Traffic and Statistical Multiplexing Characterization of 3-D Video Representation Formats

Abstract

Abstract-The network transport of 3D video, which contains two views of a video scene, poses significant challenges due to the increased video data compared to conventional single-view video. Addressing these challenges requires a thorough understanding of the traffic and multiplexing characteristics of the different representation formats of 3D video. We examine the average bitrate-distortion (RD) and bitrate variability-distortion (VD) characteristics of three main representation formats. Specifically, we compare multiview video (MV) representation and encoding, frame sequential (FS) representation, and side-by-side (SBS) representation, whereby conventional single-view encoding is employed for the FS and SBS representations. Our results for long 3D videos in full HD format indicate that the MV representation and encoding achieves the highest RD efficiency, while exhibiting the highest bitrate variabilities. We examine the impact of these bitrate variabilities on network transport through extensive statistical multiplexing simulations. We find that when multiplexing a small number of streams, the MV and FS representations require the same bandwidth. However, when multiplexing a large number of streams or smoothing traffic, the MV representation and encoding reduces the bandwidth requirement relative to the FS representation.

Adaptive delivery of immersive 3D multi-view video over the Internet

“…Video sequences that contain large frame size differences over the time may exceed the network capacity and result in packet losses. For instance, Pulipaka et al in [35] showed that inter-view predicted, i.e., MVC encoded, streams present larger variations in the encoded frame sizes compared to the independently, i.e., no inter-view prediction, encoded video streams. Also, inter-view predicted video streams come at the expense of reduced error robustness due to the massive coding dependencies between the views.…”

A Poisson Hidden Markov Model for Multiview Video Traffic

“…end for 14: end for probabilities above. We can write the initial probability of being in state as (8) Then, we can express the transition probabilities as (9) The frame size distribution in each state is given by (10) Finally, the state duration is expressed as (11) Note that a single iteration of the estimation algorithm consists of calculating first the state sequence probabilities (4)- (7), which is the expectation step. Subsequently, we compute the new parameter estimates by averaging the observations weighted with the state probabilities (8)- (11), which corresponds to the maximization step of an iteration of the algorithm.…”

“…The type of data and the traffic characteristics [9] are quite different in multiview video services compared to classical video streaming. The traffic observed during a multiview video communication session is generated by dynamically multiplexing different encoded streams, corresponding to the hierarchically organized encoding of different views.…”

A novel approach for characterizing multimedia 3D video streams by means of quasiperiodic processes

“…The drawback is the increased amount of data needed to describe the extra information of 3D frames. To overcome this problem, several efficient video compression and coding techniques and formats have been implemented to provide 3D video services on transmission links with a relatively limited bandwidth [12], which complicate the design of efficient bandwidth management and estimation schemes, needed to reach the quality of service (QoS) and quality of experience (QoE) guaranteed for end users [8,12].…”

“…The study performed in this work focuses on the lowfrequency (LF) behaviour of compressed streams because, as well known from literature, 2D and 3D compressed videos are typically variable bit rate (VBR) streams [3,12], and they are typically characterized by "low-frequency" components for bandwidth allocation and management, and QoS guarantees [6,9,17]. For this reason, the high-frequency part of the spectrum of VBR streams (i.e., in the time domain, strong data variations over smaller time scales) is usually removed before modeling VBR sources [3].…”