This paper proposes a service for multicast video distribution based on the active network paradigm. Active nodes can provide soft-state and perform customized computations on a per-user or per-application basis. These functionalities allow the deployment of an efficient and flexible service that can dynamically adjusts itself to the bandwidth availability. The source learns about the best transmission rate of the video by receiving feedback messages from receivers.A fusion mechanism is used to prevent feedback-message implosion.Simulation results show that a capsule filtering level of 96% can be achieved by using the fusion mechanism.
video distribution; active networks; multicast; adaptive service
