Modeling and Enhancing Resilience against Cascading Failures in Large-Scale Unmanned Aerial Vehicle Swarm Networks

During cooperative combat missions, large-scale Unmanned Aerial Vehicle (UAV) swarms are highly susceptible to network-wide cascading failures. Due to node functional heterogeneity and resource constraints, external perturbations can trigger load redistribution that precipitates the functional collapse of the entire combat system. To systematically analyze and mitigate such risks, this study addresses two core issues: cascading failure modeling and resilience enhancement. First, addressing the limitations of existing models in characterizing heterogeneity and task-orientation, we construct a dynamics model integrating non-uniform task load generation with hierarchical node capacity constraints, and propose a load-weighted survival rate metric to quantify functional survivability in a load-aware manner. Second, targeting load imbalance, we propose a synergistic active defense framework combining physical topology reconfiguration with logical traffic scheduling. Within this framework, a C2-UAV deployment optimization model is established to minimize average load and load deviation. Additionally, an adaptive multi-path load allocation and fault recovery mechanism is designed, predicated on node survival probabilities. To solve the resulting non-convex optimization problem, an improved Wolf Pack Algorithm is developed. Finally, extensive simulations show that the proposed resilience enhancement scheme reduces cascading-failure propagation in the tested scenarios. In representative single-point failure scenarios, the number of consequent node failures is near zero; under 5% multi-node removal, the proposed scheme attains the highest mean survivability among the evaluated strategies across the tested task fields. The resulting framework links task-induced load, failure propagation, and resilience control for risk assessment in large UAV swarms.