A Metaheuristic for Autonomous and Self-Organizing Consensus Formation Inspired by Honey Bee Nest Site Selection Behavior
Type of Degreedissertation
MetadataShow full item record
This research presents a self-organizing system for engineering consensus among cooperating, distributed processes and studying consensus negotiation in natural systems that use the quorum sensing mechanism. The presented method combines metaheuristic techniques and a negotiation protocol based on honey bee nest site selection behavior in order to optimize the collective social utility of a group decision. Unlike existing negotiation and voting protocols, the proposed metaheuristic accommodates negotiation of consensus from among two or more decision values. It is decentralized, self-organizing, and does not require a fully-connected network in which each process can communicate directly with each other. Benefits of these attributes are that there is no central point of failure, and processes can make informed decisions using only information acquired from immediate neighbors, rather than requiring the consolidation and evaluation of global preferences. The proposed metaheuristic is modeled as an agent-based system using the Repast modeling and simulation framework, and this model is used to conduct simulation experiments in accordance with the Design of Experiments methodology to analyze the Honey Bee Consensus metaheuristic and its performance on multiple social network models. As the Honey Bee Consensus metaheuristic is an extension and new application of a recently-proposed agent-oriented Quorum Sensing pattern, the presented understanding of its parameter and topology influence is of value in steering the behavior of self-organizing systems based on the Quorum Sensing pattern language. Additionally, understanding how network topology influences consensus formation has applications spanning decision theory, social choice theory, network science, and control theory. Significant outcomes of this research include a description of the proposed metaheuristic and the identification of a quorum size to population ratio that provides optimal speed-accuracy tradeoff for a range of population sizes and social network models.