Performance studies of networked video-on-demand systems

In a multimedia service such as video-on-demand system, video streaming requires allocation of several communication channels. The overall cost of establishing a video-on-demand system also includes the channel cost. Several caching and batching schemes have been proposed in literature for the optimal utilization of these channels. A viewer is served with a mix of main and small (e.g. advertisement) movies. The requested movies not necessarily will be of same duration. This study deals with the system having limited number of channels and servicing of requests for mixed movies of different durations. The system is modeled for its behavior with a continuous time Markov chain (CTMC) process. The performance measures such as channel utilization, number of channels in use, blocking probability of channels, cost and duration after which the proposed system starts yielding profit etc. are quantified in terms of various design parameters like, total number of channels, users request arrival rate, and batching interval, request service rate etc.