Voice Over IP

Behind
the wallpaper of every networked office lies a myriad of cables and wires.
There’s network cabling for data, and there’s telephone wiring for voice
communication. There’s electrical wiring too, but here we’re interested
in the first two only. Efforts are underway to integrate these two, so that
voice can be transferred over your data network as well. The concept is
known as Voice over IP–or VoIP for short–and uses TCP/IP to carry voice
packets over a data network. It uses various modules to integrate your
existing telephone/PBX (Private Branch Exchange)/fax lines with your data
network.

VoIP can be set up using what’s
called a VoIP module. This plugs into your network just like an ordinary PC.
You can then either connect regular telephones to it, or connect it to your
company’s EPABX and let everyone in the office use it. However, VoIP isn’t
feasible for intra-office communication–because why would you want to
communicate within your office through VoIP when you can directly dial? But
it could become a cost saver for inter-office communication. For example,
suppose you have offices in different cities across the country, which are
already interconnected through leased lines. You can then connect VoIP
modules at each location and use your existing WAN for inter-office voice
communication. This way, you can save on STD calls between your various
offices.

Extending this argument
further, VoIP could also be used over the Internet if you have VPN
connectivity. Here, you would have servers in different offices connected to
their respective local ISPs through static IP addresses. They would have VPN
software running and would use the Internet for inter-office data
communication. VoIP modules would connect to these servers, allowing voice
communication over the Internet as well. However, voice communication over
the Internet is currently not possible in India due to legal implications.

Inside a VoIP module

A VoIP module
has four sub-modules that are based on two chips–a microprocessor and a
DSP (Digital Signal Processor). It has embedded software for performing all
its functions. The first sub-module–the voice packet sub-module–runs on
the DSP. This performs various voice functions like tone generation,
handling echo, compressing voice data for transmission, etc.

The other three sub-modules–the
network protocol module, the network management module, and the telephony
signaling gateway module–run on the microprocessor. The network protocol
module consists of the IP signaling stack for IP networks. In case of ATM or
Frame relay networks, it would consist of their signaling protocol stacks.
The network management module takes care of the physical interface to the
telephone jack, and generates the required signals at a level compatible
with the telephone. The telephony signaling gateway module takes care of
debouncing in case someone happens to plug in an old rotary type of
telephone. It also converts telephony signaling information into a format
compatible with the signaling protocol used by the VoIP module.

Drawbacks of VoIP

Since voice
has to be packetized and transmitted over a data network, it raises some
problems not experienced over ordinary telephone lines. These problems
affect the overall speech quality, and include:

  • Delay: An IP
    network sends packets along the shortest route available at the time of
    transmission. So there’s a possibility of packets arriving out of
    order or being delayed. This, coupled with the time taken to re-arrange
    and re-assemble the packets, causes delay in the transmitted/received
    speech. The ITU (International Telecommunication Union) recommends that
    one-way delay should not exceed 150 ms for achieving good quality voice.

  • Jitter: One of the
    consequences of delay is jitter. This is also caused when packets are
    lost over the network. It causes the speech to sound choppy and
    difficult to understand. To counteract this, a jitter buffer is
    used, which buffers the received packets before they’re sent to the
    voice codec. A ratio of the time difference between each received packet
    is calculated and used to dynamically adjust the size of the jitter
    buffer.

  • Packet loss: This
    occurs due to degradation of packets over a network, or due to
    congestion. Continual losses of about 5-10 percent of the received
    packets can make a significant difference in voice quality. For small
    losses, the same principle is used as in your audio CD player. That is,
    the last bit of data is replayed as a substitute for the missing data.
    This usually goes unnoticed, unless a large amount of data is missing.

Currently, there are a few
solutions available for VoIP. However, the use of VoIP is still limited
because not everyone can interconnect their offices through WANs. It could
become a feasible solution if more bandwidth is available with ISPs, and the
legal implications of using voice over the Internet are taken care of.

Leave a Reply

Your email address will not be published. Required fields are marked *