Unmanned aerial vehicles (UAVs) with multi- connectivity (MC) capabilities efficiently and reliably transfer data between terrestrial networks (TNs) and non-terrestrial networks (NTNs). However, optimally sharing and allocating spectrum and power resources to maintain MC while ensuring reliable connectivity and optimal performance remains challeng- ing in such networks. Channel variations induced by mobility in UAV networks, coupled with the varying quality of service (QoS) demands of heterogeneous devices, resource sharing, and fairness requirements in capacity distribution pose challenges to optimal resource allocation. Thus, this paper investigates resource allocation for QoS-constrained, MC-enabled, dynamic UAVs in an integrated TN-NTN environment with spectrum sharing and fairness considerations. To this end, we consider three types of links: UAV-to-radio base station (RBS), UAV-to-UAV, and UAV-to-HAP. We also assume two types of UAVs with diverse QoS requirements to reflect a practical scenario. Consequently, we propose two algorithms. The first algorithm maximizes the capacity of UAVs-RBS and UAVs-HAP links while ensuring the reliability of the UAV-UAV link. To achieve this, the algorithm maximizes the collective throughput of the UAVs by optimizing the sum capacity of all the UAV-RBS and UAV-HAP links. Next, to provide constant capacity to all links and ensure fairness, we propose another algorithm that maximizes the minimum capacity across all links. We validate the performance of both algorithms through simulation

翻译：具备多连接（MC）能力的无人机（UAV）能够高效可靠地在陆地网络（TN）与非陆地网络（NTN）之间传输数据。然而，在此类网络中，如何在确保可靠连接和最优性能的同时，最佳地共享和分配频谱与功率资源以维持多连接，仍然是一个挑战。无人机网络中由移动性引起的信道变化，加上异构设备不断变化的服务质量（QoS）需求、资源分配以及容量分布中的公平性要求，都对最优资源分配构成了挑战。因此，本文研究了在具有频谱共享和公平性考量的集成TN-NTN环境中，面向QoS受限、支持MC的动态无人机的资源分配问题。为此，我们考虑了三种类型的链路：无人机到无线基站（RBS）、无人机到无人机以及无人机到高空平台（HAP）。我们还假设存在两种具有不同QoS需求的无人机类型，以反映实际场景。基于此，我们提出了两种算法。第一种算法在确保无人机-无人机链路可靠性的同时，最大化无人机-RBS链路和无人机-HAP链路的容量。为实现这一目标，该算法通过优化所有无人机-RBS和无人机-HAP链路的总容量来最大化无人机的集体吞吐量。接着，为了向所有链路提供恒定容量并确保公平性，我们提出了另一种算法，该算法最大化所有链路中的最小容量。我们通过仿真验证了两种算法的性能。