A Study of the Meta-Level, Emergent Behavior in a Synthetic Ant Colony through Modeling and Simulation

Peter Heck, Dept. of Comp. Science & Eng., Arizona State University, Tempe, AZ, USA
pheck@asu.edu

S. Ghosh, Dept. of Comp. Science & Eng., Arizona State University, Tempe, AZ, USA
sumit.ghosh@asu.edu

Current literature in ant colony behavior includes efforts to study the dynamic movement of ants through computer modeling, generating synthetic ant trails through simple rules of behavior, modeling the influence of genetic selection in the evolutionary behavior in ants, investigating the division of labor in ant colonies, and examining the potential trapping of ants in pheromone-induced trails, when the available food sources maintain unlimited supplies. This paper focuses on understanding, objectively, the emergence of the meta-level behavior of an ant colony as a function of the rules that govern the behavior of the individual ants, the interactions between them, and other environmental factors. A computer model of a synthetic ant colony is developed that permits the precise description of the unique behavior of each individual synthetic ant and its interactions with other ants. Under individual ant behaviors, this paper studies different search algorithms that are utilized uniformly by all ants, during foraging. In addition, this paper introduces the notion of creativity, wherein a few select ants adopt a very different search algorithm for foraging and expend greater resources in an effort to locate more food sources, where available. With respect to the key environmental factors, this paper analyzes the influence of food sources with finite supplies and the duration of the pheromone, before it vaporizes, on the foraging behavior of ants. The performance metrics, obtained through simulation, reflect the meta-level, emergent behavior, and include the simulation time within which all available food sources are located, the rate of identification of the food sources as a function of simulation time, and the energy expended by the ants in the course of foraging. Performance results reveal the presence of an optimal value for the duration of the pheromone that yields the fastest foraging behavior for a given search algorithm and a given number of ants within a fixed geography. Results also reveal that, for all search strategies, while the presence of creative ants may enhance the efficiency of foraging under long lasting pheromone trails, the advantage is reversed for trails with optimal pheromone duration.