VoIP Basic Concept (2): Principle
VoIP principle and technology is a very complex system engineering through the Internet, which is very wide, so the technologies involved in the technology are very large, and the most fundamental technology is VoIP (Voice Over IP) technology, which can be said, the Internet Voice communication is a most typical and most promising application areas of VoIP technology. Therefore, before discussing voice communication with the Internet, it is necessary to first analyze the basic principles of VoIP, and related technical issues in VoIP. I. The basic transmission of VoIP is traditional telephone network to transmit voice in a circuit switching, and the required transmission broadband is 64kbit / s. The so-called VoIP is a series of special treatments that compress the analog speech signals in the IP packet switched network, which can be transmitted without a connected UDP protocol. In order to transmit speech signals on an IP network, there are several elements and functions. The simplest form of network consists of two or more devices with VoIP functions, which is connected through an IP network. The basic configuration diagram of the VoIP model is shown in Figure 2-18. From the figure, you can find how the VoIP device converts the voice signal to an IP data stream and forwards the data stream to the IP destination, and the IP destination converts them back to the voice signal. The network of the two must support IP transmission, and can be any combination of IP routers and network links. Therefore, the transport process of VoIP can be simply divided into the following stages.
Figure 2-18 VoIP Model Structure 1, Voice - Data Conversion Speech Signal is analog waveform, transmit voice via IP mode, whether it is real-time application service or non-real-time application service, the appearance of the voice signal first to simulate data conversion That is to quantify the analog speech signals to quantify, and then fed into the buffer storage area, the size of the buffer can be selected according to the requirements of delay and encoding. Many low bit rate encoders are coded in units of frames. Typical frame length is 10 ~ 30ms. Considering the cost during the transmission, the speech package is usually composed of 60, 120 or 240 ms speech data. Digital can be implemented using a variety of speech coding schemes. The currently used voice coding standards mainly have ITU-T G.711. The voice encoder of the source and destination must implement the same algorithm, so that the speech device can restore the analog voice signal. 2, the original data to the IP conversion once the voice signal is digitally encoded, the next step is to compress the voice package with a specific frame length. Most of the encoders have a specific frame length. If an encoder uses a 15MS frame, divide the package from the first 60 ms into 4 frames, and encoded in order. 120 speech points per frame (8 kHz). After encoding, 4 compressed frames are synthesized into a compressed voice package to send to the network processor. The network processor is added to the other endpoint after adding the header, timeclant, and other information to the voice. The voice network simply establishes the physical connection between the communication endpoint (a line) and transmits the encoded signal between the endpoint. The IP network is not like a circuit switched network. It does not form a connection. It requires the data to be placed in a growing datagonal or group, then give each data report to address and control information, and send it through the network, one stop one Stand forward to the destination. 3. Transfer in this channel, all networks are viewed as a voice package from the input, and then transmit it to the network output from the time (t). T can be varied in a full range, reflecting jitter in network transmission. The same node in the network checks the addressing information attached to each IP data, and uses this information to forward the datagna to the next stop on the destination path. The network link can be any development structure or access method that supports IP data streams. 4, IP package - data conversion destination VoIP device receives this IP data and start processing. The network level provides a variable length buffer to adjust the jitter generated by the network. The buffer can hold many speech bags, and the user can select the size of the buffer. Small buffers produce a smaller delay, but cannot adjust large jitter. Second, the decoder will compress the encoded speech to generate a new voice package, which can also be operated by frames, completely and the length of the decoder. If the frame length is 15 ms, the 60ms of the voice package is divided into 4 frames, and then they are decoded to be reduced to 60ms of the voice data stream to send the decoded buffer. During the processing of the datagram, the addressing and control information is removed, and the original original data is retained, and then the original data is supplied to the decoder. 5. Digital speech is converted to an analog voice play drive to remove the voice spots (480) in the buffer to the sound card, and broadcast through the speaker (for example, 8kHz). Briefly, the transfer of speech signals on the IP network should pass through the conversion from analog signals to digital signals, and the digital voice encapsulation into IP packets. The IP packet is transmitted through the network, and the IP packet is unpacking and digital speech to analog signals. The process is equal. The whole process is shown in Figure 2-19.
Figure 2-19 Basic Process of VoIP Transmission Second, driving the driving force of VoIP development due to many of the development and technical breakthroughs in the relevant hardware, software, agreements, and standards, making VoIP's broad use soon become reality. The technological advances and development in these areas plays a role in the role of a more effective, functional, and more interoperable VoIP network. Table 2-2 briefly lists the main developments in these areas. As can be seen from the table, the technical factors that drive VoIP rapid development and even widespread applications can be summarized as the following aspects. 1. Digital Signal Processor, DSP) performs voice and data integration requirements. The DSP processing digital signal is mainly used to perform complex calculations, otherwise these calculations may have to be executed by the General CPU. Their specialized processing capabilities and low-cost combinations make DSPs well suitable for performing signal processing functions in the VoIP system. The calculation overhead of G.729 voice compression on a single voice stream is generally large, requiring 20MIPS, if a central CPU is required while processing multiple voice streams, it is not realistic, so it is unrealistic, so Use one or more DSPs to uninstall the computational tasks of the complex voice compression algorithm in the central CPU. In addition, DSP is also suitable for voice activity detection and echo cancellation of such a function, which handles voice data streams in real time, and quickly access the memory on the board, therefore. In this chapter, it is more detailed to describe how to implement voice coding and echo cancellation in the TMS320C6201DSP platform. Table 2-2 Progress Progress Agreement and Standard Software H.323 Weighted Fair Queuing Method DSP MPLS Marks Exchange Weighted Early Detection Advanced ASIC RTP CAR Cisco Quick Forward CPU Processing Power G.729, G.729A: CS-ACELP Extension Access Table ADSL, RADSL, SDSL FRF.11 / FRF.12 token Bucket Algorithm
MultiLink PPP Frame Relay Data Standard SIP Based on Priority COS Packet Over SONET IP and ATM QoS / COS Integration Protocol and Standard Software H.323 Weighted Fair Queuing Method DSP MPLS Tag Exchange Weighted Random Early Detection Advanced ASIC RTP, RTCP Dual funnel universal cell rate algorithm DWDM RSVP Rated Access Speed SONET DIFFSERV, Car Cisco Quick Forward CPU Processing Power G.729, G.729A: CS-ACELP Extended Access Table ADSL, RADSL, SDSL FRF.11 / FRF.12 Bucket Algoric MultiLink PPP Frame Relay Data Standfinder SIP Based on Priority COS Packet Over SONET IP and ATM QoS / COS Integration 2, Advanced Specific Integrated Circait, ASICs have developed Fast, more complex, and more functional ASICs. ASIC is a single application or a small set of functional application chips. Since concentrated in a very narrow application goal, they can optimize specific functions, usually double universal CPU fast one or several orders. Just like a streamlined command set computer (RSIC) chip focuses on fast performing throwing, ASIC is pre-programmed to make it possible to perform a limited number of functions faster. Once the development is completed, the cost of ASIC bulk production is not high, which is used to include network devices such as routers and switches, perform routing tables, packet forwarding, packet categories, and check, and other functions. The use of ASICs makes the performance of the device, and the cost is lower. They provide additional broadband and better QoS support for the network, so it has a great promotion of VoIP development. 3. Most of the IP transmission holding transmission telecommunications network uses time division multiplexing methods. The Internet must adopt statistical reuse becoming a plurality of packet switched modes, the latter is high, and the interconnection is simple. Effective, it is very applicable to the data service, which is one of the important reasons for the development of the Internet. However, broadband IP network communication makes a momentary requirement for QoS and delay characteristics, so the technological development of statistical multiplexing growth group exchange is concerned. At present, in addition to the new generation of IP protocols that have been introduced, the World Internet Engineering Task Force (IETF) proposes multi-protocol mark exchange technology (MPLS), which is a variety of tag / labels based on network layer selection. Exchange, can improve the flexibility of the circuit, extend the network layer selection ability, simplify the router and the integration based on cell exchange, and improve network performance. MPLS can be used as a standalone selection protocol, but also compatible with existing network routing protocols, supporting various operations, management and maintenance features of IP networks, making IP network communication QoS, routing, signaling and other performance Improve, meet or close the level of statistical multiplexing packet exchange (ATM), while it is simpler, efficient, cheap and applicable than ATM. IETF also seizes new grouping institutions to implement QoS selection. Among them, the "Tunnel Technology" is to achieve broadband transmission of one-way links. In addition, how to choose an IP network transmission platform is an important area of research in recent years. It has emerged in IP over ATM, IP Over SDH, IP OVER DWDM and other technologies. The currently recognized broadband network analysis model is shown in Figure 2-20.
Figure 2-20 The first layer of the layered model of the broadband IP network is a grassroots, providing high-speed data transmission backbone. The IP layer provides high quality, IP access services that have a certain service guarantee to IP users. The user layer provides access form (IP access and broadband access) and service content form. In the basic layer, Ethernet as the physical layer of IP networks is a matter of course, but IP overdwdm has the latest technology and has great development potential. Dense Wave Division Multiplexing, DWDM is injecting new vitality for fiber optic networks and provides amazing bandwidth in the new fiber main network in Telecom. DWDM technology utilizes fiber optics and advanced optical transmission equipment. The name of the wave division multiplexed is obtained from a single-stranded optical fiber (Laser). The current system can send and identify 16 wavelengths, while future systems can support 40-96 full wavelength. This is of great significance because every time a wavelength increases a stream. Therefore, 2.6Gbit / s (OC-48) network can be expanded 16 times without having to lay a new fiber. Most new fiber networks run the OC-192 at a speed of (9.6 gbit / s), which generates 150 Gbit / s capacity on a pair of fibers when combined with DWDM. In addition, DWDM provides the protocols and speed of the interface, which can support the transmission of ATM, SDH, and Gigabit Ethernet signals on one fiber, which is compatible with the various networks that have been built, so DWDM can be Protecting existing suppliers, can also provide a more functional backbone network with its huge bandwidth for ISP and telecommunications, making the broadband costs and more accessibility, which provides powerful requirements for the bandwidth requirements of VoIP solutions support. The increased transmission rate can not only provide a more coarse pipe, which makes the occlusion opportunity, but also reduces the delay, so it can largely reduce QoS requirements on the IP network. 4. User Access to Broadband Access Technology IP Network has become a bottleneck that restricts the development of the whole network. From a long-term development, the ultimate goal of user access is to fiber to the household (ftth). Optical access networks include both optical digital loop carrier systems and passive optical networks. The former is mainly in the United States, combined with open port V5.1 / v5.2, and transmits its integrated system on the fiber, which shows a lot of vitality. The latter is primarily targeted and Germany. Japan persisted in unremitting research for more than ten years, take a series of measures to reduce passive light network costs to a level similar to copper and metal twisted pairs and a lot. In particular, recent ITU proposes ATM-based passive optical network (APON), complementing ATM and passive optical network advantages, access rate up to 622M bit / s, is very advantageous for broadband IP multimedia business, and can reduce The number of fault rates and nodes, expand coverage. At present, ITU has completed standardization, and various manufacturers are actively developing, soon there will soon be available, which will become the main development direction of broadband access technology for the 21st century. The mainly used access technologies are: PSTN, IADN, ADSL, CM, DDN, X.25 and Ethernet, and broadband wireless access system columns. These access technologies have characteristics, the fastest development is ADSL and CM; CM (Cable Modem) adopts coaxial cable, high transmission rate, strong anti-interference ability; but cannot transmit two-way transmission, no uniform standard. ADSL (ASYMMERAL DIGITAL LOOP) exclusive access broadband, fully contributing to an existing telephone network, providing a non-symmetrical transmission rate, the download rate of the user side can reach 8 mbit / s, and the upload rate of the user can reach 1m bit / s. .
ADSL provides the necessary broadband for companies and users, and greatly reduces costs. Using a lower cost ADSL area loop, the company can now access the Internet and Internet service providers with VPNs at a higher speed, allowing higher VoIP calls capacity. 5. Central processing unit technology central processing unit (CPU) continues to develop in terms of function, power and speed. This allows multimedia PCs to be widely used and improves the performance of system functions restricted by CPU power. The power of PC processing streaming audio and video data is expected in the user, so transmitting speech calls on the data network will serve as the next goal. This computing function allows advanced multimedia desktop applications to support voice applications in advanced features in network components.