Transcript
QualiVision™ RADVISION’s improved packet loss video compensation technology
A RADVISION White Paper March 2005
Table of Contents 1. Introduction ....................................................................................................................3 Videoconferencing—the technology of the 21st century............................................ 3 Video quality vs. network bandwidth ......................................................................... 3 QualiVision — RADVISION’s packet loss handling technology ..................................... 3
2. How videoconferencing works .......................................................................................4 Videoconferencing components.................................................................................. 4 Video transmission: an overview ................................................................................ 4 Video transmission: compression standards ............................................................... 5 Compression standards problems ............................................................................... 6
3. The QualiVision Solution ................................................................................................7 QualiVision Mode......................................................................................................... 7
4. Summary..........................................................................................................................9
1. Introduction Videoconferencing—the technology of the 21st century Videoconferencing has emerged in recent years as one of the more exciting technologies of the Internet age, bringing with it an impressive set of diverse applications and benefits. The advent of IP-based systems enabled unprecedented media integration features along with reduced operating costs. This included the ability to integrate streaming video, multipoint conference management, remote Internet management and voice over IP, all into one system.
Video quality vs. network bandwidth The holy grail of videoconferencing is to provide video quality as close to life-like as possible but ever-growing traffic demands placed on enterprise networks create ongoing tension between available bandwidth and the prodigious amounts of data needed for video transmission. Meanwhile the main players in the videoconferencing industry have spent years developing video compression algorithms, adopting new transmission protocols and putting an impressive array of standards in place. Video quality is contingent on these three major parameters 1. Compression Video codecs are used to compress the large amount of data necessary to provide smooth audio and video. The most common video compression standards published by the ITU (International Telecom Union) and used in today’s videoconferencing are. H.261, H.263 and H.264. 2. Bandwidth In today's enterprise, network bandwidth has become a scarce commodity. Virtually all organizations today face bandwidth constraints due to the large amount of e-commerce, inter-office multimedia, data, voice and videoconferencing traffic that is squeezing itself onto the enterprise’s intranet and extranet. 3. Packet loss The Achilles’ heel of videoconferencing — packet loss — is a naturally occurring phenomenon. A poorly implemented IP network can prevent a percentage of IP packets (each packet containing video frame data) from reaching their final destination. How videoconferencing systems handle and compensate for packet loss is at the heart of what differentiates the quality of video between one videoconferencing system and another.
QualiVision — RADVISION’s packet loss handling technology QualiVision is state-of-the-art technology that handles and manages packet loss. This technology is capable of producing quality video even under conditions of disruptive packet loss. QualiVision is an enhanced encoding implementation that adheres completely too existing video standards such as H.261, H.263 and H.264… This ensures full interoperability between ITUcompliant endpoints made by different manufacturers.
QualiVision White Paper
3
2. How videoconferencing works Videoconferencing components MCUs (Multipoint Conferencing Units) are devices that enable multipoint video conferencing. The MCU mixes the video and voice data sent from each participating endpoint (see Figure 2-1) and then transmits the mixed data back to the endpoint device. The MCU acts as a receiver and a transmitter of video and voice data.
Figure 2-1: MCU with various types of participating endpoints
Video transmission: an overview Video data is transmitted from one endpoint to another via IP packets. One video frame is made up of pixels that are divided into 16x16 pixel blocks (macroblocks), coded, packaged into IP packets, transmitted, reassembled and decoded at the other end. Endpoint
Endpoint
Coding
Decoding IP Packet
n
IP Packet 2
IP Packet 1
Figure 2-2: Video transmission over IP
QualiVision White Paper
4
Video transmission: compression standards The most common video compression algorithms used in today’s videoconferencing environment are H.261, H.263 and H.264. These standards use a compression technique known as interframe compression. Interframe compression is applied to a sequence of video frames rather than a single image. In general, relatively little change occurs from one video frame to the next. Interframe compression exploits the similarities between successive frames, reducing the volume of data required to reproduce that sequence of frames.
I = Intraframe (full frame transmission) P = Interframe (partial frame transmission)
Figure 2-3: Interframe compression
How it works Initially an intraframe (full frame) is sent followed by a sequence of interframes (partial frame data). An intraframe does not require any information from previous frames. Interframes contain only information pertaining to changes that occurred, relative to the previously sent frame. Due to the incremental nature of this encoding, should an interframe get dropped due to packet loss, the missing data will cause video distortion; propagated in all succeeding video frames, till the next intraframe (full frame) arrives.
The fast-update (or intraframe) request In an IP videoconferencing session, media packets may be dropped and not reach their final destination. Video control protocol within H.323 and SIP contain a command called video fastupdate, which is used by the receiver to request a full frame update (intraframe) from the transmitter. The sequence of how this works is demonstrated in the following figures.
Figure 2-4: Receiver requesting a fast-update from transmitter
1) The receiver detects packet loss 2) The receiver sends a request for fast update from the transmitter QualiVision White Paper
5
Figure 2-5: Transmitter replies with an intraframe in response to a fast-update request
3) The transmitter sends an intraframe (full frame)
Compression standards problems A network sustaining a high rate of packet loss will result in excessive fast-update requests from the receiving endpoint, i.e., intraframes (full frames). In this case the benefits of frame sequence compression are lost. Furthermore the system will experience the following degradations:
Video quality (blurry images)
Jerky and uneven motion (lower frame rate)
Flicker effect
Finally, in the case of network congestion, this fast update mechanism results in poor bandwidth utilization. A vicious cycle will occur due to the fact that rising packet loss will create an increase in fast-update requests, resulting in more data getting pumped into an already congested network.
QualiVision White Paper
6
3. The QualiVision Solution QualiVision technology provides a solution that improves video quality in situations of disruptive packet loss and addresses some of the shortcomings of standard video compression algorithms in handling packet loss. QualiVision provides a fully ITU-compliant technological solution, ensuring that a wide variety of ITU-compliant devices made by various endpoint manufacturers will benefit from QualiVision’s technology solution.
QualiVision Mode The RADVISION MCU is designed to constantly analyze receiving and transmitting channels to each connecting endpoint. When the MCU detects packet loss occurring on a link to an endpoint, it goes into what’s known as QualiVision Mode. The RADVISION MCU independently analyzes each of the receiving and transmitting channels and then applies a separate corrective compensation mechanism to each of these channels. Receiving Channel MCU
Endpoint Transmitting Channel
Figure 3-1: Receiving and transmitting channels between MCU and endpoint
Two principal things happen when the MCU goes into QualiVision mode: 1. Receiving channel: Downspeeding — when the RADVISION MCU detects packet loss on the receiving channel (i.e., endpoint to MCU), it issues a flow control command, to the connecting endpoint to gradually reduce the rate of packets being sent in an attempt to reduce packet loss rates. 2. Transmitting channel: QualiVision optimized encoding — when the RADVISION MCU detects packet loss on the transmitting channel (i.e., MCU to endpoint), instead of sending an intraframe (full frame), it applies an intelligent encoding algorithm, which transmits a focused spray of additional packets to those areas in the frame where they are most needed.
Receiving channel: Downspeeding When the RADVISION MCU detects packet loss on the receiving channel (i.e., endpoint to MCU), it issues a flow control command to the connected endpoint. This command causes the endpoint to gradually reduce packet rate transmission. This process will continue gradually until there is no more packet loss or a minimum rate has been reached. If the minimal rate has been reached and excessive packet loss is still detected, the MCU instructs a return to the original transmission rate. The assumption is based on the fact that packet loss in this instance is not attributed to network congestion and there is some other problem with the link.
QualiVision White Paper
7
Transmitting channel: QualiVision optimized encoding When the RADVISION MCU detects packet loss on the transmitting channel (i.e., MCU to endpoint), it identifies the areas in the frame that are most affected by packet loss. Then instead of replying to a fast-update with an intraframe (Figure 2-5), it applies an intelligent encoding algorithm, whereby it transmits a focused spray (or a light “drizzle”) of intra macro blocks to those areas in the frame where they are most needed (i.e., where images are moving). In this manner the MCU is able to prevent the disruptive effects of packet loss by focusing a flow of data to those areas of the frame that are most subjected to the visible effects of packet loss.
Figure 3-2: QualiVision optimized encoding — transmits an intelligent spray of packets to those areas in the frame most impacted by packet loss
The end result is that instead of entire frames getting transmitted over and over again, an effective spray of focused data is sufficiently transmitted in order for the receiving endpoint to recompile the video image with improved quality, despite sustaining a relatively high percentage of packet loss.
QualiVision White Paper
8
4. Summary Benefits and Drawbacks Benefits
Generates improved video quality in non-perfect network environments — including networks with packet loss rates of up to 7%
Compliant with ITU standard-based packet loss compensation algorithm
Fully tested to assure improved video quality with most endpoints on the market. Contact RADVISION for a full and updated list of tested endpoint models.
Results in improved video on the transmission and reception streams.
Avoids creating a vicious cycle, where greater amounts of data get pumped onto a congested network.
Drawbacks Some endpoints, when encountering packet loss, will freeze the video image until an intraframe (full frame) is received. In this case QualiVision optimized encoding will not work. This video image freeze behavior is considered poor implementation by the endpoint. The vast majority of endpoints on the market do not implement this freeze behavior; those that do are quickly moving away from this implementation.
About RADVISION RADVISION (Nasdaq: RVSN) is the industry’s leading provider of products and technologies for videoconferencing, video telephony, voice over IP, and collaborative communications solutions. RADVISION offers the broadest and most complete set of videoconferencing networking systems and next generation protocol toolkits and platforms on the market today, enabling enterprises, equipment vendors, and service providers to develop and deploy new converged networks, services and technologies. Today, hundreds of thousands of end-users around the world communicate over a wide variety of networks using products and solutions built around RADVISION’s rich media communications platforms and/or software development solutions. These include RADVISION’s award-winning videoconferencing infrastructure solutions such as its highly scalable IP/ISDN interworked gateways, feature-rich conferencing bridges, and advanced gatekeeper applications. RADVISION’s enabling technologies for OEM systems include developer toolkits for SIP, MEGACO/H.248, MGCP, and H.323, 3G-324M wireless multimedia delivery, and the ProLab™ Test Management Suite. For more information please visit our website at www.radvision.com or contact:
USA/Americas T +1 (201) 689-6300 F +1 (201) 444-7173
[email protected]
APAC T +852 (2) 8014-070 F +852 (2) 8014-071
[email protected]
EMEA T +44 (208) 757 8817 F +44 (208) 757 8818
[email protected]
] Product specifications subject to change without notice. This document is not part of a contract or license as may be expressly agreed. RADVISION is a registered trademark of RADVISION, Ltd. QualiVision and Delivering the Visual Experience are trademarks of RADVISION, Ltd. All trademarks recognized. All rights reserved. © 2005 RADVISION, Ltd.
QualiVision White Paper
9
