建構一個行動式多重代理人系統的開發環境

Abstract

利用代理人的觀念來設計日益複雜的軟體系統是當前軟體發展的重要議題；軟體代理人方面的研究日益進步，基於分散式以及網路研究而產生的行動式代理人系統，以及基於人工智慧研究而產生的智慧型代理人系統是目前的研究重心。然而現今能處理行動式代理人以及智慧型代理人的系統，例如JAM系統，一次只能執行單一個代理人程式，是其不足的地方。本文提出了一個行動式多重代理人系統的架構，兼採以上系統的長處，並加入了可以同時執行多個代理人程式的能力，能有效地在分散式環境下協同作業解決問題。

Nowadays, utilizing agent techniques to help programming complex software systems is an important issue. The researches in agent systems evolve into mobile agent systems, based on distributed and internetwork systems, and intelligent agent systems, based on artificial intelligent systems. However, current system which owns both mobile agent and intelligent agent paradigm only allows a single agent once a time, for example, the JAM agent system. In this thesis we propose a framework for mobile multi-agent system, which supports both above features and multiple agents concurrently. In our implementation, the system allows the collaboration of mobile intelligent agents effectively and efficiently in distributed environments.