Minimize the Effects of Network Pathologies on Path Planning of Mobile Agents

• Main Authors: Yen Hsu Chen, 陳彥旭
• Other Authors: Chien Chiao Yang
• Format: Others
• Language: zh-TW
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/65489216273055820902

碩士 === 國立臺灣科技大學 === 電子工程系 === 91 === Though the network and distributed system have been extensively used now, the stability of network can''t satisfy the requirement of most of the user yet. Because of the network loading is increasing rapidly from day to day, however, the infrastructure of network can''t process the increased network loading effectively. It causes many network pathologies. On the other hand, limited by the traditional distributed technigue, a time-consuming and laborious on-line operation mode, network users need a new technique which can improve the operation mode and productivity effectively.Contrast to the traditional techniques, mobile agent, a technique in distributed system, has been proved that it has many advantages than traditional techniques. They include the load decreasing in network transmission, off-line computing, and load sharing. In this thesis, we focus on the network pathologies in mobile agent system when agents roam on Internet. We hope that the delay or even abnormal termination for mobile agent’s task made by the network pathologies can be decreased effectively. The Internet’s autonomous system is what we concerned in our research context. First, we analyze the pathology conditions and network errors in today’s Internet and then we devise a solution correspondingly. We propose a mechanism to help the mobile agents to deal with network pathologies. The process involves three steps. First, we design several kinds of acknowledgement and then compute the response time accordingly. From those computations, we can know whether mobile agents encounter transmission delay or not. If the delay did happen, a new route search will be triggered. During the searching of new route, the mobile agent will continue to detect the platform''s state if the platform can not be connected. The key point of our research is on the path planning and the route searching.The proposed method let the mobile agent''s route can change with network state immediately and dynamically.