The use of unmanned airborne agents as relays for ground agents to ensure ground network survivability is gaining traction at both theoretical and practical levels, in research that targets contexts like search and rescue in disaster areas, farming with autonomous equipment, surveillance, internet of things, military operations, autonomous transportation systems, and many others. This comes with some challenges, which include the dynamics of the ground unit behaviors, the scalability of the system, the mobility model of the airborne agents, and the integration between ground and air agents. This paper contributes to addressing the above challenges by using a bio-inspired approach that combines swarm intelligence and evolutionary computation for providing ground network survivability for a wide range of ground movement patterns. The proposed approach models ground and airborne agents as a dual air-ground swarm that uses boids-like rules, and optimizes the movement of the airborne agents using an on-line real-time genetic algorithm with shadow simulation and prediction. Arguably, the proposed approach provides system scalability both size-wise and context-wise, while also offering a certain amount of integration between the ground and air swarms of agents. The results of the investigation demonstrate that the methods employed endow the airborne agents with the needed capability to ensure network survivability for complex ground activity, including resilience to changes in the ground movement pattern and good responsiveness to activities that have no pattern at all, such as uniform random walks.

在针对诸如灾区搜寻和救援，使用自动化设备进行耕作，监视，物联网等背景下的研究中，使用无人机载代理作为地面代理的中继器以确保地面网络的生存能力在理论和实践上都越来越受到关注。军事行动，自动运输系统等。这带来了一些挑战，其中包括地面部队行为的动态变化，系统的可伸缩性，空降特工的机动性模型以及地空特工之间的集成。本文通过使用生物启发的方法来解决上述挑战，该方法结合了群体智能和进化计算，可为各种地面运动模式提供地面网络生存能力。所提出的方法将地面和空中特工建模为使用类似于布迪斯规则的双重空地群，并使用带有阴影模拟和预测的在线实时遗传算法来优化空中特工的移动。可以说，所提出的方法在规模和上下文方面都提供了系统可伸缩性，同时还提供了地面和空中特工人员之间的一定程度的集成。调查结果表明，所采用的方法使空降人员具备了确保复杂地面活动的网络生存能力所需的能力，包括对地面运动模式变化的适应能力以及对根本没有模式的活动的良好响应能力，例如统一的随机游走。