VERDIKT-Kjernekomp.og verdiskaping IKT

The primary goal of this project is to develop a robust infrastructure for real time video streaming over the Internet, making it possible for it to eventually replace broadcasting-based TV, and at the same time adding new features such as true video-on-d emand. Such a system cannot be implemented straightforwardly, since there is a significant difference between traditional broadcasting an video distribution over the Internet. In traditional broadcasting, the number of viewers will not affect the broadc aster; data is sent once, and is receivable by an unlimited number of users within reach. Distribution over the Internet will normally requires a separate channel of transfer for every single viewer! I.e., the data flow grows proportionally with the numb er of viewers, making it unusable for large scale video distribution. There are two solutions: (1) IP multicasting where a single transmitting computer can send its data once, but allowing for many receivers. (2) Traditional one-to-one network transfer technology, but make each viewer a new sender. Solution 1 is similar to broadcasting, but there are problems associated with it: (i) A multicasting infrastructure does not yet exist, and the construction of one would be difficult and costly. (ii) There i s no mechanism today that would make multicasting scalable to millions of viewers and senders. Therefore, we have chosen solution 2, an extension to today's P2P technology. It is today mostly used for sending huge amounts of data where the order in which data arrives is irrelevant, and there is no requirement that the incoming data stream is steady. Ignoring these requirements makes it easier to make efficient P2P networks, but it also means that partially downloaded files cannot be used until complete. I.e, current P2P solutions can't be used for video-on-demand, as one would not be able to start watching video until it had finished downloading. We believe this project will result in solutions