Microsoft not only designed an OTT protocol, they also released the software (IIS), the file format for storage and the protocol for communication between the encoder and the origin server.
Smooth Streaming server file format
Smooth Streaming storage file format is PIFF/ISMV (see PIFF/ISMV). The file extensions are:
• ismv for files containing video and audio tracks • isma for files containing audio tracks only
The storage can be done in a multibitrate pre-fragmented unique file or in various pre-fragmented files for each bitrate. File is virtually split up into chunks when responding to a client request.
Server manifest
The server manifest is an XML file. Its file extension is ism for VOD and isml for live. It describes the relationship between media tracks, bitrates and files on disk.
<?xml version="1.0" encoding="utf-8"?>
<smil xmlns="http://www.w3.org/2001/SMIL20/Language"> <body>
<switch>
<audio src="./mezz.ismv" systemBitrate="96000" systemLanguage="eng"> <param name="trackID" value="1257" valueType="data" />
<param name="trackName" value="audio_eng" valueType="data" /> <param name="SamplingRate" value="24000" valueType="data" /> <param name="Channels" value="2" valueType="data" />
</audio>
<video src="./mezz.ismv" systemBitrate="800000"> <param name="trackID" value="2" valueType="data" /> <param name="MaxWidth" value="626" valueType="data" /> <param name="MaxHeight" value="352" valueType="data" /> </video>
<video src="./mezz.ismv" systemBitrate="1000000"> <param name="trackID" value="1" valueType="data" /> <param name="MaxWidth" value="852" valueType="data" /> <param name="MaxHeight" value="480" valueType="data" /> </video>
</switch> </body> </smil>
Figure 64: An encoder publishing a fragment to a multibitrate pre-fragmented ismv file
Protocol
In live, the encoder continually sends new fragments via HTTP POST to the origin into the same HTTP connection (thanks to the Transfer-Encoding HTTP header).
The origin server continually appends new fragment from the encoder to the ISMV file. To prevent having an infinite file, it sometimes does file rotation.
Multibitrate TS
The idea is to stream MPEG2-TS over UDP over IP multicast.
The advantages of this method over Smooth Streaming as pivot format are:
• since it uses IP multicasts, several devices can register to the same stream. This allows redundancy and monitoring
• MPEG2-TS is commonly used by broadcasters, they can reuse their infrastructure (encoders, probes, multicast infrastructure)
To have MPEG2-TS as an input for OTT formats we must solve two problems.
Make it multibitrate. This can be done in 3 ways:
• various SPTS over various multicast IP
• various SPTS over the same multicast IP but each SPTS with a different UDP port • one single MPTS where each elementary streams corresponds to a different bitrate
Make it "chunk" ready. The origin/packager receiving the multicast must be able to cut each bitrate in same
size chunks. Since each chunk must start with an I-frame, the encoder generating the multi bitrate must therefore synchronizes each bitrate by putting IDR I-frames at the same frequency. Then, depending on the coders:
• some will explicitely flag the chunk start by using MPEG2-TS fields
• others will simply consider that if all bitrates have an IDR I-frame at the same moment, this is a chunk start There are various multibitrate implementations:
• Envivio Genesis • Harmonic MBTS
EIA-608
Designed by Electronic Industries Alliance. Subtitles used in US and Canadian TV (analogic NTSC and later digital ATSC). It allows to transmit 2 characters per image. The character set is very limited so it can not be used for most international languages.
Teletext
It allows to transmit text data. It has been designed for Analogic European TV. It has been renamed DVB-TXT for use in digital TV. In DVB, this will be a separate elementary stream. Teletext informations are splitted in "pages". It can transmit information page such as weather and sport results. It can also used for subtitles (typically on page 888, 777 or 333)
DVB-SUB
DVB-SUB (ETSI EN 300 743) has been designed for Digital TV. It is much more powerful than teletext as it
allows text, images, various colour palettes, various subtitles for different languages, various subtitles for different aspect ratio (4:3 and 16:9). It is synchronised with the video by using MPEG2-TS timestamps.
TTML and DFXP
Timed Text Markup Language (TTML) is an XML format for describing subtitles on the web. This is a rec-
ommendation by the W3C, maintainer of HTML. It allows precise placement, coloring, formatting, but for text only.
<tt xmlns="http://www.w3.org/ns/ttml" xml:lang="en"> <body>
<div>
<p begin="00:00:22" end="00:00:27"> I’ll teach thee Bugology, Ignatzes </p>
<p begin="00:00:40" end="00:00:43"> Something tells me
</p>
<p begin="00:00:58" end="00:00:64"> Look, Ignatz, a sleeping bee </p>
</div> </body> </tt>
Figure 65: TTML example
Distribution Format Exchange Profile (DFXP) is based on TTML with some more constraints. It is specific for
SRT
SRT is a text file format invented by the creators of the software SubRip. It is very basic and does not support any kind of formatting. See an example in Figure 66.
Based on the SRT format, the W3C is working on a standard called Webvtt (Web Video Text Tracks). 1
00:00:22,000 --> 00:00:27,000 I’ll teach thee Bugology, Ignatzes 2
00:00:40,000 --> 00:00:43,000 Something tells me
3
00:00:58,000 --> 00:00:64,000 Look, Ignatz, a sleeping bee
The head end is the part of the IPTV network from which the stream are distributed. It is made up of video servers.
Figure 67: The ecosystem deployed for an ADSL network
DVB-to-IPTV Gateway
As described in OSI model applied to IPTV, a DVB-to-IPTV gateway will receive the stream from a DVB source and retransmit it to the IP network.
Anevia provides the Vialive solution for Telcos and the Flamingo product for the Hospitality and Corporate markets.
Transraters and Transcoders
To adapt the stream to the network and the Set-Top Boxes, it is sometimes required to transrate, reducing the bitrate by raising the compression ratio of changing the resolution of the video. We can also transcode to change the video codec used (h.262 to h.264).
To prepare the contents for OTT, transcoders must transcode in various bitrates with various compression ratio and/or various resolutions.
Offline transrating and transcoding can be done to prepare VOD content.
Live transrating and transcoding must be done in real-time and requires huge processing power.
Here are some transraters and transcoders manufacturers Anevia work with: Allegro, Ateme, Envivio, Elecard, Grass Valley, Harmonic, Elemental.
Video-On-Demand and recording server
A Video-On-Demand server is usually a server implementing the RTSP protocol (see RTSP).
Most Video-On-Demand servers are also able to record live streams. By combining a DVB-to-IP gateway with a recording/on-demand server we provide end users applications like:
Start-Over the ability to shift from live TV to a recorded stream on a Video-On-Demand asset previously recorded
starting at the beginning of the show
Pause TV also named timeshifting. The ability to shift from live TV to a recorded stream on a Video-On-Demand
asset in a Pause state
Catch-up TV the ability to play Video-On-Demand on older programs that have been recorded previously.
OTT packager and Origin server
TODO: Sales drawing about OTT
An OTT packager, in both live and VOD, will prepare the OTT contents. That includes transcapsulate them between pivot format to various OTT formats. That includes scrambling them in relation with a DRM server. That also includes filtering the incoming tracks to suit better the device.
An OTT Origin server is a web server that host OTT files.
Anevia provides the ViaMotion Plus, that can act as both a live and VOD OTT packager and Origin server. Allowed input formats are Smooth streaming and multibitrate TS. Possible output formats are HLS version 1 and 4, MPEG DASH, Smooth streaming 2.0 and 2.2, Adobe HDS.
Digital Rights Management
As seen in ??, when the contents need to be stream scrambled, a Digital Rights Management (DRM) infrastructure must be introduced.
It is often made of 2 components: a key server and a scrambler.
Some DRM providers Anevia works with: Verimatrix, SecureMedia, Nagravision. The ViaMotion Plus can act as a scrambler.
OTT Edge servers
OTT Edge servers are HTTP caching servers dedicated for OTT delivery. A geographically distributed network of Edge servers constitute a Content Delivery Network (CDN).
Anevia provides the ViaMotion Edge as the cache server, and the ViaMotion Balancer for geographical load balancing over various ViaMotion Edge.
Monitoring
To monitor the head-end, various polling and probing systems can be added.
Anevia provides the ViaManager Monitor (based on Skyline Dataminer) that will poll the Anevia products. We can add ViaSniffer to probe live multicast streams and ViaMotion Probe to probe OTT streams (including Verimatrix scrambled HLS contents).