Internet telephony.

Internet telephony.

Internet telephony

Over the past few years, the world has seen a rapid increase in the number of Internet users. Its popularity is due to a wide range of services and capabilities: e-mail (E-mail), the World Wide Web (WWW), network conferences (NNTP), voice mail, Internet Telephony (IT), and others.

This article is devoted to the prospects and problems of IT development based on telephone gateways (phone gateway network). With their help, an ordinary telephone set becomes a means of access to the global telephone network built on the Internet. Such a network is more intelligent compared to the ordinary telephone network, since all the functions of telephone switches are performed by computers.

Until recently, it was possible to say that digital and analog information is transmitted via communication channels. With the development of the Internet, the line between these two types of information is beginning to blur and, perhaps, will disappear completely after some time.

To transmit various types of information, as a rule, various network technologies are required. Thus, the following data transmission protocols are used on the Internet: TCP/IP, X.25, Frame Relay, PPP, ISDN, ATM, etc. As a result, many technical, organizational and economic difficulties arise in the design of networks and their management, for example, when integrating networks built on different principles (circuit-switched and packet-switched networks; connection-oriented networks and broadcast networks). There are problems associated with the reception, processing and distribution of telephone and fax messages, voice and data multiplexing, quality and delays in voice transmission, channel capacity and tariffication of new services.

Before 1995, IT was not so well known. There were separate freely distributed programs (public domain), created by research programmers without the prospect of their wide application, since it was considered impossible to get a high-quality voice connection via the Internet. VocalTec, which released the Internet Phone program in February 1995, proved the opposite. Installed on a personal computer (486/33 MHz or higher), equipped with multimedia (sound card, microphone, speakers) and connected to the Internet (Fig. 1), the Internet Phone program allows you to talk to another PC user anywhere in the world.

Fig. 1. Configuration of a personal computer for VoIP

At the end of 1995, the total number of active IT users was already estimated at 500 thousand. (Active users constantly turn to IT, unlike those who, having the opportunity to work with IT, do not use it.) The lion's share of the IT market—94%—was occupied by VocalTec products.

Initially, IT users were people making long-distance telephone calls over the Internet or conducting small experiments in business lntranet applications. The potential future of the technology is interactive electronic commerce, intra-company communications, and international telephone communications over the Internet.

At present, IT is actively developing. According to IDC forecasts, by the end of 1999 the number of IT users of all categories should increase to 16 million. Dozens of companies have released software products that provide voice connections over the Internet in real time (the most famous of them are listed in Table 1).

Table 1. Software products for VoIP

In the offered products, the voice quality varies from comparable to a standard telephone connection (Digiphone) to a barely audible voice (Netphone). Currently, most programs provide acceptable voice quality even with a modem connection to the Internet at a speed of 28.8 kbps. The quality of communication is significantly reduced at a speed of 14.4 kbps.

The disadvantage of all the listed products is the mandatory availability of computers with voice cards, an Internet connection and the same type of software for end users. The next step in the development of IT is computer-network-telephone and telephone-network-telephone technologies with the ability to connect via the Internet using a regular telephone set.

IDT has announced the Net2 Phone system, the first commercial service that allows Internet users in the United States to make domestic phone calls. So far, in most US cities, the system only provides a computer-to-phone connection, where the caller connects via a computer connected to the Internet, and the called party answers on a regular phone (Fig. 2).

Fig. 2. Net2Phone Technology

IDT announced plans to provide Internet users with the ability to call the United States from anywhere in the world in 1998.

While this service is not yet available, it is expected that the fee will not exceed $0.10 per minute. It also plans to provide a phone-network-phone service, which does not require the caller to have a computer.

In March 1996, VocalTec and Dialogic joined forces to develop an Internet Telephony Gateway (ITG). As a result, a new software product was created — VocalTec Telephony Gateway (VTG), which supports up to 30 simultaneous voice connections. In Russia, its presentation took place at the Comtek-97 exhibition. At present, several corporate communication lines have already been installed via VTG between large Russian cities: Moscow, Novosibirsk, Samara, St. Petersburg, etc.

The communication possibilities via a network of telephone gateways are practically unlimited. Such gateways are capable of connecting local public switched telephone networks (PSTN) in different cities, private telephone networks of mini-PBX (PBX), and computer local networks (Ethernet) via the Internet. They make Internet telephony truly convenient, since a subscriber can call any phone number in another city or country (Fig. 3). A call via the Internet reaches the city and goes through the server to the city telephone lines. The terminal equipment can be either a personal computer or a regular telephone with tone dialing.

Fig. 3. VoIP network technology

The procedure for establishing a connection via the Internet for a subscriber is practically no different from a regular international call. The subscriber picks up the phone, dials the local ITG number, passes authorization, dials the remote ITG number and the number of the called subscriber. After the subscriber has called the local ITG via PSTN or PBX, passed authorization, dialed the country and city code, the ITG requests the IP address of the remote ITG by its phone number from the server (LDAP Directory Server). Having determined the IP address of the ITG closest to the called subscriber, the local ITG establishes a connection with it via the Internet. The remote ITG, in turn, calls its subscriber via PSTN or corporate PBX. When the subscriber hangs up, the ITG breaks the connection with the remote ITG and makes all the necessary records on the connection time, etc.

The following types of calls are possible using telephone gateways (Fig. 4,5):

Fig. 5. Connecting phone to phone via network

The call goes via the public switched telephone network (PSTN) or directly from the office digital mini-PBX to the nearest ITG. Then, after an automatic prompt, the caller dials the end subscriber's number on the telephone keypad. The call via the Internet goes to the ITG located closest to the subscriber. From here the call is routed via the ITG or office mini-PBX to the subscriber.

To send a fax, simply program the fax machine to automatically dial the ITG number and the required number. The fax goes to the nearest ITG via a PBX or office mini-PBX. Then, via the Internet, the fax goes to the ITG closest to the subscriber's fax, and from there it is sent to its destination.

If you call from a telephone to a computer, the call goes to the nearest ITG via a PBX or office mini-PBX. After the automatic prompt of the voice system, the subscriber's Internet address is entered using the telephone dial pad and the call is sent to the required computer.

Fig. 4. Connecting a computer to a telephone

A call from a computer via the Internet is received by the ITG closest to the subscriber and from there via a PBX or office mini-PBX is sent to the called subscriber.

In the first two cases, two telephone gateways are used, in the last two — one. The cost of the call in all cases is determined by the cost of connecting to the Internet plus possible costs for using the public telephone network.

To connect to an Internet user, you need to know the IP address or domain name of their host computer. Telephone gateways solve the addressing problem, since in this case it is enough to know only the subscriber's number. Each telephone gateway can be assigned a specific numbering plan. The table of correspondence between telephone numbers and gateway IP addresses is stored on the telephone server (LDAP Directory Server). The sequence of establishing a voice connection via the Internet is shown in Fig. 6.

Fig. 6. Sequence of establishing a VoIP connection

The operation of the telephone gateway is illustrated in Table 3. One interface (FXS, FXO, E&M, E1/T1) is connected to the telephone network, the other to the Internet (Ethernet, Frame Relay, ATM). The ITG receives the telephone signal, digitizes it (if it is analog), compresses it, breaks it into IP packets and sends it to the router. The ITG converts the packets coming from the router back into a telephone signal. This ensures a full-duplex connection. The vast majority of ITGs are managed via the SNMP protocol using any SNMP manager, such as HP Open View.

Table 3. Functional diagram of the telephone gateway

The ITG functions also include echo cancellation (Fig. 7), which occurs in telephone systems due to signal reflection, for example, when switching from a two- to a four-wire line. The echo effect is clearly evident during long-distance telephone calls, when the signal delay is large; during local calls, echo cancellation in the telephone system is not required. In IT, telephone systems are used only for local calls, so the function of echo cancellation during connection falls on the ITG.

Fig. 7. Eliminating echo in IT systems

Problems arise when transmitting DTMF (Dual tone multi-frequency) signals over the Internet. Segmentation into IP packets makes them indistinguishable at the remote end. ITG must distinguish DTMF, interrupt transmission and generate them at the remote end.

The key issue of using IT in business applications is the quality of transmitted voice and delays. The quality of voice transmission has significantly improved compared to earlier versions of software products for IT, distortions have been eliminated. This was achieved by improving the coding mechanisms, recreating lost packets.

A delay exceeding 250 ms significantly degrades the quality of the connection. The IP protocol was designed without taking into account the possibility of transmitting voice in real time. Unlike the telephone network, which is built on the principle of circuit switching, the Internet uses the principle of packet switching. Significant delays occur when IP voice packets are transmitted along different routes and the sequence of their reception is disrupted. The IP protocol does not allow predicting these delays.

International standards are of great importance for the development of IT. In 1996, the International Telecommunications Union (ITU) developed the H.323 recommendation (Fig. 8). It defines standards for the transmission of data, video and voice traffic when one or more sections of the network connection are local IP networks, as well as the T.120 teleconferencing standard. The ITU recommends using RTP/RTCP (Real-Time Protocol/Real-Time Control Protocol) to manage audio and video traffic. H.323 defines how delay-critical audio traffic receives higher priority on the Internet than other types of traffic (H.324 applies to dedicated lines, H.320 to ISDN).

Fig.8. H.323 specification

The Internet Engineering Task Force (IETF) and the largest Internet service providers propose using the Reservation Protocol (RSVP) for voice transmission over the Internet. RSVP allows bandwidth to be reserved and provides the so-called quality of service (QoS). However, it will take time for all existing routers to switch to RSVP support.

The H.323 recommendation also includes the G.729 voice compression standard approved by the ITU in November 1995. Most telephone networks in the world use adaptive differential pulse-code modulation (ADPCM) with a channel capacity of 32 kbps (defined by the G.724 standard). The ADPCM compression algorithm is based on encoding the difference between signal levels, as well as dynamic adjustment depending on the input signal level.

The new compression algorithm allows to achieve the same quality level when transmitting voice at a rate of 8 kbps as when transmitting using ADPCM at a rate of 32 kbps. The new standard is based on the CS-ACELP (Conjugate Structured-Algebraic Code Excited Linear Predictive) algorithm. CS-ACELP forms voice packets of 10 bits every 10 ms, based on 80 templates of the standard PCM-1 (64 kbps). The algorithm ensures high quality of voice transmission with minimal delays on DSP (Digital Signal Processors).

One example of G.729 implementation in practice is the ClearVoice technology from Micom. ClearVoice packs G.729 voice frames so that together with service information they occupy approximately 9 kbps of a leased dedicated line or 10.6 kbps of a Frame Relay connection. Channel bandwidth savings are also achieved by using the pause suppression method: in a conversation on one channel, pauses are filled with digitized voice or data from other channels. As studies confirm, voice traffic consists of 50-70% pauses, thus the requirements for the average channel bandwidth are reduced to 4 kbps, so the pause suppression method is very effective.

Due to its cost-effectiveness, ClearVoice technology allows for a larger number of voice channels to be organized on a connection with a lower bandwidth, ensuring high voice transmission quality. Here are just a few examples:

— MicroBand ATM technology can be used to create two high-quality voice channels on a dedicated line at a speed of 19.2 kbps, leaving 10-12 kbps for data transmission.

— Seven high-quality voice channels can be organized on a Frame Relay CIR 64 kbps channel.

— A set of 30 voice channels on an E1 line can be compressed to 256 kbps, transmitted over a dedicated line or Frame Relay connection.

The main equipment for IT is the computer expansion boards, including: voice, fax and switching boards, voice command recognition boards, text-to-speech and pulse-to-tone conversion boards. Depending on the functional requirements and the number of telephone lines, the required number of expansion boards connected via the SCbus are installed in the computer. SCbus is a bidirectional high-speed multiplexed bus with a capacity of up to 1024 full-duplex channels and the ability to switch any two channels. Several computers with Dialogic boards can be connected via SCbus.

Voice cards are the main element of IT systems. They provide connection via telephone lines and voice interaction. Voice cards are built on specialized DSP processors that perform a number of functions (digitization, compression, speech reproduction, recognition of tone dialing signals) without accessing the central processor. In addition, with their help, user commands are implemented by successive pressing of the keys of a telephone with tone dialing. The number of serviced channels is from 2 to 30.

Fax cards are responsible for receiving and sending fax messages (according to a list, on request), converting text files into document images and encoding them for sending by fax. Structurally, these cards can be made in the form of autonomous devices or attachments to voice cards. The number of channels is from 1 to 30.

Switching boards are designed for intelligent switching of voice channels of both local and external lines and organization of conferences with the number of participants from 2 to 8.

Voice command recognition boards are necessary for the perception of voice commands of the subscriber and are used in cases where it is impossible to give a command from the keyboard of the telephone set (for example, if the set does not support tone dialing).

Text-to-speech conversion boards allow you to synthesize a voice message from a text file, and are used to create a voice menu system.

Pulse-to-tone conversion boards convert pulse dialing signals into tone dialing signals, ensuring their recognition by voice boards.

The specified set of expansion cards for IT systems is an open technology based on the SCSA (Signal Computing System Architecture) developed by Dialogic. SCSA defines a standard for hardware and software, allowing the construction of flexible computer telephony systems. About 300 IT hardware manufacturers, software developers, and telecommunications equipment manufacturers have announced their support for SCSA.

One of the main advantages of IT is the ability to simultaneously transmit data, voice, fax messages, and local computer network traffic over the same communication channels. This maximizes the channel capacity and, given that this is the most expensive telecommunications resource, significant cost savings are achieved.

There are several multiplexing technologies, each with its own advantages and disadvantages (see Table 4).

Table 4. Main characteristics of various multiplexing technologies

Time Division Multiplexing (TDM) technology is based on dividing the entire bandwidth of a channel into fixed time intervals (time slots), each of which is allocated a certain time interval for operation in the network. Providing a very low degree of latency, TDM technology is most suitable for voice and video transmission. However, TDM has low efficiency in the network, since when a gap occurs in data transmission on one of the channels, no other channel can use its bandwidth. When transmitting data and voice, TDM efficiency is on average 10-25%.

Statistical Packet Multiplexing (SPM) combines X.25 principles with statistical multiplexing technology. SPM provides bandwidth only to active channels and dynamically distributes bandwidth between them.

With the help of a multiplexer, active data is converted into a packet, to which an identifier and a checksum are added, which are necessary for error correction and routing. The packet is transmitted over the network using the store-and-forward principle, which means that the packet begins to be transmitted to the next node only after it has been fully received at the current node.

SPM solves the main problems of TDM, but has its own disadvantages: it increases delays in data transmission over the network, and also complicates the prediction of this delay. The store-and-forward principle itself increases the transmission delay, and the error correction process and the changing packet size make its arrival time unpredictable.

Fast Packet Multiplexing (FPM) preserves all the advantages of both SPM (dynamic bandwidth allocation) and TDM (low traffic latency). FPM provides error detection/correction throughout the network, which significantly reduces packet delays. With FPM, only a few bytes of a packet need to be received before the packet is routed to the next network.

FPM solves the network delay problems inherent in SPM, but does not provide a prediction of the information arrival time. FPM packets vary in length from several tens to thousands of bytes. A voice packet, which is critical to time delays, will wait until the transmission of a long data packet is completed.

An example of FPM implementation is the relatively inexpensive and popular Frame Relay (FR) technology. Unlike ATM, which is created as a highly intelligent integrated network for transmitting various types of information, FR is designed for economical data exchange with packet switching.

FR technology assigns different priorities to different types of traffic, with voice traffic having the highest priority. Since the FR network priority system does not guarantee delivery time, we can only talk about reducing the probable delivery time of packets in all possible ways. It is usually assumed that between two voice packets in the queue there are no more than two data packets.

During the transmission of voice information, the so-called packet segmentation mode can be switched on, when the data is divided into very short packets, with the transmission time of each packet being from 5 to 10 ms. In this way, it is possible to reduce the intervals between voice packets.

One of the modern promising network technologies is the asynchronous transfer method (Asynchronous Transfer Mode — ATM). The use of ATM allows solving one of the main problems in packet-switched networks — the problem of variable delays in the process of transmitting information of all types.

ATM switches are connected to each other using trunks, which can simultaneously transmit data, voice, and video. ATM technology is based on the use of cells of a fixed length (53 bytes), which becomes effective on channels with a speed of over 2 Mbit/s (Table 5). At lower speeds, Micro Band ATM technology can be used.

Table 5. Standard transmission rates for ATM channels

When using ATM, a virtual connection is established between the source and the receiver before data transmission, cells are transmitted only along this path, and their sequence is not violated. In addition, ATM uses the concept of «quality of services»: the equipment responsible for the connection makes a request for certain priorities for traffic transmitted over a given virtual channel. It is possible to set four service levels: CBR (Constant Bit Rate), VBR (Variable Bit Rate), ABR (Available Bit Rate), UBR (Unspected Bit Rate). The highest priority CBR provides guaranteed speed and is used for voice transmission.

For low-speed communication channels, MicroBand ATM (M-ATM) technology has been developed, which is a combination of fast packet multiplexing and cell switching. MATM allows increasing the efficiency of channel bandwidth usage to 90-95%.

Table 6. Comparative characteristics of MicroBand ATM and ATM technologies

MicroBand ATM dynamically distributes the range, providing it only to active users. Just like ATM, M-ATM technology is 5-10 times more efficient than TDM, but does not require high-speed and expensive channels (Table 6).

Many factors influence the development of IT. Currently, the Internet bandwidth is insufficient for widespread use of IT throughout the world. IT is most likely to spread in corporate Intranet and commercial Extranet networks, where a single operator can control the network bandwidth.

Another important element of IT development is the evolution of ITG from a personal computer platform to a specialized switch capable of handling hundreds of voice connections simultaneously. Such switches combine data, voice, and video in a single communication channel. In this case, IP acts as a unified intermediate layer between the application and the underlying physical layer (leased line, Frame Relay, ATM).

A noticeable trend in IT development is the use of ATM technology for building backbone networks by the largest ISPs.

In the coming years, increasing channel speeds and improving the efficiency of their use may turn Internet telephony into a service as widespread and convenient as e-mail is today.