Variable Bit Rate Video: Characteristics, Modeling and Multiplexing

1 SOURCE MODELING This section describes the TES source model for MPEG-coded video, used in the performance study; TES models for other video coding schemes may be found in [8]. The model was constucted as a deterministic superposition of three component source models, one each for a subsequence of I-frames, P-frames and B-frames. The component bit rates were then interleaved (superposed) in the appropriate MPEG cycles as explained before. Each component subsequence was accurately modeled as a TES process; the details of TES process theory may be found in [6, 3, 4, 5], and an overview at an introductory level in [7]. The section is organized in two subsections. The first subsection describes the TES model construction of component bit rate subsequences for each type of frame. The second subsection describes the composite model and presents the modeling results. We begin by describing the generic construction of TES models from empirical bit rate measurements of MPEG-coded video. TES modeling differs from other modeling approaches in that it aims to fit a prescribed marginal distribution and a prescribed autocorrelation function simultaneously, via a stationary stochastic process, thereby capturing both first-order and second-order statistical properties of empirical time series. The construction of TES models is a two-phase procedure. In the first phase, one defines a background TES process fU