An Immunity-Based Distributed Multiagent Control Framework

This dissertation, "An Immunity-based Distributed Multiagent Control Framework" by Wing-ki, Vicky, Wong, 黃穎琪, was obtained from The University of Hong Kong (Pokfulam, Hong Kong) and is being sold pursuant to Creative Commons: Attribution 3.0 Hong Kong License. The content of this dissertation has not been altered in any way. We have altered the formatting in order to facilitate the ease of printing and reading of the dissertation. All rights not granted by the above license are retained by the author. Abstract: Abstract of thesis entitled An Immunity-Based Distributed Multiagent Control Framework Submitted by Wing Ki Vicky Wong for the degree of Doctor of Philosophy at The University of Hong Kong in September 2006 This dissertation presents the theoretical development of a control framework for distributed multiagent systems based on the biological metaphor, known as Artificial Immune System (AIS). The control framework aims at providing an effective platform for multiagent coordination. Most of the numerous multiagent control paradigms focus on the control at either the individual or collective levels. Agents are therefore not coordinated in a fully autonomous manner. The sophisticated mechanisms of the biological immune system have therefore been adopted to develop a generic control framework for multiple autonomous agents. By adopting the mechanisms of the immune system, a fully decentralized control framework is developed in which agents exhibit the properties of self- organization and self-regulation in a dynamic environment. The AIS-based control framework comprises four components: the AIS agent architecture, the responses manipulation algorithm, the AIS-based behavioral model and the immunity-based regulation mechanism. These four components establish a comprehensive framework for the control of mobile agents at both individual and collective levels. The control at the individual level is constituted by the AIS agent architecture and response manipulation algorithm. The architecture defines the internal control for individual agents with a set of AIS-based functions. These functions endow the agents with autonomy to perceive, communicate and make decisions. A novel algorithm is derived for the responses manipulation that controls the action generation function of the AIS agents. This algorithm exhibits the properties of specificity and diversity. Based on the innate and acquired immunities, agents can generate actions with four responses and can acquire new functionalities to solve complicated problems based on past experiences. The control at the collective level is constituted by the behavioral model and regulation mechanism. The behavioral model defines a set of immune inspired behaviors, known as primitive behaviors. These primitive behaviors are emergent properties that result from the local interaction between agents and their environment. The regulation mechanism provides the theoretical underpinnings that give rise to effective agent coordination. Following the regulation of immune responses, three threshold measures are adapted, namely wandering threshold, affinity threshold and activation threshold. These thresholds regulate the behavior and activities of AIS agents such that they are self-organized and can adapt to a dynamic environment. By integrating the four components of the control framework, a computational algorithm for solving multiagent control tasks is developed. The algorithm resembles the operation of the immune system, which effectively control agents at both individual and collective levels. The theoretical development of the AIS-based control framework is described and a series of analyses and simulation studies are carried out to evaluate and demonstrate the effectiveness of the control framework. The results of the studies demonstrate that the AIS-based control framework achieves superior performance in terms of flexibility, scalability, robustness and sel