VoIP - IP phone

4. H.323 Components 

4.1 H.323 Terminals (1/2) 

The IP phone - also called the H.323 terminal - converts speech from analog to digital format. IP packets carry this digital information through the network. The receiving phone restores the analog signal. 

 Two IP phone versions are available: the H.323 hardphone looks like a standard fixed-line phone. The H.323 softphone is an application running on a PC with headset. Video and Data support is an option in both systems. 

4.1 H.323 Terminals (2/2) 

The structure of an H.323 terminal is standardized and divided into several layers. Each layer has to fulfill a particular task. The lowest layer is represented by the Registration Admission Status (RAS). Its function is to register the terminal with the gatekeeper and control tasks prior to the call. 

H.225 includes call control messages for call setup and release. 

The next layer contains H.245. It provides information on which data compression algorithms are available. Additionally the quantity of speech information per IP packet is exchanged between the end points and logical channels for audio transmission are set up. The  highest  layer  contains  various  audio  and  video  codecs  for  the  conversion  and compression of analog voice signals and optional video data. 

The Receive-Path-Delay-Block or the "Jitter Buffer" ensures a continuous data stream to the codec's input. It arranges the received IP packets in the right order and compensates for delays in transmission. 

All H.323 terminals support real time, full duplex communication to other H.323 endpoints via a LAN interface using IP. 

4.1.1 The software IP phone (1/2) 

With the help of client software every normal PC can become a LAN phone. A loudspeaker and a microphone complete the equipment. 

The client emulates a normal telephone and offers a Graphic User Interface  (GUI) with configurable layout and functions. These configurations may not only apply to the company or department but also to each individual user. Thus each user may design his GUI to his personal  needs  almost  without  limitations,  e.g.  the  number  and  position  of  keys  for abbreviated dialing can be modified in just a few seconds. 

Such a wide variety of modification is impossible using standard telephones. The client software is completely integrated into the existing computer environment. Thus each user will find his personal GUI even when he is at a different location or using remote access via the Internet. Various lists are included, e.g. a phone number list and a list of automatic completion of calls to busy subscribers. 

4.1.1 The software IP phone (2/2) 

Microsoft  NetMeeting  is  a  widely  used  application  for  communicating  over  the  Internet. NetMeeting  supports  real  time  communication  for  both  speech  and  video.  In  addition, functions like bulletin board, text based chat and file transfer are offered. Using application sharing  several  users  may  work  e.g.  a  PowerPoint  presentation  could  be  designed  in conference, independent of the users location. 

NetMeeting  connects  different  users  via  the  Internet  using  IP  addresses  or  an  Internet Locater Server  (ILS). Because a PC is not always reachable and can also change its IP address, the ILS offers information on subscribers that are currently online. NetMeeting follows the H.323 standard and can be used as an IP softphone in every VoIP environment. 

4.1.2 The hardware IP phone (1/4) 

IP telephony must also be possible without using a software client. A hardware phone is needed if 

•    the PC is switched off

•    the user does not accept a PC based phone •    high availability in case of mains breakdown is important •    the environment does not allow a software client for any reason

The difference between an IP hardware phone and a standard phone is the interface used. Instead of a twisted pair or ISDN interface the IP phone is connected via 10-base-T or 100base-T. The protocols will also have to change. An IP phone needs TCP/IP and H.323 whereas ISDN protocols can be used in a standard phone. Furthermore the hardware phone compresses voice following G.723 and G.729 respectively. 

4.1.2 The hardware IP phone (2/4) 

One of the biggest advantages of IP telephony is the flexibility in connecting the phones: changing the physical connection of a standard phones does not necessarily change the parameters of the phone to the new outlet. Moving an IP phone from one outlet to another is just a case of disconnecting here and pluging in there, the complete functionality with all parameters remains. The only condition is network access. 

Most  H.323  terminals  support  either  static  IP  addressing  (i.e.  IP  address  is  configured manually at the phone) or the dynamic assignment of IP addresses. 

In the dynamic verison after an IP hardware phone has been connected to the network it has to broadcast a Dynamic Host Configuration Protocol (DHCP) - "Discover" message. The DHCP server receives the message and answers it with a "DHCP-offer". This message contains IP addresses from the servers database which are available for the DHCP-Client. The client accepts the address information by sending a "DHCP-request". The DHCP server acknowledges it with a "DHCP-ACK" containing an IP address and additional options. Then the phone contacts the IP-PBX via TFTP protocol, identifies itself with its unique MAC hardware address or with a user login and gets its profile including its telephone number, PBX extension, keypad layout and activated services. 

4.1.2 The hardware IP phone (3/4) 

Most users need a PC alongside an IP hardware phone at their workspace. Usually a hardware phone includes a hub for connecting the PC. When an Ethernet connection is present the transmission speed is usually negotiated to 100 Mbps half duplex, if an older interface is used the speed might be reduced to 10 Mbps. 

In  the  latter  case  it  is  useful  to  have  separated  network  connections  for  each  device. Standard  hardware  phones  from  various  manufacturers  can  be  equipped  with  a  voice interface. Analog phones, DECT base stations for cordless phones and analog facsimiles machines can be integrated into a Voice over IP environment by using a VoIP terminal adapter. 

As an IP hardware phone needs an external power input several suppliers of  Ethernet switches implement the power supply via switch ports pin 7 and 8, which are not used for data  communications.  This  is  known  as  in-line  power  supply  and  follows  IEEE-802.3af standard. Furthermore newer IP hardware phones offer switch ports for the IP phone and 100 Mbps Ethernet full duplex instead of an integrated hub. 

4.1.2 The hardware IP phone (4/4) 

An  IP  hardware  phone  is  made  up  of  several  modules.  The  central  element  is  a microcontroller for signaling. It is responsible for address translation from keypad input to IP address with the help of the gatekeeper. It also takes care of call handling and signaling for call  setup  and  teardown  including  the  negotiation  of  all  necessary  parameters.  A  user interface offers the drivers for display, keypad and ringing tone. 

A DSP performs voice processing including echo suppression, voice activity detection, voice compression and packaging as well as compensation for jitter, loss of packages and delay variations. 

The DSP contains a Pulse-Code-Modulation (PCM) interface and a DTMF dual tone multi frequency generator. Other elements in IP hardware phones include a flash memory for software updates and a random access memory RAM. 

The RJ45 interface to the IP network is a transceiver with an automatically selected data rate of 10 and 100 Mbps. 

4.2 The Gatekeeper 

The Gatekeeper, mostly integrated in gateways or IP PBX, is the brain of a Voice over IP network.  As  the  central  element  it  is  responsible  for  the  setup  and  maintenance  of connections. Basically it has the function of the switching matrix in a PBX. The gatekeeper can  be  seen  as  a  table:  each  line  manages a  user  with  phone  number,  extension,  IP address, status and bandwidth. 

A Voice over IP network, also called an H.323 network, can be split into several zones, each with  a  particular  number  of  phones,  gateways  and  MCUs.  Each  zone  needs  its  own gatekeeper  to register the  zone  members. The zones  may be  defined  by  geographical aspects (e.g. branches of a company), by the physical network structure or they may follow the organization chart of a company 

The gatekeeper processes all activities in a zone. Whenever a zone member becomes activated it sends a registration request to the gatekeeper. 

4.2.1 Main functions of the Gatekeeper

The gatekeeper supports the following functions: 

Address translation: All phone numbers dialed with an IP phone have to be translated into an IP address as the user does not normally know the IP address of the B subscriber. With the help  of  translation  tables  the  gatekeeper  changes  the  alias  address  like  E.164  phone numbers, URLs and e-mail addresses into the IP address needed for connection setup. Thus all IP terminals must be registered with the gatekeeper before they can entered into a zone. This registration includes the transmission of IP and alias addresses. 

Access control: A terminal has to send various signaling information to the gatekeeper to initiate a connection. Based on this information the gatekeeper will accept or refuse the request. A refusal may happen, e.g. if a user's pre-paid account has run out. Or he requests a service which he is not subscribed to, e.g. making an international call when he only has rights for national calls. 

Bandwidth control: The gatekeeper monitors and controls the IP terminals bandwidth needs and  making  sure,  that video  and  audio  traffic does  not  exceed  the  administrative  limit. Choosing the appropriate limit can ensure, that a WAN connection does not use up all available bandwidth but leaves a minimum to other applications like email or FTP. 

4.2.2 Additional functions of the Gatekeeper 

The gatekeeper supports the following additional functions. 

User identification and rights: The gatekeeper performs a user identification with KerberosAuthentication or via a certificate of authenticity. It allows or rejects calls from each user by checking configurable rights. These rights may depend on the zone, the gateway used, the bandwidth available and requested, or the time. 

Collection  of  charging  information:  After  closing  a  call  the  gatekeeper  transfers  all  the relevant information such as the B-number, A-number, and the calls duration to a system handling Call Detail Recording (CDR). 

Access to X.500 Directory Services: The gatekeeper grants access to directory services like Microsoft Active Directory or Novell NDS for centralized storage of user data. A directory service includes a hierarchical database with objects called attributes. Some attributes of user objects are name, phone number and e-mail address. The contents of the attributes are values. 

4.3 Das H.323 Gateway (1/4) 

A  gateway  is  hardware  and  software  interconnecting  different  networks  with  or  without protocol conversion. The function of a gateway is to transmit messages from one computer network to another. It is located on the lowest protocol layer common to both networks. In extreme cases, that layer could be the OSI application layer. 

The gateway can interpret the protocols of either side, thus it can decode the data packages of one side down to the lowest common layer and pack it again with new header information for the other sides protocols. The complete protocol conversion includes address translation, data formatting, buffering of packages and acknowledgement as well as flow control and data rate adaptation. 

4.3 Das H.323 Gateway (2/4) 

H.323 gateways enable connections between voice and data networks. They offer Ethernet interfaces for data network connection and analog or ISDN interfaces for circuit switched networks. 

The ISDN interface is either a Basic Rate Interface (BRI) with 2 voice circuits or a Primary Rate Interface (PRI) handling 30 circuits following E1 standard. Besides protocol conversion, IP telephony gateways internally work on procedures for call setup and teardown as well as voice packet modification, e.g. from IP codec G.729A (8 kbps) to G.711 (64 kbps). Thus VoIP subscribers are fully available from the public switched telephone network. The  responsibilities  of  the  gateway  as  an  endpoint  includes  the  Jitter  Buffer,  runtime optimization,  and  echo  suppression  etc.  The  translation  functions  of  the  gateway  are described in H.246. A gateway is not necessary if no connection to a standard telephony network exists. 

4.3 Das H.323 Gateway (3/4) 

The hardware- and software structure of a gateway are similar to that of an IP hardware phone. 

The hardware differs in the number of DSPs and the gateway does not have a user and voice interface. Instead it has one to the telephony network. 

4.3 Das H.323 Gateway (4/4) 

On the IP network side, the gateway supports call setup and release using H.225 with Q.931 messages. 

The protocol Q.931 is used for signaling in both ISDN and H.323. In ISDN the D-channel is the transport media. As no equivalent exists in IP, TCP is used instead. With RAS signaling a gateway is registered to the gatekeeper. The H.245 layer above is responsible for call control signaling and for opening a logical channel carrying video and audio data. Signaling enables the exchange of capability information to the H.323 terminal as well as the negotiation of functionalities such as the audio codec used on the IP network side. In the highest layer, voice frames are transcoded into a constant bit rate of 64 kbps. 

On the ISDN side, a B-channel transports this bit stream. H.225 and H.245 signaling is found on the appropriate layer of the ISDN D-channel. 

Remarkably, from an H.323 network point of view, the gateway behaves like an H.323 terminal. Correspondingly, the gateway also behaves like a terminal from the telephony network point of view. This means that when a call is made from an H.323 network into a telephony network the H.323 gateway has to convert all necessary signaling. 

4.4 The MCU (1/2) 

A conference is a virtual meeting of three or more distant partners with a real time audio or video communication. All data, i.e. audio, video and text, can be digitized, exchanged and worked on simultaneously. 

The  Multipoint  Control  Unit  MCU  supports  IP  telephony  conferences  between  several terminals. All of them have to establish a connection to the MCU. The MCU checks the voice processing abilities of each terminal, decides which codec to take and distributes the voice information, in a process called media streaming. 

The MCU is composed of the following elements: 

The Multipoint Processor (MP) receives audio and video data streams from the conference participants, mixes them into one data stream and then distributes this to the terminals that are part of the multipoint conference.

The  Multipoint  Controller  (MC)  offers  basic  functions  for  multipoint  conferences.  It  is responsible for the negotiation of capabilities and the control of the conference. The MC function can be located in a terminal, a gateway, the gatekeeper or in an MCU. The MCU is an endpoint supporting multipoint conferences and consists of at least one MC and one or more MPs.

4.4 The MCU (2/2) 

During a centralized multipoint conference all terminals send their multimedia data to the MCU  using  a  point-to-point  connection.  In  addition,  the  data  stream  contains  different signaling information to be processed by the MC. For the MCU to manage a centralized conference, it has to have at least one MP as well as the MC. The task of this MP is mixing the received data streams into one and distributing it to all terminals within the conference. Using  multicast  technology  terminals  in  a  decentralized  conference  can  exchange  data directly. In contrast to a centralized conference, the MCU in a decentralized conference only has to perform different control tasks. 

In a decentralized conference the multimedia data streams have to be combined in the terminals. The MC may be located in one terminal. When several terminals hold an MC the one  of  highest  value  controls  the  conference.  This  process  is  called  master  and  slave designation.