Abstract
Recently, by reason of progressing of the internet technology and high-speed network bandwith, it will come true to provide services of video-on-demand on internet. Now, Internet not only provides text and picture medias but also videoes.For a typical example, Chunghwa Telecom‘s MOD service. Because more people using internet than before, application of internet begins toward multimedia services field. Because of this, we use the technology of data mining and PreStore Mechanism (PSM) to help solving cache problem in VOD (video-on-demand) system.Using PSM, we can provide a simple and small video delivery system. It can help us reducing response time of waiting and plenty of human resources. Finally, it can also help us avoiding human fault on VOD delivery system. This thesis discusses the issue of cache mechanism of VOD/MOD system. Because the limit of internet bandwith, it is always necessary for VOD system putting hot moive on local video server/cache video server. In this case, VOD system will have good reponse time and reducing instant flush of network flow.We use data mining method to find out hot movies which general users maybe want to see. We also use data mining skill to prestore hot movies in local video server/cache video server. That did work for reponse time of waiting to seeing movies. The chapter 4 of this thesis provides simulation methods to analysis performance of VOD system.We find reponse time and performance of VOD system having massive progress. We also compare our PSM to LFU and LRU. We find the results of PSM’s hit rate better than LFU and LRU. The improvement on LRU and LFU for hit ratio is 26.6% averagely. The study establishes the cache mechanism in TVOD environment. Users can query the movie more efficiently on the TVOD system. It provides the TVOD system makers with a more perfect decision environment and services as a further reference for the development of an advanced management mechanism of TVOD system.