Network-Aware Genetic Algorithms for the Coordination of MALE UAV Networks

Abstract

Maintaining an ad hoc network infrastructure to cover multiple ground-based users can be achieved by autonomous groups of hydrocarbon powered medium-altitude, long-endurance (MALE) unmanned aerial vehicles (UAVs). This can be seen as an optimisation problem to maximise the number of users supported by a quality network while making efficient use of the available power. We present an architecture that combines genetic algorithms with a network simulator to evolve flying solutions for groups of UAVs. Results indicate that our system generates physical network topologies that are usable and offer consistent network quality. It offers a higher goodput than the non-network-aware equivalent when covering the communication demands of multiple ground-based users. Most importantly, the proposed architecture flies the UAVs at lower altitudes making sure that downstream links remain active throughout the duration of the mission.