| Multi-agent systems (MASs) have emerged as the most natural intrinsic paradigm for modeling complex systems, such as supply chains (SCs). However, their development process remains quite involved and extremely time consuming, hindering wide-spread adoption in industrial-strength applications. This dissertation research focused on developing methodological frameworks addressing three important aspects in particular.; In the first part of the research, we develop a generic process-centered methodological framework, "Multi-Agent Supply Chain Framework (MASCF)", to fill the crucial gap of "lack of generic methodologies" for modeling SCs using MASs. The framework introduces the notion of process-centered organization metaphor, and creatively adopts Supply Chain Operations Reference (SCOR) model to a well-structured generic MAS development methodology, Gaia, for analysis and design of multi-agent supply chain systems (MASCSs).; Next, we develop an implementation framework, "software agent-component based framework", to simplify and speedup MAS development. Based on Java Agent DEvelopment Framework (JADE), it integrates multiple tools and technologies providing the necessary infrastructure support. With the help of pre-developed libraries of reusable components---organizational agents (OAs), SC agents (SCAs), behaviour and policy objects; the framework allows model developers to quickly configure a MASCS to simulate dynamics and study control and coordination issues. Being generic, flexible, and scalable; it supports development of either pseudo-centralized models by a single model developer, or distributed models by either a single or group of enterprises constituting a SC. The framework is unique; based on requirements it allows for representing different segments of SC network at either aggregated or detailed levels resulting in models of hybrid resolution. It facilitates studies involving intra- and inter-organizational dynamics, considering information asymmetry explicitly.; The frameworks were validated using "Tamagotchi" case study. Its SC was designed and analyzed, the generated output specifications were implemented, and multi-agent simulations were carried out. Finally, we extend the capability of our framework agents by incorporating "goal-directed" rationality through the well-known Belief-Desire-Intention (BDI) model of human practical reasoning and implement it using Jadex. As an elementary step, we convert Tamagotchi "Facility Agent" into a BDI-agent and conduct preliminary experiments to study the impact of rational investment decisions. |
