With the rapid development of low-altitude intelligent networks (LAINs), the growing demand for data services poses significant challenges for the existing networks. At the same time, the airspace becomes increasingly complex due to the escalating count of low-altitude users. Although unmanned aerial vehicles (UAVs) carrying mobile base stations to provide communication services can effectively alleviate pressure on existing network infrastructure, they unfortunately face the dual challenge of sustaining reliable data transmission and guaranteeing UAV flight safety. Therefore, in this paper, we propose a collision-free UAVs communication model specifically designed for the hybrid low-altitude environment, incorporating both static and dynamic, as well as known and unknown obstacles. To efficiently support safe flight operations of UAVs, an artificial potential field (APF)-based collision probability map is constructed, enabling the UAVs to dynamically evaluate and avoid obstacles while maintaining high communication performance constrained by limited energy resources. To maximize energy efficiency in low-altitude environments with hybrid obstacles, an adaptive association multi-agent deep deterministic policy gradient (AA-MADDPG) algorithm is proposed to enable collaborative trajectory planning among multiple UAVs. Simulation results confirm that the proposed strategy enhances energy efficiency by 58.06% and reduces collision probability by 86.18%, achieving significant improvements in both communication performance and flight safety.

With the development of unmanned aerial vehicles (UAVs) and the diversification of low-altitude applications, the cooperation among UAVs with different capabilities and objectives offers an exciting prospect for achieving efficient and ubiquitous communication coverage. However, coordinating the cooperation and competition among heterogeneous UAVs is an intractable challenge. In this paper, we propose a novel centralized-distributed heterogeneous-UAVs intelligent communication network system, which addresses the cooperation-competition issue among heterogeneous UAVs through reasonable task allocation. Specifically, a hub UAV makes ground users (GUs) scheduling decisions based on global information and provides backhaul link support through trajectory optimization. Meanwhile, high-mobility distributed UAVs cooperate to ensure fair, efficient communication for assigned GUs. Although centralized user scheduling offers greater flexibility and better performance, it also faces the serious problems which includes time-varying local observation spaces, hybrid action spaces, heterogeneous state spaces, and reward discrepancies. To solve these problems, we propose a convolutional neural network-assisted heterogeneous-UAVs proximal policy optimization algorithm, which aims to jointly optimize UAV trajectories and user scheduling, maximizing the system’s total fair energy efficiency. The simulation results demonstrate that the proposed CNN-HUPPO algorithm outperforms the four multi-agent deep reinforcement learning (MADRL) benchmark algorithms and two baseline algorithms in terms of fairness and accumulative fair energy efficiency.

In the next-generation emergency communication networks, unmanned aerial vehicles (UAVs), serving as aerial base stations, have attracted increasing attention recently due to their high mobility and low cost. Therefore, this paper conceives a heterogeneous UAVs assisted intelligent communication network system, in which tethered UAVs (T-UAV) make ground users (GUs) scheduling decisions to avoid resource competition, while other UAVs cooperate to provide communication services. However, the limited onboard resources of UAVs and cooperativecompetitive problems among heterogeneous UAVs becomes key challenges in UAV-assisted intelligent emergency communication networks. In order to tackle these issues, we propose a heterogeneous multi-agent approximate policy optimization (HMAPPO) algorithm to maximize the total fair energy efficiency by jointly optimizing UAV trajectories and user scheduling. Simulation results demonstrate that HMAPPO outperforms other baseline algorithms in terms of energy efficiency of the system and fairness of GUs. Furthermore, benefit to the partial parameter sharing mechanism, the proposed HMAPPO significantly accelerates the training process compared to other benchmark algorithms.