Swarm-Intelligent Formation Reconfiguration for UAV Networks Under Wind Disturbances and Communication Dropouts

Abstract

Unmanned aerial vehicle swarms are increasingly deployed for tasks that require cooperative motion and agile adaptation of formation patterns. In realistic low-altitude environments, the dynamics of each vehicle are significantly influenced by spatially and temporally varying wind fields, while the communication network experiences intermittent connectivity and packet dropouts. These effects interfere with formation keeping and make formation reconfiguration particularly challenging when agents must respond to mission-level events, obstacles, or vehicle failures. This article examines a swarm-intelligent formation reconfiguration scheme for multirotor-type UAV networks subject to linearized wind disturbances and stochastic communication dropouts. A discrete-time linear state-space model is formulated at both agent and network scales, incorporating wind as a structured disturbance and dropouts as random multiplicative factors on the communication graph. Formation reconfiguration is driven by distributed, swarm-inspired rules combining consensus interactions, role allocation, and local disturbance compensation. The resulting closed-loop dynamics are expressed in a compact linear form, which enables analysis of stability and convergence properties under time-varying communication graphs. Mean-square stability conditions are derived in terms of spectral properties of the expected interaction matrices and bounds on the disturbance energy. Simulation scenarios are conceptually discussed to illustrate the interplay between wind intensity, dropout rates, formation geometry, and reconfiguration transients. The discussion focuses on how local rules and linear feedback gains shape the global behavior of the swarm, and on how design parameters influence formation error, control effort, and robustness against combined environmental and communication uncertainties.