@article{XIONG2026, 
author = {Kai XIONG and Dunxing LONG and Ke ZHANG and Supeng LENG},
title = {Optimization strategy for UAV swarm topology based on over-the-air computation},
year = {2026},
journal = {Acta Aeronautica et Astronautica Sinica},
volume = {47},
number = {18},
keywords = {UAV swarms, state information consensus, Over-the-Air Computation (OAC), topology optimization, spectral graph theory},
url = {https://www.sciopen.com/article/10.7527/S1000-6893.2026.32902},
doi = {10.7527/S1000-6893.2026.32902},
abstract = {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.}
}