VoIP Network Components

VoIP network components are discussed here in more detail. Special focus is on the standards that VoIP components use to provide specific features and functions. These standards are often listed in the product sheets for these VoIP components.

VoIP Gatekeeper/Router/PBX

The central VoIP network component is the VoIP PBX. It may be called a gatekeeper, a router, or a PBX. We refer to it here as a VoIP PBX. Whatever the name, it performs the core inter-networking functions. These inter-networking functions are implemented in hardware and software that combines voice and data transmission over an IP network. The VoIP PBX supports analog telephones, emulates a telephony PBX by implementing signaling software functions, and connects the VoIP networking functions to the PSTN.

The VoIP PBX replaces other PBXs, uses the Internet or a private IP network to replace TIE trunks between facilities, and sometimes interfaces to cellular networks.

A VoIP PBX is typically administered using a Web browser like Internet Explorer or Netscape Navigator. This means that it must support the Hypertext Transfer Protocol (HTTP) to communicate with the Web browser software.

A VoIP PBX would implement hunt and call groups. A hunt group is a group or set of telephones that ring when a specific number is dialed. They ring in a designated sequence from the first phone in the hunt group to the last phone in the hunt group. Hunt groups are used to implement call centers.

In contrast, call groups are a set of telephones that all ring simultaneously when a specific number is dialed. The first phone answering the call services the call. Our phones use a call group so that anyone calling causes several phones to ring simultaneously. Anyone answering the phone first then answers the call. An alternative to call groups is automated attendant answering, a process in which the robot voice gives a menu listing and then finally asks for the extension number of the person whom you are calling.

Call detail reporting is used by call centers to track work and telephone activity. It is also used to allocate telephone costs to the different enterprise activities using the telephone system.

Unified voice mail/e-mail messaging is implemented using Internet Message Access Protocol version 4 (IMAP4). IMAP4 is a client/server protocol that works with e-mail received and held on an Internet or VoIP server. An e-mail client views just the heading and the sender information for the combined voice/e-mail message. The user then decides whether to download and listen to the voice mail or just read the e-mail message. Folders or mailboxes can be created and manipulated, voice mail and e-mail messages can be deleted, and specific information can be searched for in messages on the VoIP PBX server using IMAP4. IMAP4 requires constant access to the VoIP PBX server while the unified messaging is being worked upon.

Unified messaging also uses Microsoft's Telephony Application Program Interface (TAPI). TAPI is a standard program interface for PCs sending voice or video to other PCs or phone-connected resources. PCs equipped with TAPI can perform unified messaging tasks such as the following:

1.	Initiating a call by clicking on an icon or other image
2.	Setting up and attending conference calls
3.	Viewing other callers individually or in a conference call
4.	Adding a voice message to e-mail or listening to a voice message attached to e-mail received
5.	Automatically receiving phone calls from specified numbers
6.	Sending and receiving facsimile messages

TAPI permits PCs and other VoIP components to work with different telephone systems, including the PSTN, the Integrated Services Digital Network (ISDN), and PBXs, without having to understand their details because phone system hardware makers provide specific software drivers that interface directly with their hardware. TAPI/WAV (TAPI/Wave Form) is an audio file used to store and exchange voice information with VoIP components. TAPI/WAV files carry good quality audio that is not compressed. TAPI 2.1 adds client/server functionality to Windows TAPI. With TAPI 2.1 telephony hardware can be installed on a telephony server and then be accessed from any computer on a SOHO network. TAPI 2.1 includes administration software that controls access to remote resources.

VoIP PBXs use Internet Group Management Protocol (IGMP) to multicast phone number dialing information to other VoIP PBXs. In this manner they coordinate signaling information between VoIP PBXs.

VoIP PBX systems must resolve several differences between traditional circuit-switched telephony and VoIP telephony. These differences include latency, or delay, in the voice stream, jitter in the packet arrivals, lost packets, and echo compensation.

Latency is most noticeable in speaking to listening transitions. The latency, or delay, is the time it takes the voice signal to travel through the VoIP network connection. Such latency includes an accumulation delay from the voice codec, a processing delay in packaging digital speech into IP packets, and a network delay. Round-trip delays may be as high as several hundred milliseconds.

Latency makes satellite IP network services much less usable for implementing VoIP. When there is no other choice, satellite networks do work. But acceptable voice call latency (network delays) is a maximum of 250 ms one way. The recommended one way delay is 150 ms. These meet ITU (International Telecommunications Union) recommendations.

Typically a round trip delay over any satellite link like the Starband satellite links runs around 600 ms for geosynchronous or geostationary orbit satellites and much, much less for low earth orbit (LEO) satellites. The latency, or delay, is simply computed. The distance from an earth-bound antenna to the satellite is 22,300 miles. A single round-trip transmission must travel up and down from the satellite twice for a total of 89,200 miles. It takes light and radio waves traveling at 186,000 miles per second 480 ms (89,200/186,000) to make a round trip. Add a 100 ms electronic processing delay and the total is 580 ms or 600 ms (close enough for engineering work) round trip.

At a quarter of a second (250 ms) callers start to become uncomfortable with telephone voice communication because they can both speak simultaneously but not have a means to gracefully recover because they have been speaking too long. When the one-way delay is longer than 250 ms, callers must make a very conscious effort not to step on one another's speech. They almost must say “Over to you” when they complete speaking.

A second significant area of concern in VoIP telephony is jitter management. One approach to jitter management is using adaptive jitter management that measures jitter over time and adjusts a buffer size to match the jitter. This approach works well when jitter is consistent like jitter from Asynchronous Transfer Mode (ATM) or cell relay networks. An alternative approach uses an allowable late packet ratio to determine the jitter buffer size. This approach counts late packets and determines the ratio to good packets, and it then sets a buffer to match a specified allowable late packet to good packet ratio. This approach works well with highly variable times between packet arrivals such as is commonly found in IP networks. Some devices use a combined or hybrid approach.

The third significant area of concern for VoIP telephony is lost packets. Lost packets translate into dropped speech. Two approaches are used to compensate for lost packets' interpolation and redundancy. Interpolation guesses the lost packet speech using the information in received packets. This works well when there are infrequently lost frames. It is not very effective when large numbers of packets or bursts of packets are lost. Redundancy sends added data from which the lost packet speech can be reconstructed. The drawback here is that this uses extra transmission capacity or bandwidth. Some VoIP PBXs use a combination of approaches.

The final significant area is voice echoes interfering with speech. Voice echoes are caused by long round-trip delays for VoIP speech. In telephony, echoes are masked by side tones. With round-trip delays in excess of 50 milliseconds the side tones are prevented from masking the echoes. Echo compensation is described by G.165 Echo Cancellers (normal) and G.168 Digital Network Echo Cancellers (more stringent) ITU specifications.

VoIP PBXs connect to loop-start analog, digital, and VoIP telephones. The VoIP telephones conform to the H.323 ITU specification. The H.323 specification describes multimedia communications among terminals, network equipment, and services. H.323 belongs to the H.3x group of ITU recommendations for multimedia interoperability. H.323 originally provided consistency in audio, video, and data packet transmissions for IP networks, but it did not guarantee quality of service (QoS). H.323 is now the standard for interoperability in audio, video, and data transmission as well as Internet phone and VoIP because H.323 specifies call control and management for point-to-point connections and multipoint conference connections. H.323 also specifies gateway administration for traffic, bandwidth, and user participation. The most recent version is ITU-T H.323-V3.

Calls are set up using the Media Gateway Control Protocol (MGCP). This is also called H.248 and Megaco. The MGCP-H.248-Megaco standard is a protocol for signaling and session management needed for multimedia conferencing. The MGCP-H.248-Megaco defines communication between a media gateway that converts data from a circuit-switched network format to packet-switched network and media gateway controller formats. MGCP-H.248-Megaco is used to establish, maintain, and terminate calls among multiple endpoints. Megaco and H.248 refer to enhanced versions of MGCP. MGCP-H.248-Megaco provides a single standard for controlling multimedia IP transmission gateway devices and connecting calls from a VoIP LAN to the PSTN.

The Session Initiation Protocol (SIP) initiates an interactive connection or session for video, voice, chat, gaming, and virtual reality communication. SIP establishes multimedia sessions or VoIP calls and terminates them. Both H.323 and SIP use Real-Time Protocol (RTP) for VoIP communications.

Real-Time Protocol or Real-time Transport Protocol (RTP) manages real-time transmission of multimedia communications. RTP was designed to support video conferences. RTP is now commonly used in VoIP applications. RTP does not guarantee real-time delivery of multimedia communications. Instead it provides the ability to manage multimedia transmission as it arrives at its destination to provide the best transmission quality over an IP network. RTP combines data transport with a Real-Time Control Protocol (RTCP), making possible data delivery monitoring. Such monitoring permits a receiver to detect packet losses and to compensate for latency and jitter. RTP headers tell receivers how to reconstruct the multimedia transmission and describe how codec-produced bit streams are placed into IP packets.

Trivial File Transfer Protocol (TFTP) provides a simple file transfer mechanism. VoIP components use TFTP to upgrade firmware and transfer other files containing control information.

VoIP PBXs work with different codecs used to compress VoIP communications for low bit rate (LBR) transmission. The codecs employ G.711 Pulse Code Modulation (PCM), producing a 64 Kbps digital data stream for a single voice call; G.726 Adaptive Pulse Code Modulation (ADPCM), producing 16, 24, 32, or 40 Kbps digital streams; G.728 Low Delay-Code Excited Linear Prediction (LD-CELP), producing 16 Kbps; and G.729 Conjugate-Structure Algebraic Code Excited Linear Prediction (CS-ACELP), producing an eight-Kbps digital data stream for a single voice call. The bottom line here is the resulting transmission speed required to carry a single voice call. The lower the speed the more calls that can be carried on a VoIP communications link. The trade-off is that greater compression may impact voice quality.

When an analog signal is converted to a digital signal by a codec, the signal is divided into a discrete number of smaller parts. This is called quantization. Quantization produces the output signal, which is an approximation of the input signal. Quantization distortion is perceived as noise. The magnitude of the quantization noise problem depends on the number of quantization steps used in encoding. The effect of quantization noise is more severe with small amplitude signals with a small signal-to-noise ratio. This problem is significantly reduced when logarithmic quantizing is used. This is called companding, for compressing and expanding. Companding is a means to represent a large dynamic range of speech sound using fewer bits. Two main quantization laws are used in telephony:

Both approaches to companding map a signal onto a logarithmic curve approximated by straight-line segments. Mu-Law (µ-Law) encoding is the ITU G.711 standard. In Europe, some systems still use A-Law. VoIP PBXs should work with both types of companding to provide effective international telephony.

Telephones connecting into VoIP PBXs use different signaling techniques to dial telephone calls. These techniques include Dual Tone Multi Frequency (DTMF) and Multi Frequency (MF, MF R1/R2) analog signaling, other Channel Associated Signaling (CAS), ISDN's Common Channel Signaling (CCS), and Network Call Signaling (NCS) used in cable modem networks. Signaling conforms to the E.164 ITU international public telecommunication numbering plan.

VoIP calls need a QoS guarantee when they are transported by an IP network with best effort delivery. Such guarantees are provided by IP-ToS, IP-DiffServ, and 802.1p/Q transport mechanisms. IP-ToS, IP-DiffServ, and 802.1p/Q are all approaches that assign greater priority to VoIP packets. Greater priority means that VoIPs are processed more quickly in IP network routers and other IP network components. In this manner a transmission speed or quality of service level is guaranteed for VoIP communications through an IP network.

VoIP components can use Digital Speech Interpolation (DSI) to increase the transmission efficiency by insertion of additional signals from other sources in the transmission stream during periods of silence or inactivity. DSI fills in gaps during speech pauses to increase transmission efficiency because silence comprises about 60 percent of voice communications. DSI transmits packets from one source only during voice bursts. Another active source uses the bandwidth during silent periods. Voice Activation Detection (VAD) prevents sending packets when no voice activity is present. Comfort Noise Generation (CNG) removes discomforting periods of dead air at the receiver caused by VAD. Comfort Noise Generation matches the natural background noise and uses it to fill in discomforting dead air periods.

VoIP PBXs work with Routing Information Protocol to perform routing in a SOHO LAN. RIP is based on a set of algorithms that use distance vectors to mathematically compare routes. The route with the lowest distance vector becomes the best path to a specific destination address.

Simple Network Management Protocol (SNMP) is used by VoIP components to perform component configuration and network management functions. SNMP works with specially designed management software.

Remote Authentication Dial-In Service (RADIUS) is a client/server protocol and software that remote access servers use to communicate with a central server, to authenticate dial-in users, and to authorize access to a requested system or service. RADIUS supports maintaining user profiles in a central database that all remote servers share. In this manner RADIUS provides better security by supporting the setup of a security policy applied at a single administered network database. A central RADIUS service makes it easier to track usage for cost allocation and to keep network operational statistics.

Data sheets for VoIP PBXs, gatekeepers, and routers may contain references to the features and specifications identified here. In some cases standards compliance may not be listed. VoIP components that comply or use Internet and other telephony standards tend to work better than components not conforming to those standards. This, however, does not guarantee interoperability with VoIP network components made by other vendors. To get the most out of VoIP components the best strategy is to use all components from a single vendor.

Figure 8.6 shows an Inter-Tel Axxess PBX with a VoIP card installed. The PBX uses two interconnected chassis, with the lower chassis being opened to expose the VoIP Ethernet card. This card connects to the SOHO LAN and to VoIP telephones. The Inter-Tel Axxess PBX also connects to other digital and analog phones as well as a local PSTN trunk line into a Class 5 central office. Other VoIP cards are installed in the top chassis as well.

PSTN-WAN Connections

The VoIP PBX uses standard interfaces to connect to the PSTN. These interfaces include high-speed T-1 (United States), E-1 (international), and ISDN-PRI interfaces. Such interfaces are common to all PBXs. The T-1 interface connects using a V.35 connector, fiber connectors, or an RJ-48C connector. RJ-48C means the T-1 or E-1 digital telephony wires and pin-outs physically terminate in an eight-pin modular plug (MOD-8 or RJ-45 connector). The ISDN Basic Rate Interface S/T is ISDN electrical signaling terminated in an RJ-45 connector.

Lower-speed interfaces include TR08, TR303, Loop Start Analog, and E&M. TR303 or TR08 interfaces are for Digital Loop Carrier (DLC) connections to the PSTN for voice traffic. A Loop Start Analog connection is the same as a single two-wire analog telephone line. Similarly, E&M and tie-line connections are analog connections between switching equipment. The term E&M denotes the telephony tip and ring wires. E&M is sometimes referred to as ear and mouth, indicating the receive function and the transmit function of the leads. E&M may have originated from rEceive and transMit. E&M and tie-line interfaces are four-wire interfaces as opposed to the loop start two-wire analog interface.

To complete PSTN calls a VoIP PBX must interface with SS7. Signaling System 7 performs out-of-band (common channel signaling) signaling to support the call-establishment, billing, routing, and information-exchange functions of the PSTN. It identifies functions to be performed by a signaling-system network and a protocol to perform them.

VoIP PBXs use QSIG (ISDN-PRI-QSIG) to provide signaling for Private Integrated Services Network Exchange (PINX) devices; for example, one corporate VoIP SOHO networked facility to another corporate VoIP SOHO networked facility. QSIG is a global signaling system that links corporate telecommunications networks. QSIG is based on the ISDN Q.931 standard. Using QSIG PRI signaling, a VoIP PBX can route incoming voice calls from a PINX across a WAN to a peer VoIP PBX. This VoIP PBX then transports the signaling and voice packets to its PINX.

A VoIP PBX may also support high-speed interfaces into the Internet, other private IP networks, or directly to other VoIP PBX systems. These connections can run over Frame Relay or ATM networks. Voice running over a Frame Relay network connection uses Q.922 data link protocols and Q.933 Voice over Frame Relay Signaling specifications. The Logical Management Interface (LMI) manages frame relay permanent virtual circuits. Voice IP packet encapsulation into Frame Relay frames is specified by RFC 1490 (RFC 2427) encapsulation. ATM networks use ATM Adaptation Layer 2 (AAL2) or ATM IP encapsulation to support VoIP communications.

LAN Connections

VoIP PBX SOHO LAN connections include 10 Base T connections, 10 Base T-MDI (Medium Dependent Interface) ports for direct connection to 10 Base T hubs, 100 Base T connections, and 100 Base TX connections. Some VoIP PBXs support only 10-Mbps Ethernet, while others support both 10-Mbps and 100-Mbps variations.

Some VoIP components interface to cable modems. They use the Data Over Cable Service Interface Specification (DOCSIS). The newest variation is DOCSIS 1.1, which adds QoS and security features for constant bit rate VoIP telephony. DOCSIS 1.2 is on the horizon. It is aimed at providing higher-speed service for future applications. The DOCSIS specifications are also known as ITU J.112 Data-over-Cable Service Interface Specifications.

Other components are needed to implement a VoIP SOHO LAN, as discussed below.

Multimedia PCs

Multimedia PCs become a VoIP network component with the addition of a soft phone application program. The soft phone program uses a microphone and speakers attached to the PC to turn it into a VoIP telephone. When installed on a laptop PC the soft phone application creates a mobile VoIP telephone that uses a dialup or a high-speed Internet connection to become a remote extension of the central facility VoIP system. Soft phone application software is unique for each VoIP PBX vendor, so that the soft phone-enabled computer fully integrates with its vendor's VoIP PBX features.

VoIP Telephones

VoIP telephones look and operate like a normal digital PBX telephone. This means that, similar to VoIP soft phones, they are designed to integrate fully with the specific features and functions of the VoIP PBX. Inter-Tel Axxess digital Executive phones have the same (or equivalent) control buttons that a regular digital Inter-Tel Axxess VoIP Executive phone has. Consequently, the VoIP telephones integrate with the features and functions provided by the VoIP PBX.

VoIP telephones are commonly configured using a Web browser. The configuration would set up how the IP address was assigned, the codec used (translating into the required VoIP bandwidth), and other similar functions. A typical VoIP telephone is shown in Figure 8.7.

The VoIP telephone requires a separate power source that runs through the CAT-5 UTP Ethernet cable into the phone. Ethernet uses two of the four pairs of wires in the CAT-5 Ethernet cable. One pair is the transmit pair and the other is the receive pair. This leaves two unused pairs, one of which Inter-Tel uses to power the phone from a wall socket transformer, as shown in Figure 8.7. The phones can be powered by the PBX when attached to a single head-end run back to the PBX, as long as the cable does not run through any intermediate switches or hubs.

Figure 8.8 shows a close-up of the LAN/power connection and a second hub connection on the VoIP telephone.

This phone permits one CAT-5 cable to be run from a hub to the phone and then extended to a PC host nearby. The only drawback here is that this VoIP phone only supports 10 Base T Ethernet and not 100 Base TX Ethernet. Consequently, connecting a PC host to the SOHO LAN would reduce its speed to 10 MBps.

Brain Teaser: VoIP PBX Go to the Inter-Tel Web site and download PDF data sheets for its Axxess PBX, SoftPhone, and IP networking module. What specifications referenced here appear in these data sheets?