Cast Your Vote in the Streaming Media Readers' Choice Awards.

Me TV Part 2: The Technical Challenges of Personal Streams

Article Featured Image

In my last post, I focused on the content programming aspects of leanback programming. In this post, I'll focus on the technical components.

How Do We Do It?

From an implementation perspective, the overall approach needs to be considered in regard to both of the following:

  1. The orchestration of multiple systems to support the content-decisioning process
  2. The end-to-end process of how content is delivered and consumed


What does it take to create unique content programming for an audience of one with the ability to scale to hundreds, thousands, or hundreds of thousands of simultaneous viewers? The content delivery network (CDN) still plays a role as the core content delivery mechanism to reach scale, enabling the bits and bytes to reach playback devices -- desktop, mobile, connected TVs -- in an efficient, cost effective, and consistent manner.

The content decisioning process is new. As we think about all the factors that can make content programming unique and personal, it requires a complex set of data that extends beyond basic content scheduling and electronic program guides (EPGs). The inputs to the content-decisioning process may include any or all of the following:

EPGs/Scheduling would define generic blocks of programming, mixing live and prerecorded content, allowing for videos that must be viewed at a specific time, and ensuring content can be dynamically programmed based on any of the other following factors.

User Personalization would define what content is available based on user preferences or targeting. User preferences may include filtering or prioritizing content based on metadata (e.g., TV rating, duration, genre, air date) or preference (e.g., specifying a specific video bitrate quality or bandwidth limit to control data usage). Targeting would extend to a model that measures the engagement of specific content types (based on core characteristics or metadata) and using either individual information or aggregated contextual data (e.g., geographical trends, device trends, social network trends) to affect programming.

User Management/Entitlements would define which videos a specific user is authorized to view. Entitlements could include any number of models: gamification to earn the right to access specific levels of content, pay-per-view, subscription video-on-demand, or TV Everywhere.

Content Policy would define the content rights for a particular video, including geographic restrictions, placement (e.g., viewable only on specific web domains), device or form factor limitations, and whether or not specific entitlements must be enforced.

Ad Decisioning would define the rules for when and how advertising should be added as part of content programming. For linear content from a broadcast feed, ad decisioning would include management of SCTE-35/SCTE-104 signal-based triggers. For true live content from an event, ad decisioning would include management of manual (i.e., human) triggers. This process would require integration with a publisher’s third-party ad server or a third-party ad network to manage the actual ad request and ad response, containing information about the underlying ad creative and beaconing.


How does all of this affect the actual video consumption experience? When using a desktop or laptop with a browser, it’s likely that Flash is installed and used to manage a wide range of experiences, including video. The advantage is that Flash can be used to deliver a consistent, efficient, and customized video experience.

However, when thinking about how video content is consumed elsewhere, fragmentation is the name of the game. The wide range of mobile devices, connected TVs, and game consoles means that it’s not a straightforward process to generalize the level of interactivity the viewer may have. The publisher may also have limited information about the viewer and the context of the video-consumption experience.

The publisher needs to ensure that Me TV can respond to both a dumb client -- playback of programming with limited or no user interaction -- and a smart client -- playback of programming with extensive opportunities to interact with the user, present options, and monitor the experience.

Consequently, this typically refers to a model known as server-side content programming, as this combines both video content and advertising content. Outside of iOS, Apple’s HLS protocol has been adopted as a de facto media streaming standard by other mobile platforms (including Android to a degree), connected TVs, and game consoles. HLS can even be supported on the desktop -- natively by some browsers or even by using Flash as a client-side proxy.

From a technical perspective, HLS provides adaptive bitrate delivery of content over HTTP via segmented streaming. Not only can the segments be easily cached by CDNs, allowing for scalability to a large audience, but server-side manipulation of the m3u8 manifests -- the master playlist and variant playlists -- enable frame-accurate video manipulation on the server. For the client, content is consumed as a single video without the need to load discrete videos.

By removing the chance of buffering or pauses between content -- videos or advertisements -- Me TV mimics a broadcast experience.

The publisher is effectively creating a unique channel for every viewer. Simplified, this means that, behind the scenes, every viewer has his or her own logical set of HLS manifests that dynamically stitch together the individual (TS) segments of content.

On the client side, publishers need to handle cases where there is limited client-side capabilities for viewer interaction and beaconing. For those cases where there is capability, publishers should instrument the Me TV model for more accurate measurement, provide the viewer with options to influence programming, enhance the leanback experience, and provide integrated monetization beyond the standard in-stream creative.

Since It’s All About Me, When Will I Get It?

The discrete pieces of the technical solution are already available to publishers. However, the concept of leanback programming and Me TV -- even if desired by viewers -- raises a challenge to publishers to review, reject, retire, or rebuild their existing content and advertising strategy. This is no easy task, as Me TV fundamentally runs counter to many of the established rules that govern the broadcast and entertainment industries. While several publishers have already tested leanback programming in various forms, they haven't gotten to the point of offering a unique and addressable Me TV experience.

The DVD, iPod, and iPad were catalysts for change within established models of distributing content. Me TV has become a rallying cry for consumers as they demand more choice and freedom, but as too much choice leads to confusion, studies have repeatedly shown this presents an opportunity for publishers to maximize their core strengths -- great content and intelligent curation -- while satisfying the individual desires of the viewer, implicitly turning the me into a we.

Editor's Note: This is a vendor-written case study. StreamingMedia.com accepts vendor-written case studies based upon their usefulness to our readers. Albert Lai is CTO of broadcast and media at Brightcove.

Streaming Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues
Related Articles

Media Companies Reassess Their Identities as Expectations Change

For streaming video providers, the choice now is to either build in-house or to look to the cloud and get out of the technology business.

Me TV: Programming Content for an Audience of One

The next evolution of television will be personal television, delivering individual streams optimized for the viewer, device, time of day, and location. Learn what to expect from Me TV.

Companies and Suppliers Mentioned