Predictive Encoder and Buffer Control for Statistical Multiplexing of Multimedia Contents

Abstract

International audienceStatistical multiplexing of video contents aims at transmitting several variable bit rate (VBR) encoded video streams over a band-limited channel. Rate-distortion (RD) models for the encoded streams are often used to control the video encoders. Buffering at the output of encoders is one of the several techniques used to smooth out the fluctuating bit rate of compressed video due to variations in the activity of video contents. In this paper, a statistical multiplexer is proposed where a closed-loop control of both video encoders and buffers is performed jointly. First, a predictive joint video encoder controller accounting for minimum quality, fairness, and smoothness constraints is considered. Second, all buffers are controlled simultaneously to regulate the buffering delays. This delay is adjusted according to a reference delay constraint. The main idea is to update the encoding rate for each video unit according to the average level of the buffers, to maximize the quality of each program and effectively use the available channel rate. Simulation results show that the proposed scheme yields a smooth and fair video quality among programs thanks to the predictive control. A similar buffering delay for all programs and an efficient use of the available channel rate are ensured thanks to the buffer management and to the predictive closed-loop control

Impact of Playout Buffering on Mobile TV Performance

“…However, the operator must guarantee a minimum bitrate to these services, so as to have minimum video quality level [18]. This latter bitrate, which we denote by d min T V is the parameter that defines the admission control before applying statistical multiplexing policies to re-allocate the unused bandwidth to enhance the video quality [19]. Hence, the actual bitrate d T V is equal to d T V which is itself equal to d min T V for constant bitrate services.…”

“…where Q b is the transition probabilities matrix defined in equations (17), (18), (19) and (20) and Π 0 is the initial state distribution (no doubt the system evolution will begin by a state (∀a, b = 0, ω = 0)).…”

A consensus approach to PI gains tuning for quality‐fair video delivery

“…In [35], we compared different PI tuning methods; here, we extend the comparison to other controller types. In particular, we compare the performance of PI controllers, tuned with our new method, with the PI controllers tuned as in [7], and with the controller schemes in [2] and [5].…”

“…of stored videos. Lagrangian optimization frameworks are considered in [4,5]. In [4], the sum of the achievable rates is maximized while minimizing the distortion difference among streams.…”

“…In [4], the sum of the achievable rates is maximized while minimizing the distortion difference among streams. In [5], the average quality is maximized under total rate constraints, and constraints on the peak signal-tonoise ratio (PSNR) discrepancy between streams. Both approaches require to gather all RUCs of the streams at the control unit.…”