Some sites:
Http://www.vchelp.net/cndevforum/subject_list.aspforum_id=28&order=view_num
http://www.voip99.com/old/index_ku_00021.htm
http://www.vchelp.net/cndevforum/subject_view.asp?subject_id=64672&Forum_ID=28
http://www.1000bbs.com/dispbbs.asp?boardid=83&id=35534&page=1
http://www.1000bbs.com/dispbbs.asp?boardid=83&id=119137&page=9
http://www.cer.net/Article/20040726/3111127.shtml
http://www.vchelp.net/cndevforum/subject_view.asp?susubject_id=42847&Forum_ID=28
http://www.liumiti.com/art/product/2002122150940192.htm
1 Introduction
With the continuous growth of Internet bandwidth, video applications based on stream media transmission have developed rapidly. The related technologies transmitted on the Internet also become the hotspots for the Internet section research and development. Currently, many experimental high-speed broadband networks use the technical and application of video transmission as a key topic of research. The Digital Video Initiative in the Internet 2 (I2) research plan is an important part of the I2 application development is established in the INTERNET 2 (I2) research program. The I2 application development group believes that digital video is the most widely used interest and the most widely used I2 capabilities, which can override the application from point to remote resource control. China's 3TNET research program also clearly proposes streaming media related technologies in high-speed networks to realize video applications such as HDTV.
There are many difficulties in transmitting video on the Internet. The fundamental reason is that the Internet's unconnected per package forwarding mechanism is primarily a burst data transmission design, which is not suitable for transmission of continuous media streams. In order to be effective, high quality transmission video streams on the Internet, support for multiple technologies. This article mainly introduces some key technologies in Internet video transmission, including video compression, encoding technology, application layer quality control technology, continuous media distribution service technology, media synchronization technology, and digital copyright management technology.
2, video compression, encoding technology
The usual video compression can be divided into compression of scarable and unscalable. Unlatable compression refers to a bit-stream, and scalable compression generates multiple sub-bit flows, one of which is a base stream, which can decode independently, output rough quality. Video sequence. Other subtragmifers work from quality enhancements. All sub-bit flows together to restore the best quality video sequence. Correspondingly, only partial sub-bit streams (must contain basic sub-bits), the output of the video is in the image of the image, or the size or the frame rate is poor. These three aspects constitute a compressed encoding in a mass signal-to-noise ratio (SNR), space, and time. Specific scalable compression selects some or few of them to achieve scalability. In order to provide better flexibility, a different delay requirement is met, a new compression mechanism called fine-grained scalability (FGS) is submitted to MPEG-4. FGS compression The video is compressed into two bit streams: the basic layer of bitstream and enhanced layer flow. Unlike SNR scales, FGS is encoded by BitPlane. It enables continuous reinforcing layer rate control. A FGS encoded variant is a PFGS (Progressive FGS) encoding, in addition to the basic advantages of FGS, PFGS can achieve more than two bit streams, with better prediction and error recovery capabilities. A variety of encoded formats have been developed in the actual system, mainly including RM, WMA, ASF, DIVX, etc. Windows Mediaz's latest Windows Media ENCODING UTILITY V8.0 Beta includes Windows Media Audio 8 (audio) and Windows Media Video 8 (video) two-part compression coding systems. The MPEG-4 video compression technology adopted by Windows Media Video 8 achieves a screen (a resolution 640 × 480, 24 frames per second) close to DVD Quality at 500 kbps transfer rate. Windows Media Video 8 is optimized for movie online downloads, support TRUE-VBR (true dynamic variable rate coding). Real Networks's latest Real Producer 8.5 uses its own REAL G2 CODEC to make RM files, which have many advanced design, such as SVT (Scalable Video Technology, Two-Encoding). Bidirectional coding is similar to VBR, which can be optimized based on the bandwidth limit by pre-scanning the entire movie.
3, application layer quality control technology
The main purpose of application layer quality control technology is to ensure the improvement of quality in the video transmission process. It mainly includes congestion control and error control. There are many reasons for quality control, one main reason is that the current Internet only provides Best-Effort service, without quality assurance. Therefore, quality control is required to achieve the mechanism of the application layer. The purpose of congestion control is to avoid decline due to the loss caused by network congestion. For video streams, the main method of congestion control is rate control. The purpose of the rate control is based on the predicted network bandwidth determines the rate of transmission. There are two basic rate control means, based on the rate control of the transmitting end and the rate control based on the receiving end. The former is mainly based on feedback information. Can be applied to the way of unicast or multicast. For unicasting, there are two rate control methods, based on probes-based methods, and model-based methods (Model-based). The method based on the detection allows the loss rate of the package to a fixed probability value by constantly adjusting the rate. Model-based method directly utilizes TCP throughput model calculation rate. Therefore, the model method is also referred to as a TCP friendly rate control method. For single-channel multicast, only probing-based methods can be employed. In the rate control based on the recipient, the sender does not participate in the rate control, and the acceptor is adjusted by increasing and reducing the number of channels. Typically, this method is used to adopt a wide range of multicast, and the multicast video stream is divided into a multi-layer, and each layer corresponds to a channel of the multicast tree. The reception side has a different quality of the channel to obtain different channels. The rate control based on the recipient can also be divided into a method based on the method and a model-based method. In addition to the rate control of the sender and the reception-based rate control method, there is a method called a mixed rate control, and the characteristics of the two are characterized, that is, the accepted part increases reduces the channel, and the sender is also adjusted according to feedback. The rate of each channel. An example of a mixed rate control method is a method of target set packet. When the rate of transmission is determined based on the rate control of the sender, a necessary operation is to adjust the rate of the original compressed stream to meet the requirements of the target rate, which is achieved by a rate shaping. A typical rate adjustment method is implemented based on the encoded scalability. Includes: a) Frame discard filter, which can distinguish between different frames, such as MPEG encoded I frames, B frames, and P frames. Discard the frame (first B frame, last I frame) according to the importance of the frame. b) Sublayer discard filtration; C) frequency filtering.
The purpose of congestion control is to reduce the loss of the package, but it is not possible to avoid the loss of the package. In this case, a certain error control mechanism may be required. The error control mechanism includes: a) FEC, FEC's purpose is to restore correct information by other packets after the package is lost by increasing redundant information. b) retransmission of delay constraints. The usual flow of the broadcast has time limit, so the retransmission is valuable when the retransmission time is less than the normal play time. c) Error-resilient encoding: In the encoding, by appropriate control makes it possible to reduce the impact on the quality after the loss of data is lost. In the Internet environment, the most typical method is to describe encoding (MDC). MDC compresses the original video sequence into a multi-bit stream, and each stream corresponds to a description, it can provide acceptable visual quality. Multiple descriptions are combined to provide better quality. The advantage of this method is to achieve robustness and enhanced quality of data loss. Its disadvantage is that it describes the encoding (SDC) compared to compression efficiency. Moreover, a certain amount of correlation information must be added between multi-description, which further reduces the efficiency of the compression. D) Cancealment: The wrong cancellation means that the receiving end is weakened to the human visual impact by a certain approach after an error has occurred. The main method is the interpolation of time and space. Research in recent years also includes maximum smooth recovery, motion compensation time forecasting, etc. In the current commercial streaming media system, Real Networks uses a large number of application layer quality control technology to improve the quality of streaming. In RealVideo, the Damage-Resistant encoding is used to reduce the impact of the package loss. When the rate is small, FEC technology is used. Realvideo supports two coding: Realvideo Standard and Realvideo Fractal. The former can support the coding rate from 10K to 500K, and is especially Optimized for 28.8k and 56K. In order to adapt to the change of the network rate, Real Networks uses Sure Stream technology that avoids the shortcomings of the rate usage mechanism cannot be optimized for each rate. The specific method is to first store media streams encoded at different speeds in a file, followed by dynamic streams of that rate by monitoring bandwidth between servers and customers. The main disadvantage of this technology is that the continuous control of the rate is not adapted.
4, continuous media distribution service
Traditional video transfer is based on a single streaming server. However, a single server is not expandable in performance, and the quality of the transmission is limited by the main network transmission quality. In order to improve the performance of the transmission, there are some techniques to improve the quality of streaming. The purpose of the Continous Media Distribution Services is to provide service quality and efficient media transfer based on the Best-Effort basis. Typical methods include:
Network filter, the basic method of network filtering is to insert a network filter on a network device between the transmitting end and the receiving end, which realizes the rate of the rate based on the congestion status of the network. Unlike the speed of the transmitting end, the network filtering is performed on the path of streaming, so the rate shaping can be performed in the congestion point, thus improving the efficiency of the rate control and the response time of the congestion control. And you can place more than a plurality of filtering devices in the network. Network filtering usually uses a Frame-Dropping Filter.
Application-Level Multicast: Based on multicast technology for IP layers, despite years of research, it has difficulty implementation. The main problems include support for scalability, network management, implementation and high-level applications. The objective of the application layer is to build multicast services on the Internet. It allows you to build an independent media multicast network. Media Multicast Networks can be built by interconnecting of content distribution networks, and is built by peer-to-peer relationships between different types of service providers. In the media multicast network, each node with multicast capabilities (called Media Bridge: Mediabridge) Performs the route of the application layer. In addition, each media bridge and one or more neighbor media bridges have established an application layer overlapping topology through a clear configuration interconnection. The Media Bridge performs a distributed application layer multicast routing algorithm in the media multicast network to determine an optimized virtual multicast path. The advantage of applying layer multicast is to break some obstacles to IP multicast. Multicast control can be implemented in a more flexible manner. Content Pass Network (CDN): CDN can improve the performance and service quality of network transmission with the increase in total traffic, and can maintain the user's latest content in a rapidly changing environment. The content transfer network can also support flexible distributed applications. Due to these features, the content transfer network can be used to pass streaming content. A content delivery network mainly includes the following parts: Caching and content delivery, content distribution and management, request routing, accounting, and billing. Caching Server Storage The Objects Request to provide users with faster and easier delivery. Caching and content transfer are the main features of the CDN. A total of three contents of the Caching Server: Forward Proxy, transparent forward, reverse proxy. The distribution and management of content will distribute the content from the source server to the CACHE of the CDN, that is, the content can be transmitted or pushed to the CDN node by a certain policy. Content distribution strategies the key role of optimization of CDN. Request Routing Technologies include content redirection over the global, and load balancing of Cache servers in the same CDN network. There are three typical technologies to implement request routes, including: using dynamic DNS global request routes, global request routing of the transport layer, global request routing in the application layer.
Currently, REAL NETWORKS and Microsoft provide Caching / Proxy based solutions, such as Real Networks RealSystem Proxy. RealSystem Proxy maintains the integrity of Caching content through a back channel back channel (Back Channel).
5, media synchronous control
In some streaming applications, video streams and other forms of media streams need to be integrated together in a synchronous manner. For example, in remote learning, the slide description of the learning content needs to be synchronized with the audio or video stream of teachers. There are usually three types of synchronous controls: intra-stream synchronization, inter-stream synchronization, inter-Object) synchronization. The core of the media synchronization mechanism is to indicate the time relationship between the media or between the media. The method of explaining the time relationship comprises an interval based method, a method based on the axis, a method based on a method of controlling and a time-based method. Commonly description of the explanation of continuous media is an explanation or timestamp of the shaft. For the terminal system, the synchronization mechanism includes the preventive mechanism and the Corrective mechanism. The former is mainly to reduce synchronous errors by reducing delay and jitter, and the latter is mainly resumed after the synchronization error occurs. Taking into account the latency randomness of the Internet transmission, synchronous errors are inevitable. Therefore, the error compensation of the recipient is necessary. A corrective mechanism is a stream synchronization protocol (Stream Synchronization Protocol), which uses the concept of an expected delay to adjust the expression time between different media streams to recover the network delay. Synchronous Multimedia Integration Language (SMIL: SYNCHRONIZED MULTIMEDIA Integration Language) is a multimedia manipulation language specified by 3W (WORLD WIDE Web Consortium). Multiple streams and text messages can be implemented in the time synchronization control and spatial location arrangement when playback. A certain user interaction function can also be achieved by SMIL.
6, Digital Rights Management (DRM)
DRM is a way to protect multimedia content from unauthorized playback and replication. It protects their private music or other data for content providers to provide means for illegal replication and use. DRM technology protects digital content by encrypting and adding rules for digital content, where using rules can determine if the user meets the conditions of playing digital content. Using rules typically prevent content from being replicated or restricted. The operating system and multimedia middleware are responsible for enforcing these rules.
Currently, Microsoft provides a complete DRM solution, including the implementation of the DRM architecture and SDK to achieve secondary development. Software encryption in Microsoft's DRM scheme, speed can reach 10m / s, using RC4, DES, 3DES symmetric encryption algorithm encrypted digital content, use asymmetric algorithm RSA, ECC to authenticate and encrypt content encryption key.
7. Conclude
At present, streaming applications and services based on Internet video transmission have developed rapidly in China. Various applications, including video conferencing, video on demand, internet, remote teaching, etc. are gradually promoted. The techniques of transmitting videos in the Internet have also received more and more applications and is constantly developing. It is expected that broadband streaming media applications based on video transmission will become one of the mainstream applications of future Internet.
