There are two major factors that affect the video experience: bandwidth, and delay. The components that produce delay are both endpoints as well as the network connecting them together. The accepted delay goal (ITU recommendation) is 150 milliseconds (ms) from camera/microphone to speaker/display. There is a conflict: more bandwidth, less delay; less bandwidth, more delay.
There are camera and coding delays at one end, with decoding and display delays at the other end. Video applications present not only sound and picture quality problems, but can also cause the lips and picture to be out-of-sync -- very distracting. These four delay components can have a significant influence on the end-to-end delay and the user experience.
I learned more about the camera and display delays from Yoav Nativ, Senior Manager of the Avaya Engagement evangelist team, who presented "Best Practices for Preparing Your Network for Video Solutions" at the June International Avaya User Group (IAUG) conference in Denver.
The delay in the camera can be short -- a few milliseconds. There is no processing performed on a pixel-by-pixel basis. Processing that includes multiple pixel processing increases the delay, which occurs with functions such as filtered digital zoom.
Video compression causes delay, which will depend upon the compression technique used to reduce the bandwidth requirements. The following digitization standards, from highest to lowest bit rate are:
The better the compression algorithm, the less bandwidth required.
There are two numbers associated with display capabilities. For example, a picture can be described as 1080p60. The first number is the resolution of 1,920 by 1080 pixels. The second number after the 'p' means the progressive scan rate in frames per second. Each frame of the image is created in a single pass of the picture from top to bottom. Coding delay will depend on the digital resolution. Yoav provided some example delays for the Sony EVI-HD7V.
As the frame rate decreases, the coding delay increases, but the required bandwidth decreases. As the number of pixels decreases, the delay increases.
Yoav presented several components that are part of the network delay, some which occur in the endpoints:
All of these network delays can add up to 4 ms to over 130 ms if the packets traverse multiple ISP networks.
The receiving endpoint has to accomplish four functions: decoding, scaling, output with an overlay, and the display. The overlay delay is the longest, producing delays that range from approximately 24 to 30 milliseconds. The last function, display, generates the shortest delay of 1 to 2 milliseconds.
Yoav provided some examples of the approximate one way endpoint delays for the Avaya Scopia XT 5000 Video Conferencing systems. The table below shows the total equipment delays. The delay figures do not include the network delays.
Although the network delay is a significant delay contributor, it is important to analyze the delays produced by the endpoints. The endpoint delays can consume most of the recommended 150 ms delay goal. Select your endpoints and resolutions carefully. The selection may reduce the bandwidth requirements but penalize the quality of experience for the user.
