ParallelStream (H.264/AVC)

The fastest growing segment of Internet Applications is Streaming Video. Already, digital video accounts for more than 60% of all data transferred over Internet. Popular consumer video sources like Netflix, Hulu, iTunes, YouTube and many more are expanding rapidly, and will overtake Broadcast by 2020 according to The Diffusion Group (DTG). AVC Adaptive Streaming is the accepted way to deliver streaming video in best possible quality and reliability. VSS ParallelStream™ is a patent pending technique of VSS AVC Encoder, specifically developed to address the challenges of AVC Adaptive Streaming.

AVC Adaptive Streaming Systems

The challenge in delivery of video to consumers over the internet is the constantly changing bandwidth of the client's connection. To address this shortcoming without changing the already accepted AVC video format, video serving has migrated to Adaptive Streaming architecture. Such systems continually monitor client player conditions and dynamically adjust video content being served. Each such system is proprietary and with matching set of server and player components. For example: Microsoft Silverlight, Move Networks, Adobe Flash, Apple HTTP (iTunes) and others. Typically each such system consists of three major components, Encoding/Adaptation of video content; delivery of ready content to the appropriate player and the internet player itself. Here forth these will be called Adaptive Stream Encoder, Streaming Server and Streaming Player. VSS ParallelStream™ addresses the first component requirement, the Adaptive Stream Encoder.

Adaptive Streaming Encoder Requirements

To enable Adaptive Streaming, the source video must be transcoded into a number of lower resolution, lower bitrate video clips. Each of these videos are used during streaming process when the bandwidth of client connection is reduced. All of the generated video streams must be encoded with aligned GOP structure so that Adaptive Streaming Server can switch from one video stream to another on-the-fly. When this happens, client's player seamlessly transitions from one video quality to another. The objective of this system is to make a smooth transition between different streams so that viewer does not notice it. With more video streams generated from original video, the transitions become more smooth and less noticeable.

The challenge of Adaptive Streaming Encoding is in efficiency of implementation. The traditional system will use as many individual AVC Encoders as required by number of streams. Thus, to create 8 streams, the operator must run 8 simultaneous encoders on a single PC system. Processing of such application requires a very powerful and therefore expansive PC system.

VSS has developed unique algorithm, ParallelStream™, which reduces the CPU requirement for Adaptive Encoding in 4 times. The largest stream is fully encoded as previously, but all smaller resolutions require 4x less CPU cycles than before. With ParallelStream™ provider needs to increase encoder CPU spec only in 1.25 times to achieve the real-time encoding for Adaptive Streaming.

CPU saving for transcoding case is even higher. With ParallelStream™ CPU requirement is one decode (1/4 of 640x480 encode) plus 1/4 encode, i.e. total is only 0.5x CPU, or 2.25 times less than before.

VSS ParallelStream™ solution brings substantial benefit to users, as it allows to achieve dramatic cost saving in hardware equipment as well as in power consumption. It enables the Adaptive Streaming at a fraction of the cost, thus allowing content producers to move to the most compelling technology without a doubt. More detailed information is available in VSS ParallelStream™ white paper.