摘要:We compare the performance of two real-time systems for communication of distortion-tolerant content such as audio and video over rate-varying channels. The basic difference between the systems is the placement of a buffer either before or after the rate-controlled encoder. For our evaluation we target signal quality versus end-to-end delay. We develop an analytical framework for their comparison, when the systems use a deterministic rate-varying channel. Furthermore, for stochastic rate-varying channels we use the Gilbert-Elliott channel model and suggest some heuristic rate-control strategies for the two systems when communicating over such a channel.Contrary to intuition it is shown that one of the systems performs better than the other system. As the analysis uses approximations, performance is also validated through simulations.