1 相关工作
1.1 无人机集群拓扑优化
1.2 空中计算
2 系统模型
3 系统优化
3.1 基于谱图理论的优化方案
3.1.1 拓扑优化策略
3.1.2 时帧长度优化
3.2 基于DQN的优化方案
3.2.1 状 态
3.2.2 动 作
3.2.3 奖 励
4 仿真结果
表1 仿真参数Table 1 Simulation parameters |
| 参数 | 数值 |
|---|---|
| 总带宽 | |
| 信道数 | |
| 信号功率 | |
| 优化权重 | |
| 时隙最小值 | |
| 奖励折扣 | |
| 学习率 |
ACTA AERONAUTICAET ASTRONAUTICA SINICA >
Optimization strategy for UAV swarm topology based on over-the-air computation
Received date: 2025-10-13
Revised date: 2025-11-06
Accepted date: 2026-02-27
Online published: 2026-03-19
Supported by
AVIC United Technology Center for Intelligent Decision-making and Coordinated Control Mechanism Model Research
To address the urgent demand for efficient information exchange in Unmanned Aerial Vehicle (UAV) swarms for emergency rescue scenarios, this paper innovatively integrates Over-the-Air Computation (OAC) technology with spectral graph theory and proposes a novel topology optimization strategy oriented to low-latency and high-consistency communication. The inherent correlation mechanism among communication topology structure, time synchronization accuracy and state information consensus convergence is explicitly uncovered, and a corresponding theoretical quantitative model is established to characterize the interactions among the three factors. On this basis, a bi-level joint optimization framework is developed: the upper layer optimizes the network topology based on spectral graph theory criteria, and synergistically improves synchronization performance and consensus convergence rate by regulating the node degree distribution and loop characteristics of the swarm network; the lower layer adaptively adjusts the MAC-layer frame length to minimize the end-to-end communication latency. The proposed topology design criterion breaks through the limitations of traditional point-to-point communication paradigms and realizes the balanced optimization between the convergence condition and convergence rate of state information consensus. Extensive experimental results show that compared with the Deep Q-Network (DQN) based reinforcement learning benchmark, the proposed method achieves a maximum improvement of 37% in information collaboration efficiency, with drastically reduced algorithm execution time. Compared with traditional schemes, the proposed method exhibits prominent latency advantages in large-scale swarm scenarios, which provides solid theoretical support for efficient collaboration of UAV swarms in high time-critical application scenarios such as disaster emergency response.
Kai XIONG , Dunxing LONG , Ke ZHANG , Supeng LENG . Optimization strategy for UAV swarm topology based on over-the-air computation[J]. ACTA AERONAUTICAET ASTRONAUTICA SINICA, 2026 , 47(18) : 532902 -532902 . DOI: 10.7527/S1000-6893.2026.32902
表1 仿真参数Table 1 Simulation parameters |
| 参数 | 数值 |
|---|---|
| 总带宽 | |
| 信道数 | |
| 信号功率 | |
| 优化权重 | |
| 时隙最小值 | |
| 奖励折扣 | |
| 学习率 |
| [1] |
|
| [2] |
|
| [3] |
|
| [4] |
|
| [5] |
|
| [6] |
|
| [7] |
|
| [8] |
|
| [9] |
|
| [10] |
贾永楠, 田似营, 李擎. 无人机集群研究进展综述[J]. 航空学报, 2020, 41(S1): 723738.
|
| [11] |
刘雷, 刘大卫, 王晓光, 等. 无人机集群与反无人机集群发展现状及展望[J]. 航空学报, 2022, 43(S1): 726908.
|
| [12] |
|
| [13] |
|
| [14] |
|
| [15] |
|
| [16] |
|
| [17] |
|
| [18] |
|
| [19] |
|
| [20] |
|
| [21] |
|
| [22] |
|
/
| 〈 |
|
〉 |