the study of constellation system model, traffic model and dynamic resource management methods, and investigates the capacity and dynamic resource management methods for uplink communications. The main contributions are given as follows:Firstly, a method for uplink capacity analysis of NGSO satellite networks is proposed to study the ergodic capacity under intra-constellation or inter-constellation interference, and the relationship between the uplink capacity and the main parameters of the system is given. The uplink communication model is established, and the uplink capacity under intra-constellation interference or inter-constellation interference is analyzed using stochastic geometry. The analytical expressions for ergodic capacity are derived, and the relationship between the ergodic capacity and the main parameters of the system is given, which is verified by simulations, providing theoretical reference for the construction of NGSO satellite networks.Secondly, a random access resource management method based on the satellite oppertating pattern and the traffic prediction is given, and the dynamic resource management method for random access is designed, which can forecast traffic precisely and reduce the system access overhead effectively. At first, the traffic prediction and resource management scheme for NGSO satellite networks is designed. Based on the scheme, a redundancy optimized repetition slotted ALOHA protocol is raised, which is able to predict traffic and optimize the number of packets in random access procedure accordingly, improving 24% the energy efficiency on average while ensuring the throughput remain unchanged. Besides, an access control and resrouce allocation scheme with adaptive update interval is designed to improve the random access efficiency and decrease signalling overhead at the same time, which is able to optimize the resource management and its update interval. Simlulation results show that it can reduce 97% signalling overhead while keeping high random access efficiency.Finally, a joint resource allocation method for mixed traffic is designed, which is able to forecast real time traffic and best effort traffic separately and allocate resources for mixed traffic adaptively, improving the network efficiency. The network utility is introduced to measure the uplink resource efficiency of different kinds of traffic, and a mixed integer optimization problem is set up to maximize the total network utility. The problem is solved based on predicted traffic by optimization methods to get the optimized results for mixed traffic, including access control paprameters and resource allocations for random access and data transmission. Results show that the proposed method is able to improve the network utility by 57% on average compared with the traditional method under different mixed traffic demands, achieving efficient optimization of uplink resources.
