Video smoothing is a promising technique for reducing the bandwidth variability of video in order to improve network efficiency. This paper presents a general optimal video smoothing algorithm based on the concept of dynamic programming. The algorithm generates the optimum transmission schedule for different requirements by setting the constraints and the cost function accordingly. It can be used to study the smoothing of both stored video and real time video. In particular, for stored video, we show how the number of rate changes in the smoothed video is affected by the renegotiation cost and buffer size, assuming that the transmission rate is allowed to be lower than the reserved rate. For the real time system, we study the impact of various system parameters, including playout delay, client buffer site, and server buffer size, on the performance of video smoothing.
展开▼