Although low earth orbit (LEO) satellites can provide high-capacity backhaul to serve the terrestrial network, the performance of terrestrial-satellite communication systems is critically influenced by the coupling of user association and resource allocation in this integrating system, where user association includes small-cell base station (SBS)-user association and SBS-satellite association. In this work, we consider a cache-enabled terrestrial-satellite integrating network, in which LEO satellites provide backhaul for cache-enabled SBSs to serve ground users. Targeting at maximizing the downlink sum rate of the system and the number of accessed ground users, we formulate an optimization problem where user association and resource allocation of both terrestrial and satellite networks are joint optimized. Owing to the coupling relationship and integer programming nature of this optimization problem, we use Lagrangian relaxation to decouple and decompose it into two subproblems. We propose a user-division matching (UDM) algorithm by dividing all users into multiple user groups, which skillfully solves the first subproblem with multi-objectives. Afterward, to depict the nature of multi-connectivity sufficiently, the second subproblem is converted into a many-to-one matching game and solved by a modified Gale-Shapely (MGS) algorithm, which is highly efficient for different satellite constellations. Simulation results demonstrate the proposed algorithms can significantly improve the downlink sum rate of the system by 28.5–120.7 compared to the benchmark algorithms in the typical settings and balance the tradeoff between the downlink sum rate of the system and the number of accessed ground users. Moreover, it also shows that < 1% system performance loss can be obtained by the proposed method compared to the optimal solution.

尽管低地球轨道（LEO）卫星可以提供大容量回程服务于地面网络，但在这个集成系统中，用户关联和资源分配的耦合严重影响了地面卫星通信系统的性能，其中用户关联包括小-小区基站(SBS)-用户关联和SBS-卫星关联。在这项工作中，我们考虑启用缓存的地面卫星集成网络，其中 LEO 卫星为启用缓存的 SBS 提供回程，以服务地面用户。以最大化系统下行总速率和接入地面用户数为目标，提出了地面和卫星网络用户关联和资源分配联合优化的优化问题。由于该优化问题的耦合关系和整数规划性质，我们使用拉格朗日松弛将其解耦并分解为两个子问题。我们提出了一种用户划分匹配（UDM）算法，将所有用户划分为多个用户组，巧妙地解决了多目标的第一个子问题。之后，为了充分描述多连接的性质，将第二个子问题转化为多对一匹配博弈，并通过改进的 Gale-Shapely (MGS) 算法求解，该算法对不同的卫星星座非常有效。仿真结果表明，所提出的算法可以显着提高系统的下行链路总速率28.5-120。图 7 与典型设置中的基准算法进行比较，并平衡系统的下行链路总速率和接入地面用户数量之间的权衡。此外，它还表明，与最佳解决方案相比，所提出的方法可以获得 < 1% 的系统性能损失。