Survivable Networks via UAV Swarms Guided by Decentralized Real-Time Evolutionary Computation

Abstract

The survivable network concept refers to contexts where the wireless communication between ground agents needs to be maintained as much as possible at all times, regardless of any adverse conditions that may arise. In this paper we propose a nature-inspired approach to survivable networks, in which we bring together swarm intelligence and evolutionary computation. We use an on-line real-time Genetic Algorithm to optimize the movements of an UAV swarm towards maintaining communication between the ground agents. The proposed approach models the ground agents and the UAVs as boidsbased swarms, and optimizes the movement of the UAVs using different instances of the GA running independently on each UAV. The UAV coordination mechanism is an implicit one, embedded in the fitness function of the Genetic Algorithm instances. The behaviors of the individual UAVs emerge into an aggregated optimization of the overall network survivability. The results show that the proposed approach is able to maintain satisfactory network survivability levels regardless of the ground agents' movements, including for cases as complex as random walks.

Analysis of UAV Thermal Soaring via Hawk-Inspired Swarm Interaction

“…In another model made for the application of mapping, adding additional behaviors, including boundary, boundary avoidance, velocity gradients, and eccentricity, caused a great increase in the efficiency of mapping tasks [ 20 ]. Others have considered adding genetic algorithms for the application of increasing the durability of the communication networks established by a UAV swarm [ 21 ]. Following these recent developments from exploring the use of Boids and modified agent-level interaction rules, our work specifically considers an analysis of the Boids behaviors and variants for the specific application of exploiting thermals to extend the operational lifetime of the swarm through soaring.…”