Open Mic - New Requirements for Linear Transcoding
Multi-screen viewing is driving them.
Having eclipsed peer-to-peer traffic, video now generates more Internet traffic than any other category. More than 50 percent of consumers use the Web to watch TV on a weekly basis. Over-the-top (OTT) video traffic continues to increase as more and more consumers start to use long-form video portals such as Hulu, Netflix and HBO Go.
Our Ericsson ConsumerLab research has already confirmed that the Internet has become a necessity in life for all manner of social, practical and leisure experiences. The amount of people consuming content on-demand every week continues to increase year-over-year.
However, scheduled TV, or linear TV, still remains at the core of the viewing experience, while recorded programs, streamed TV and movie content, and downloaded material continue to attract more widespread adoption and acceptance among consumers.
Cable ops are responding to the demand from consumers as they start to change their behavior and use new devices to watch video whenever and wherever they want. In 2011, TWC's iPad app racked up 360,000 downloads in the first month.
Clearly, operators understand how important it is to provide both the video and applications to preserve their brand, as well as to eventually generate revenues from advertising. In addition, the operator has the back office hooks to provide additional navigation and control functions for legacy set-top boxes, unifying the experience for the consumer.
HTTP Adaptive Streaming has greatly improved the experience of watching video over an unmanaged network by dynamically adapting the video bit rate to provide the best quality according to the instantaneous bandwidth available to the client. The video is encoded at multiple bit rates, called profiles, and the client requests the highest bit rate that the network can support, increasing or decreasing the bit rate as network conditions vary over time. By segmenting the video into chunks that are aligned to and commence with corresponding I-frames, the client can seamlessly play the video across transitions in bit rate. The client will also limit the maximum bit rate it requests based on its processing power and screen resolution. HTTP Adaptive Streaming clients are now becoming ubiquitous across PCs, tablets and smartphones.
Software transcoding is generally used to prepare on-demand content so that it can be delivered using HTTP Adaptive Streaming. Since the content is prepared in advance, the actual transcoding of the video into different profiles (at varying bit rates) is not time-critical. However, linear programming has a completely different set of requirements, which are much better addressed by hardware transcoders. In fact, the scale of the linear transcoding challenge for HTTP Adaptive Streaming is only now being realized. It is much larger than for conventional compression systems, since each channel must be encoded at multiple profiles:
• Each class of device (e.g., iOS, Android, Windows) has different resolution and DRM requirements.
• Each class of device may require a large number of different encoding rates to support smooth transitions (as changes in network conditions trigger adaptive bit rate changes).
So each channel may ultimately require as many as 20 to 30 profiles, and hundreds of channels will require thousands of multirate transcoders.
Resolution, picture quality and compression efficiency are just as important for HTTP Adaptive Streaming as for conventional compression systems. One at a time:
• Resolutions of up to 720p60 are required for connected TVs today; 1080p60 and even higher resolutions will be soon required.
• The highest picture quality transcoding is important because it can differentiate the operator-supplied video from OTT video.
• Finally, compression efficiency is all about making the best possible use of the available bits. This is crucial given the relatively high cost of DOCSIS bandwidth.
Fortunately, advances in hardware-based transcoding solutions have continued to push the limits of highest-possible picture quality for a given bit rate, to reduce power consumption and to increase density. Hardware transcoders can provide a 10-fold increase in density over software transcoding and are software-configurable to process video of different resolutions, so the processing power can be optimally applied to a mix of screen sizes and refresh rates.
Currently, there are three main competing and proprietary HTTP Adaptive Streaming protocols in the marketplace: Microsoft Smooth Streaming, Adobe Dynamic Streaming and Apple HTTP Live Streaming. An effort has been underway to standardize a single approach called MPEG DASH (Dynamic Adaptive Streaming over HTTP) as an ISO standard (ISO/IEC 23009-1). Work is almost complete, and MPEG DASH should be finalized by early 2012.
MPEG DASH has the potential to unify the industry around a single HTTP Adaptive Streaming standard, with the promise to greatly simplify the operations of IP video service providers – reducing the streams by a factor of three. Nevertheless, operators will still need to support all of the resolutions, devices and bit rates.
Technicolor will be writing February's Open Mic column.