Video Adaptive Streaming over HTTP
Nowadays, video traffic has contributed a significant part of transferred data over the Internet, based on recent publications of reliable organizations and network pioneers. Following that trend, service providers are keen on improving the quality of service through various ways, such as highly efficient video compression approaches, good-and-fair bandwidth management strategies, and delivery technologies. Especially, delivery technologies have increasingly attracted many attentions of scientists and developers recently: Thanks to broadband connections and the abundance of Web platform, video streaming over HTTP has become a cost-effective means. In particular, MPEG/3GPP introduced a new standard for video streaming: Dynamic Adaptive Streaming over HTTP (DASH) to enable the interoperability in the industry [1].
Media delivery hierarchy in HTTP adaptive streaming
The general architecture of an HTTP adaptive streaming system consists of servers, delivery networks, and clients. In MPEG DASH terminology, to support adaptivity, a video is encoded in multiple versions (also called alternatives or representations), each of which is further divided into short segments. Video segments together with metadata are hosted at a server and will be requested by the client. In most cases, for each request from the client, the server will send one segment. Therefore, a video will be delivered by a sequence of HTTP request-response transactions. The version (low or high) of a requested segment is decided based on the metadata and status of terminals/networks [1, 2]. This opens an effective way to deal with the bandwidth variations of today's heterogeneity networks, but simultaneously poses a problem: How should a good adaptation strategy be defined and then designed for ensuring satisfied user experience?
[1]. T. Stockhammer, “Dynamic Adaptive Streaming over HTTP - Standards and Design Principles,” in Proc. ACM MMSys'11, California, US.
[2]. T. C. Thang et. al, "Adaptive Streaming of Audiovisual Content using MPEG DASH," IEEE TCE, 2012
Quality-driven adaptive streaming with JND (Just Noticeable Difference) quality metric:
Most of the existing methods only focus on how to provide as highest video bitrate as possible. We discussed that more attention on quality of experience should be needed. More particularly, bitrate may not reflect the effectiveness of an adaptation strategy. Thus, we brought the use of perceptual quality metrric to the table. In this context, Just Noticeable Difference is considered. We conducted quality tests to see how the changes of bitrate impact on end users’ perception. The data of quality is mapped to discrete video bitrate level (of the representations). The results prove that video bitrate in fact can be reduced remarkably without any significant distortion of experience (~ 1 JND, which is "essentially undistinguishable").
We present the adaptation problem in a graph where version selection is illustrated as a sequence of linear sections (path) connecting points (representing video versions) in a trellis. To balance the trade-off between video quality and buffer level without making the stream interrupted, we defined a cost function based on preset criteria and then design an adaptation algorithm choosing the best path with a desired cost (to satisfy the criteria).
Publication(s):
H. T. Le, H. N. Nguyen, N. P. Ngoc, A. T. Pham, H. L. Minh and T. C. Thang, "Quality-driven Bitrate Adaptation Method for HTTP Live-streaming," in Proc. IEEE ICC'15, London, UK.
Selected related papers:
T. C. Thang et. al, "An evaluation of bitrate adaptation methods for HTTP live streaming,", IEEE JSAC, 2014.
H. T. Le et. al, "Buffer-based bitrate adaptation for adaptive HTTP streaming", in Proc. ATC'13, HCM City, Vietnam.
A. B. Watson and L. Kreslake, “Measurement of visual impairment scales for digital video,” in Proc. SPIE 4299, 2001.
Adaptation strategy with smooth quality transition for VBR (Variable Bitrate) videos:
Various adaptation methods for HTTP adaptive streaming have been proposed in the past few years, but so far most of them only focuses on CBR (constant bitrate) videos, while research on HTTP adaptive streaming for VBR (variable bitrate) video is still limited. The video quality might be improved if the content is VBR-encoded, instead of CBR-encoded, since the video bitrate needs to be dynamically changed for periods of detail when degradation in quality would significantly impact viewers. However, the strong variations of video bitrate over time may result in buffer underflows even though the throughput is stable, together with throughput fluctuations, are challenging for video streaming research.
The proposed adaptation method must be able to cope with variations of throughput as well as video bitrate. To do that, we (1) used a long-term bitrate value as the representative bitrate of a version to help choosing appropriate quality at each time instance, (2) designed a quality-adaptive algorithm based on current buffer occupancy for suitable version selection to provide good Quality of Experience: (+) Conservative in quality switching (for quality stability) and (+) Prompt against undesired bitrate-throughput mismatch.
Publication(s)
H. T. Le, H. N. Nguyen, N. P. Ngoc, A. T. Pham and T. C. Thang, ”A Novel Adaptation Method for HTTP Streaming of VBR Videos over Mobile Networks”, Journal of Mobile Information System, 2016.
H. N. Nguyen, T. Vu, H. T. Le, N. P. Ngoc and T. C. Thang, "Smooth Quality Adaptation Method for VBR Video Streaming over HTTP," in Proc. ComManTel'15, Danang, Vietnam.
Selected related papers:
M. Seufert et. al. "A Survey on Quality of Experience of HTTP Adaptive Streaming," IEEE Comm. Surveys and Tutorials, 2015.
T. C. Thang et. al, "Adaptive Video Streaming over HTTP with Dynamic Resource Estimation," IEEE/KICS JCN, 2013.