國立臺灣大學電機工程學系暨研究所顏嗣鈞2006-07-252018-07-062006-07-252018-07-062003-07-31http://ntur.lib.ntu.edu.tw//handle/246246/7914協同計算（Collaborative Computing）是一個術語，我們將會用它來取代一 般傳統群體軟體（groupware）的觀念，而群體軟體就我們所知，是指在非同步 方面合作，以電子郵件、討論區及一些相似性的軟體幫助我們以電子手段交換資 訊。協同運算的觀念可以將一些網路的資源分享出來給所需要的人使用，而這些 資源是可以即時方式所取得。 本研究的主要目的在發展一套跨資源的分散式資源搜尋與管理系統，並結合 智慧型軟體代理人進行實作。在第一年，本團隊完成一些理論上的分析，將網路 上資源量化並以ER 派區網路（ER Petri net）語言來描述網路上各種不同的資源 以及資源上的限制﹍等，加上引進了配對演算法（matchmaking）解決一些變動 環境極大的網路資源分配問題，網路資源的配置可以藉由配對演算法所造成的配 對樹中就可以反映出一些特性。第二年，本團隊規劃了一個以Linux 為主軸的系 統來模擬協同運算環境，利用修改一些公開共享軟體並將理論部份實做化，完成 我們所提出的資源管理服務，這些我們會在第二部份中說明，並進一步在第三部 分中簡述系統分析結果與討論。 本團隊未來將針對所設計的系統，做更進一步的測試與評估，並期以此系 統，驗證更深一層的學理研究，提供更多樣的資源管理服務。Collaborative computing, nowadays, is gradually used to replace the traditional concept of “groupware.” Generally speaking, groupware focuses on the asynchronous aspect of human collaboration such as email, discussion lists, etc. which helps people to electronically exchange information. Collaborative computing is a term describing a variety of resources where people need it; more importantly, people are able to use these resources in the real time aspect. The main purpose of our project is to develop a distributed resource-discovery and resource-manage system. In order to managing heterogeneous resources more intelligently, our implementation will combine software agents as a whole. Beginning in the first year, we have surveyed and analyzed currently worldwide activities about resource management. By introducing the ER Petri net, we established our system model to further describe the resource language and its limitation. When a job tells its requirement, matchmaking algorithm provides us the way to allocate resources with respect to their different availability. During second year, we implemented our system architecture based on PCs with RedHat Linux 8.0. With some open-source softwares, we designed three kinds of services: monitor, scheduling, and matchmaking. We would show some details and also analyze them in the second and third part of the report. In the future, our team will estimate and evaluate our system. Further more, we will be able to verify the system and supply more versatile resource management services.application/pdf215567 bytesapplication/pdfzh-TW國立臺灣大學電機工程學系暨研究所協同計算智慧型代理人派區網路資源分配配對演算法Collaborative computingsoftware agentER Petri netmatchmaking algorithmmonitorschedulingreporthttp://ntur.lib.ntu.edu.tw/bitstream/246246/7914/1/912213E002041.pdf