Jitter and Latency ----------------------- Latency can be defined as the time between a node sending a message and receipt of the message by another node. The TANDBERG systems can handle any value of latency, however, the higher the latency, the longer the delay in video and audio. This may lead to conferences with undesirable delays causing participants to interrupt and speak over each other.
Jitter can be defined as the difference in latency. Where constant latency simply produces delays in audio and video, jitter can have a more adverse effect. Jitter can cause packets to arrive out of order or at the wrong times. TANDBERG systems can manage packets with jitter up to 100ms; packets not received within this timeframe will be considered lost packets. If excessive packet loss is detected, the TANDBERG systems will make use of IPLRTF (see document D50165, TANDBERG and IPLR, for more information) or downspeeding (flow control) to counteract the packet loss.
The C-series endpoint utilizes a dynamic jitter buffer that can increase in value depending on the performance of the network. To minimize introduced latency, this buffer will begin at 20ms and continue to 100ms if sufficient packet loss or jitter is occurring. Additionally, the C-series endpoints supports RTP time stamping into the audio packets in order to help reduce lip sync issues that may occur over an H.323 call.
To further improve lip sync with high resolution images (including XGA, w720p and other high resolution video formats), the C-series endpoints support dynamic buffering of video packets in an attempt to place information on the wire as fast as possible. Without this functionality, the endpoint would attempt to maintain a consistent packet size when placing the information on the wire, which would result in video being buffered internally to ensure that the entire packet could be filled prior to transmission. This potential buffering created a potential lip sync issue at the far end of the H.323 call as the time between the actual capture of the visual image and placing the information on the wire was not a constant and, therefore, the far end system cannot adjust for any time differences between the arrival of the video information and the arrival of the audio information.
The endpoint will now, by default, not buffer the high resolution images prior to transmission, which will ensure a constant time delta between the arrival of the video and audio information to the far end, allowing for an adjustment as necessary and improved lip sync. This change in behavior, though, can cause the endpoint to send out consecutive packets that have a relatively large difference in size. For example, one packet can come out at 1400 bytes while the packet behind that can be sent out at 800 bytes followed by a 1200 byte packet and so on. Some QoS configurations improperly handle the large adjustments in packet size, thereby dropping packets within the QoS buffer and causing packet loss in the call. If, for any reason, it is necessary to reduce the impact of this behavior, it can be done through the API command ‘xConfiguration Network [1..1] TrafficControl Mode: <On/Off>’. When this setting is ‘On,’ the endpoint will buffer the traffic in order to provide more consistent packet sizing; ‘Off’ (default) will not buffer any video internally.
I hope this answers your questions.
Regards
John Regan Product Specialist (Endpoints) CDO Escalations Cisco
Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community: