Proceedings of the international workshop on Workshop on multimedia information retrieval
| Hi-index | 0.00 |
The growing interest in Internet has resulted in an increasing deployment of parallel and distributed applications over wide area environments. While the traditional client-server model is limited by the numbers of concurrent users it can support and the need for dedicated servers, the new emerging Peer-to-Peer (P2P) model offers an attractive alternative for developing large scale applications that share processing and computing resources among computer nodes on wide area networks. In this thesis, we investigate how distributed real-time multimedia applications that require both processing and communication can be supported on peer-to-peer systems. We first propose a novel distributed and dynamic resource utilization based urgency (RUBEN) scheduling algorithm whose goal is to maximize the probability that the real-time applications meet their deadlines. Next we extend this work and study scenarios in which a multimedia streaming session consists of multiple sub-streaming paths and the streaming receiver's quality-of-service requirements need to be coordinated through feedback information to streaming source nodes and transcoding nodes. In the last chapter, we study another class of distributed applications that require real-time data dissemination among interested receivers in order to maintain a consistent application view. We demonstrate how to utilize filtering techniques to reduce message overhead in such scenarios while maintaining acceptable consistency ratio for the participating peers.