Technology

Voice over IP (VoIP) is a technology that allows the transmission of language data over a network that uses the Internet Protocol (IP) as transport protocol. That's how VoIP allows the integration of two formerly separated networks – the telephone and the data network.

VoIP has its origin in the year 1995 when a software, which allowed voice communication with computers with simple mediums (microphon, sound card and PC), hit the market. Nowadays quite a lot of producers offer various hard- and software possibilities that are aimed at bringing VoIP not only to companies but also to private households. Even though VoIP is by now already approved in a lot of companies, the future will show to what extend this technology is able to gain ground in private areas.

VoIP operation mode

Speech is digitized, encoded and packaged in so-called "samples" before it can be transmitted. Subsequently the samples are transmitted to the addressee over the network. There they are recomposed to language/speech.

Protocols

Behind this simplified presentation of the operating mode, several protocols work together. The Internet Protocol, a so-called "connectionless protocol", is responsible for the transport as mentioned above. The advantage is that no unnecessary delay emerges during the transmission of speech. Other protocols relevant for VoIP are described in the following:

  • RTP/RTCP

    The Realtime Protocol (RTP) is specified in the RFC 1889. This standard is responsible for the transmission of real-time data. To fulfill this purpose, the RTP is based on the User Datagram Protocol (UDP). Apart from that, the RTP packets are provided with a time stamp and a sequence number in order to reproduce the original order of the packets when carried to the addressee. The RTP-protocol is completed by the RTP Control Protocol (RTCP).

But before sending those VoIP packets on their travel through the net, first a call has to be signalized and established. So called "signalization protocols" are responsible for that. Those are defined in the H.323-protocol family and SIP.

  • H.323

    H.323 has been developed in 1995 by the ITU-T. It should help in getting inter-compatibility problems under control. H.323 includes a couple of sub-standards in which separate partitions, like connection establishment for example, are specified.
  • H.225

    This protocol takes over the connection establishment and the connection termination between the terminals.
  • H.450.X

    This protocol is necessary in order to provide services and features known from PSTN (call diversion, call waiting etc.). Every service is defined in its own H.450.x-specification.
  • H.245

    Using this protocol, information about the capability is exchanged and the necessary logical channels between the terminals are established.
  • T.120

    This protocol is responsible for pure data transmission. Apart from that it supports data conferences between several participants.
  • Audio-codecs - G.7XX

    Codecs are standardized compressing procedures for transporting speech/language in an appropriate quality from sender to receiver. The single codec standards are distinguished by following criteria: data rate, processor load, speech quality and delay. The delay is especially important when it comes to speech transmission. If the threshold level of 150 ms is exceeded at the so called one way end-to-end delay, the speech quality will degrade.
  • Video-codecs

    Video-codecs are specified in the H.26x-standard. The two most common are:

    • H.261 – this codec is especially suitable for "higher" bandwidth.
    • This codec is intended for lower bit rates, that is to say lower than 64 kBit/s. With the H.263, about the same quality is possible as with the H.261 codec, but with about half the data rate.

The communication in a H.323-network takes place between the following system components: terminal, gateway, gatekeeper and multipoint control unit. The terminal is the remote station in the H.323-network. The gateway, on the other hand, is the connecting piece between two different networks. The conference circuit of several terminals is facilitated by the MCU, whereas the gatekeeper is responsible for the administration of a H.323-zone.
  • SIP - Session Initiation Protocol

    SIP has been standardized by the IETF in 1999 and is functionally comparable with H.323.

    SIP is based upon HTTP. For the transmission of VoIP-packets RTP/RTCP is used.
  • SAP/SDP

    The Session Announcement Protocol (SAP) is responsible for continuous emanation of multicasts with descriptions of the involved sessions (SDP – Session Description Protocol).

    Four logical units are defined: user agents, registrars, proxy servers and redirect servers.

    User agents are terminal endpoint applications, that send and receive SIP-requests. Registrars keep an internal overview over the users inside their network domain and accept their registrations. Proxy servers initiate the forwarding of arrived SIP-requests, the act as SIP-routers. Redirect servers execute address decomposition which means that they re-deliver all those addresses on which the designated user may "stay".

Conclusion

With some basic knowledge one is already able to establish a functional VoIP-network with simple programs (like free VoIP-software from the internet). But problems may arise as soon as one ore more NAT-routers are involved in the network. Here one has to consider that the NAT-router (in the case of H.323) supports H.323-nating and that all the necessary ports are enabled. If more NAT-routers are in the network, it is especially important to be sure that the VoIP-packets are sent back to that specific NAT-router that they originally came from.

In bigger networks, the "quality of service" may play an important role as well. QOS-mechanisms should be applied, where it could come to problems with the speech quality due to high IP-traffic.

top  Top