CMAF- The Future of OTT Streaming

The OTT industry has evolved from the legacy RTP, RTMP and Flash streaming to more advanced streaming protocols like HLS, DASH, Smooth Streaming, and HDS with support for adaptive bitrate streaming over the past couple of years. HLS and DASH emerged as the most popular choice among OTT service providers and have been used widely. 

Challenges with Current Streaming Technologies

Even though the OTT world has seen many technological improvements, video streaming to various devices running different operating systems mandates the use of different streaming protocols has always posed a challenge to the service providers. While they need to figure out the optimal streaming resolutions & respective bitrates to support content streaming on said devices the content has to be packaged differently to deliver HLS and DASH streams to respective devices. This introduces a lot of commercial and operational overhead on OTT service providers owing to additional processing, storage, and distribution.

Sports streaming is an integral part of the entire OTT fraternity which is driving mass user engagement to the tune of 25 million concurrent users in a single event and while the traditional broadcasting via satellite has about 5 sec of pipeline delay, the same is not true for OTT streaming. The standard delay of live OTT streaming is more than 10-15 seconds in practice. The need for low-latency streaming (~5-sec stream delay from origin till end-user) for such sports events was being felt strongly among service providers and the end-users.

All of these paved ways for the development of a common streaming format which can be used independently on any device as well as it should be capable of offering low-latency OTT streaming.

Introduction to CMAF

Common Media Application Format abbreviated as CMAF is jointly developed by Apple and Microsoft. This idea of developing a common and less-complex streaming format was tabled to Moving Pictures Expert Group (MPEG) in Feb 2016. The companies progressed very fast where Apple added support for fragmented MP4 in HLS within 4 months followed by submission of standard in July 2017 which finally got released in January 2018.

1. CMAF Format: CMAF specified the use of fragmented MP4 (fMP4) or ISO Base Media File Format (ISO BMFF) which is also used by DASH. It can be represented by HLS and DASH manifests and hence avoids the need to create separate packaging for different streaming formats saving cost and operational overhead.
2. CMAF Specifications

Container format: ISO BMFF (fMP4)

Codecs-Video: AVC, HEVC, VP9 (later)

Codecs-Audio: AAC-LC, HE-AAC & HE-AACv2

Encryption Algorithm: Common Encryption (CENC)

Encryption Modes: CBC and CTR

Subtitles: WebVTT & IMSC-1

Closed Captions: CC608 & CC 708 (via SEI)

Image showing the CMAF layer starting from protocol level down to codec.

Ref: Pallycon

CMAF Packetization. Ref. Wowoza

CMAF & DRM

Content deliveries where DRM is not being used, HLS prevailed as a single content format that can be played by all devices and there is no need to maintain a second content processing pipeline, however, for DRM protected content this became a challenge.

Initially, since the DRMs had support for either Counter (CTR) or Cipher Block Chaining (CBC) encryption modes depending on the target platforms which necessitated the need of using different encryption methods and hence, required a separate packaging with supported encryption.

CMAF specifies the use of CENC for the encryption which can support CTR and CBC modes, thus streamlining the use of a single media processing pipeline. Widevine has started supporting CBC while PlayReady has added support for CBC from v4.0+. This makes CBC as the common cipher algorithm which can be used by all DRMs.

CMAF: Benefits

1. Reduced operating cost: Single content packaging using ISO BMFF container eliminates the need for producing multiple copies of the same content to support various streaming formats. Thus cost related to content re-packaging/re-transcoding and maintaining content copies in CDN cache are nullified.
2. Reduced complexity of operation: Use of a common container format and CENC for encryption requires a single media pipeline to be maintained and thus reduces the operating complexity.
3. Reduced latency: Use of CMAF container eliminated the muxing and other typical overhead carried by TS packets while at the same time delivery of presentable data chunks reduced the decoding time making way for low-latency playback.

Conclusion

CMAF has simplified the OTT streaming workflow enabling low-latency streaming and is certainly the future of OTT streaming. Being a new technology, it would need fast adoption of its standards from the entire OTT ecosystem starting from packagers to client devices.

In the present scenario, there are many legacy devices (like Android v6.0 or below) and browsers which do not support CMAF (with CBC) as of now, and hence, a wise decision has to be made balancing all the parts while adopting CMAF.

References:

• https://blogs.akamai.com/2016/06/cmaf-what-it-is-and-why-it-may-change-your-ott-future.html
• https://blog.irdeto.com/2019/02/05/cmaf-may-be-ready-but-are-you-ready-for-cmaf/
• https://www.adobe.com/experience-cloud/articles/premium-video-content-protection-strategies-with-cmaf.html
• https://www.wowza.com/blog/what-is-cmaf
• https://www.theoplayer.com/blog/how-cmaf-will-influence-the-online-streaming-industry
• https://www.unified-streaming.com/blog/promises-cmaf-and-its-compatibility-unified-streaming-platform
• https://mpeg.chiariglione.org/sites/default/files/files/standards/parts/docs/N18093_ISOFF%28TS%29.pptx
• https://castlabs.com/resources/faq/drm/
• https://www.nexplayersdk.com/widevine-player-sdk/
• https://pallycon.com/blog/cmaf-the-quest-for-single-format-streaming/