The State of MPEG-DASH 2015
Presenting the past, present, and future of MPEG-DASH, the standard that's transforming the entire video landscape in a quiet revolution.
Learn more about the companies mentioned in this article in the Sourcebook:
As it is the first vendor-independent streaming standard, MPEG-DASH is a revolution. It’s a quiet but powerful one, and it’s transforming the digital video landscape the way MPEG-TS did during the past 20 years. How was it forged? What is the maturity level of the standard and its main differentiators? What are its main announced evolutions and innovations?
Let’s rewind to the last quarter of 2010, the year when Microsoft, Apple, and Adobe finally concluded a sort of Yalta agreement that applied to the technologies of adaptive bitrate (ABR) streaming over HTTP. Each of them provided a reference technology to reach a specific family of devices respectively through Smooth Streaming (HSS or MSS), HTTP Live Streaming (HLS), and HTTP Dynamic Streaming (HDS). HDS then benefited from the strong traction of Flash Player but suffered from being the latest entrant, whereas Smooth Streaming was strongly established in the premium video world. While Smooth Streaming and HDS relied on the MP4 container variants, HLS was built upon the foundations of the radically different MPEG-TS container. There was no real sign of convergence between the MP4-based ABR flavors; the necessary support of three formats was starting to inconvenience most content providers with wide content catalogs.
At this time, Netflix was already using its own HTML5-oriented ABR implementation based on Fragmented MP4, a prestandardization implementation of DASH. This standardization task was initiated in 2009 by the ISO MPEG Group. It gathered 15 industry proposals as candidates to be the main technology basis for DASH and worked closely with the 3GPP Group for almost 2 years, with Microsoft and Apple strongly pushing for their respective container approaches. Iraj Sodagar, DASH Industry Forum President and MPEG’s DASH Subgroup Chair, who also represents Microsoft in both organizations, remembers the beginning of the standardization process: “The market’s need and, in fact, demand for such standard were felt very strongly in the early days. The level of participation in the technical group also indicated that there is huge interest by technology companies to drive the standard. It was very competitive, intense, and at same time cooperative to get to results.”
In January 2011, MPEG Dynamic Adaptive Streaming over HTTP, mostly known as MPEG-DASH, finally reached the Draft International Standard status and allowed both TS and MP4 container approaches with common high-level fragmentation mechanisms. Even if not fully unified, MPEG-DASH was a concrete step toward simplification of the workflows. Apple will soon step back from this standardization movement, but the rest of the industry will stay focused on the ISO Base Media File Format (the standard succeeding MP4) track and its DASH lineage.
Building MPEG-DASH is an industrywide effort; while the MPEG group defines the core features of MPEG-DASH and its reference profiles, other standardization bodies are supplementing the core specification with additional recommendations and extended profiles in order to allow deployments in a specific context. For example, the DASH Industry Forum (IF), which consists of more than 80 companies from the ecosystem, proposes Interoperability Points (IOPs) covering over-the-top use cases; 3GPP and ETSI cover the use of MPEG-DASH in the mobile LTE Broadcast context; and DVB, EBU, and HbbTV cover the use of DASH in the digital and interactive TV contexts. All of these organizations communicate with each other through liaison documents and regular meetings, in order to ensure that they align their specifications in the best possible way.
The history of MPEG-DASH is all about shifting from proprietary standards to an open one.
“Initially the multiple consortia and standard groups working on dynamic adaptive streaming was a double-edged sword,” Sodagar says. “Its positive edge was the demand, need, and appetite for a standard. Its risky edge would have been to end up with multiple standards. One of the key successes of this project is alignment of all consortia and standards groups to collaborate with MPEG and in fact to let MPEG develop the core standard (one place and one path rather than parallel paths), and then work with DASH-IF on deployment, which is on-going.”
Some 16 organizations are now involved, as ATSC is working on integrating MPEG-DASH in its next specifications version. The obvious difference with previous proprietary IP video technologies precisely relies in this multilateralism. No one fully owns the MPEG-DASH standard, so no company can dictate its views as community reviews always precede final documents adoption. Some might object that this multilateralism is also a source of dilution of the standard, as several flavors of MPEG-DASH can be used in different contexts, but the reality is that, so far, the situation is under control. Each delivery scope (OTT, mobile, broadcast) builds its appropriate MPEG-DASH variant in respect to the core specification and reuses other existing open standards and interoperability points created by other groups. MPEG-DASH quickly managed to become the Swiss Army knife of the streaming video industry by adopting a flexible framework approach rather than a rigid and monolithic one.
But how can DASH stay consistent considering the various flavors implemented throughout the industry? “Define a few good flavors, use the same core standards and same core features as much as possible for all industry segments, and harmonize interoperability points as much as possible to a few good universal ones,” Sodagar says. “Make it easy to convert one flavor to another. Make sure that different flavors (interop points) are easily repackageable to each other. Both MPEG in the core standardization and also DASH IF along with other consortia have worked hard to achieve consistency and in fact reduce the number of Interop points.”
From DASH V1 to V2, DASH IF Fosters Interoperability
At the end of November 2011, MPEG finally ratified the MPEG-DASH standard, which became ISO/IEC 23009-1 in the ISO codes references system. This first version specifies the manifest format (Media Presentation Description, or MPD in the DASH vocabulary) based on XML and the media segments format based either on ISO Base Media File Format (ISO/IEC 14496-12) or MPEG-2 Transport Stream (ISO/IEC 13818-2). As the target is streaming services over the internet, HTTP is used as the default transport mechanism but is not considered mandatory. The core specification also defines some profiles: ISO Base Media File Format On Demand, Live, and Main, as well as MPEG-2 TS Main and Simple, and finally the Full profile, which includes all of the others. But the lack of mandatory audio/ video codecs in the specification ended up creating an interesting opening to all kinds of technology assemblies, but not a precise set that could be implemented on a lot of devices in an interoperable manner. Thomas Stockhammer, director of technical standards at Qualcomm, editor of the MPEG-DASH specification, and chairman of the DASH-IF IOP group, explains why the core DASH specification was designed so open: “At the time when DASH standardization was initiated, Adaptive Bitrate Streaming was already launched with proprietary formats including HLS, Smooth Streaming, and Adobe HDS. In addition, legacy formats such as the MPEG-2 TS were required by broadcast-centric organizations. This led to the approach to focus on core commonalities, but at the same time permit application profiles that are aligned with existing ecosystems.”
The core DASH specification scope (marked in red)
This is why the role of the DASH IF was critical in transforming DASH from a good idea into a sustainable industry solution. In February 2012, streaming companies (now more than 80 of them) had gathered together into an industry alliance aimed at promoting DASH and creating recommendations for implementation. A year later, they released the draft version of the DASH-AVC/264 Implementation Guidelines, which defined two interoperability points (DASH-AVC/264 and DASH-AVC/264 SD) combining the DASH v1 core specification requirements restricted to ISO BMFF with H.264/AVC Progressive High or Main(for SD) Profile up to 720p as the video codec, HE-AAC v2 as the audio codec, SMPTE Timed Text as the closed captioning format, Common Encryption as the protection scheme with multiple DRMs signaling possibility, and HTTP1.1 as the transport protocol. In June 2013, the guidelines were released officially, but a new 2.0 version quickly superseded it in in August, adding the HD support to 1080p (through the new interoperability point DASH-AVC/264 HD) and bringing multi-channel audio extensions from Dolby, DTS, MPEG Surround, and HE-AACv2 level 4 and 6.
The DASH IF DASH-AVC/264 v2 guidelines are the standard when it comes to streaming over-the-top in DASH. However, DASH IF didn’t stop its efforts toward propelling DASH to the forefront of streaming innovation; it released a preliminary DASH-HEVC/265 guidelines document in September 2013, defining three new interoperability points (DASH-HEVC/265, DASH-HEVC/265 1080p 8bit and 1080p 10bit) with support for HEVC up to 1080p 10bit. Since that time, DASH IF has merged the two families of AVC and HEVC guidelines into one DASH IF Implementation Guidelines document, added the support for closed captioning through CEA-608/708, and aligned the guidelines with the second version of the DASH core specification that was published in May 2014. This specification extends the first edition to include features such as media timeline events to support server-driven interactions and updates, empty Periods to support media blackouts signaling, content asset identifiers to categorize main and advertising content, and improved ad-insertion support.
The map of MPEG-DASH profiles after the second edition of ISO/IEC 23009-1
New Extended ISO-BMFF On Demand and Extended ISO-BMFF Live profiles have been created to cover the new features of the DASH core specification v2. DASH-IF Implementation Guidelines v3 should be published in early 2015, as DASH IF now requires offering companion test vectors to illustrate each part of the guidelines. Two new sections (consolidating ISO/IEC PDTR 23009-3 about DASH Implementation guidelines) are also to appear in the DASH IF guidelines, one dedicated to live services, the other to ad insertion, in order to cover most requirements of modern streaming services and provide a set of best practices up to a detailed operational level. Stockhammer sums up the situation after the guidelines v3 release and the continuous commitment from DASH IF: “The relevant DASH specifications for OTT delivery may be considered complete, but at the same time as DASH is used in different domains, so we need to be able to react to this. And the integration of new codecs, DRMs, formats, and metadata will continue. We already know that DASH is considered to be used in broadcast services, for lower low-latency live services, as well as for the distribution of new audio-visual experiences such as UHD video and immersive audio. The DASH IF will continue to support these efforts.”
The DASH IF Guidelines: standards combinations and core DASH restrictions at a glance
Upcoming DASH IF Guidelines Evolutions
One of the other main areas of work and discussion during the past year at DASH IF has been the interoperability proposal for backend DRM systems—something that has historically depended on proprietary implementations. The main traction factor for this rapprochement of competing companies is the perspective of Common Encryption delivering the promise of having multiple DRMs used in a unique set of files for each content, as well as the predictable use cases of content providers combining several DRM License Service Providers, or adding new DRMs when they appear.
MPEG-DASH is slowly but surely becoming the main competitor to HLS, driven by adoption by major players and intrinsic strengths. Here's who's using it now, who's going to be soon, and what challenges still need to be addressed.
With the rise of MPEG-DASH and the NAB-related announcements from both Microsoft and Adobe on the topic, we sat down with both companies to discuss the status of DASH support as well as their legacy ABR protocols, Smooth Streaming and HDS
What the online video industry needs is simple standards for reaching all viewers. But when have standards ever simplified online video?
The industry is turning away from plug-ins and embracing HTML5 everywhere. Here's how the vendor-independent streaming standard is gaining momentum.