Distributed Control of Emergence: Local and Global Anti-Component Strategies in Particle Swarms and Ant Colonies

Abstract

Although emergent systems have presented us with powerful modeling and optimization mechanisms, often measures of control are necessary to prevent unwanted emergence. In cases where messaging has to be kept to a minimum, central control approaches are not always favorable or feasible; while distributed control approaches could provide an alternative. In this paper, we present a new distributed control strategy: "lying agents'', whose functionality can be switched on to distract other agents from harmful behavior. We implement this strategy in the particle swarm and the ant colony optimization algorithms to study the problem of the tragedy of commons. The simulation results show that the distributed control strategy is powerful in achieving trade-off behaviors, although dependant on various factors in different algorithms, such as lying locations, liars' pheromone strength or the number of liars. The strategy can be used in many applications, some of which are given throughout the paper.