Multiscreen delivery

A unified media server platform provides a cost-effective approach.

For broadcasters and other content providers, the dropping price of bandwidth combined with the proliferation of video-capable devices — PCs, smartphones, set-top boxes, gaming consoles and tablets — ought to be entirely positive. Indeed, consumers seem to see it that way. According to the Cisco Visual Networking Index, the number of devices connected to IP networks will be twice as high as the global population in 2015. In other words, there will be two networked devices per capita, up from one networked device per capita in 2010.

Driven in part by the increase in devices and the capabilities of those devices, IP traffic per capita will reach 11GB per capita in 2015, up from 3GB per capita in 2010. Meanwhile, AccuStream Research continues to forecast unabated demand for online video with a double-digit year-to-year growth forecast through 2012.

The fast adoption of Android-based phones in the United States is just one illustration of the speed at which the marketplace balance can change. The first Android was introduced in October 2008. By April 2011, the operating system had captured a 36-percent share of the U.S. smartphone market. At the same time, Apple's iOS dropped to just 26 percent, while BlackBerry slid from 33 percent share in November 2010 to 23 percent, according to ComScore. With the introduction of Windows 7, the balance is expected to shift again.


Need more indicators of the fast pace of both growth and change? The global shipments of tablets are expected to rise at least tenfold over the next four years, with estimates in excess of 240 million units by 2015, according to IHS. Here the “balance of power” is also shifting. According to Strategy Analytics, by the fourth quarter of 2010, Android tablets represented 22 percent of global shipments, while the iPad had dropped 20 points to 75 percent of the market. Gartner has forecast that the Android market share will top out at 39 percent by 2015.

According to IHS, the fragmentation of the tablet market will continue as the new generation of tablets, including Motorola's high-end Xoom and RIM's BlackBerry PlayBook, debut and new models based on Windows appear. In addition, tablets will continue to expand their penetration into what were previously traditional computer markets.

More viewers with more screens watching more content more of the time is great news. For broadcasters, however, the changing and fragmented video streaming landscape poses technical challenges because of the wide variety of media platforms, transport protocols, device screen resolutions and decoding abilities. For content owners striving to ensure that their streaming infrastructure can support whatever new gadget with whatever new feature captures the consumer fancy, this can mean an endless round of running just to stay in place. Content owners that want to take maximum advantage of the trends can encounter high costs both for infrastructure and operations, and considerable risk. What if you guess wrong on which platforms and devices will rule the future? Will you be left in the silicon dust?

The streaming challenge

Mobile delivery is just part of the problem. Streaming is increasingly a delivery method of choice to TVs as well. A new generation of set-top boxes, including the popular Roku media player, is shifting away from the traditional RTSP and MPEG-TS protocols to web-centric alternatives such as Apple's HTTP Live Streaming (HLS), Microsoft's Smooth Streaming and/or other variants of HTTP streaming.

Encoding (or transcoding) for such a large variety of destinations adds another layer of complexity. In the United States, for example, 3G-connected mobile phones may require resolutions of 240p at 20fps, or 360p at 30fps, and average bit rates from 125kb/s to 250kb/s. While desktop computers may be able to consume 1080p at 30fps video, the bit rates (and quality) may also be constrained by DSL connections that average 400kb/s to 1200kb/s. A lower bit-rate stream that looks fine on an iPhone probably will not be acceptably sharp on a television equipped with a set-top box. All these possibilities may also require multiple encoding renditions of the same content per media platform, adding a potentially significant cost in live encoding equipment and storage for VOD assets.

To maintain the best possible viewing experience across multiple devices and network connections, approaches like adaptive streaming are becoming increasingly important. Here there are also numerous platform-specific variants — Flash dynamic streaming, HLS, smooth streaming and the newly minted Dynamic Adaptive Streaming over HTTP (DASH) standard. A system that accomplishes any-screen streaming must support all of the above.

Fragmented streaming

For the content provider that wants to take advantage of any-screen streaming opportunities, the traditional approach has been implementation of discrete media server infrastructures, each targeted at delivery to specific media platforms. This so-called segregated workflow streaming has a number of limitations, chief among them complexity and high cost. In fact, the approach requires buying, implementing and maintaining multiple separate technical infrastructures — for instance, Flash media servers to support Flash player-equipped devices, Microsoft IIS for Silverlight and so on. (See Figure 1)

An additional drawback of segregated workflow is inflexibility, which makes planning for the future a nightmare. Guess wrong now, and the content provider may have to replace everything much sooner than necessary. And even if the provider guesses right, an inflexible system will fail to keep pace with inevitable market shifts. In any event, the adoption rate of individual media platforms and devices is sure to change, making it difficult to predict the future.

Unified streaming

Media server software forms a crucial link in the streaming media delivery chain. The right media server software can provide a unified infrastructure for multiscreen video delivery, streaming content simultaneously to the viewer's chosen device over any connection. A single multiprotocol platform thus eliminates complexity while also saving money. Unlike the fragmented approach — which requires separate sets of encodes for each media platform — a unified server can ingest a single set of live streams or on-demand H.264 video files and stream it over multiple protocols simultaneously. This eliminates the need for specialized, client-specific encoders, and reduces VOD storage requirements because all content can be stored in one common file format.

Typically, a unified platform offers support for the most relevant live transport protocols, such as RTSP/RTP, MPEG-TS, RTMP on the ingress, and the same in addition to various flavors of HTTP streaming on the egress. (See Figure 2.) The architecture of the unified platform should be flexible enough to allow migration to new protocols as they evolve, and it should not be constrained by potential changes in encoder preferences.

Innovations in a unified server platform also enable users to enhance it with value-added features. These may bring incremental revenues, reduce cost or further simplify delivery workflows. For example, the software platform may integrate a transcoding function capable of transforming high-bit-rate live broadcast streams into multiple H.264 stream sets suitable for multiscreen adaptive bit-rate delivery. Adding network personal video recorder (PVR) functionality enables time-shifted viewing with live pause, rewind and resume features, as well as catch-up TV services. Broadcasters that source premium studio content may value the ability to implement digital rights management (DRM) protection that spans all served screens.

An important consideration in setting up the delivery system is server capacity. It typically is constrained by the server hardware and network bandwidth. To avoid a “success disaster” scenario, the unified media software platform should scale to fit capacity requirements as small as a few hundred simultaneous subscribers up to much larger audiences. (See Figure 3) Software that runs on standard, off-the-shelf hardware enables broadcasters to scale up to 10Gb/s of streaming capacity per server. For large-scale streaming, multiple servers must provide the clustering ability to deliver live or on-demand streams. If it includes dynamic caching capability, unified media server software can transfer content on demand to points closest to the viewer, enabling operators to centralize content storage while also improving the performance of edge servers.

The unified server's APIs must also enable integration with backend systems. Of course, a broadcaster wishing to implement streaming media does not have to go it alone. Depending on size and aspirations, a broadcaster may choose to employ a content delivery network (CDN) or a streaming service provider. In assessing the suitability of a service provider, the flexibility, scalability, reliability and diversity of media platforms supported by its infrastructure should be considered. For example, does the CDN offer the ability to deliver to multiple devices? How fast can it adapt to changing market conditions? What does the broadcaster need to do to prepare content for the CDN? For consumers, the availability of Web-based media on any screen at any time is all good. For broadcasters and other content providers, the Web's evolution presents both new opportunities for business expansion and new challenges, particularly keeping costs and revenues in line now and for the future. Implementation of a flexible unified media server platform is one strategy for supporting cost-effective streaming services that reliably deliver the goods to a fragmented marketplace today while keeping up with changes tomorrow.

Alex Dobrushin is chief marketing officer at Wowza Media Systems.