Voice over Internet Protocol

Introduction
Voice over Internet Protocol (VoIP) is used as another medium for communication. Phone calls are through the data network instead of the Public Switched Telephone Network or analog lines.

Voice Over IP is the technology that uses communications protocol to carry voice packets over an IP link. The end-point are typically IP phones or can be computers that uses headset and speaker. Voice before being sent to destination via IP link is converted from analog to IP by means of DSP resources that reside in the IP phones. VoIP technology has brought a revolution in the networking industry by enabling voice packets to traverse through the same link as the data link, thus helping organization to reduce call cost. Companies are able to remove the need to use a separate voice path, remove recurring cost of PBX installation, since VoIP and more preferably it's subset IP Telephony provides centralised call controlling system. The voice packets traverse over the internet/IP link using UDP protocol riding on RTP packets. The stages that are involved from converting the analog to IP voice and then transmitting it through the IP link are mentioned below:

1. Analog signal (Voice that user generates to speak) is sampled as per Nyquist theorem

2. The sample voice is Quantised (Called Quantization)

3. Quantised sample is encoded using 8 bit code

4. The encoded voice is compressed (optional)

5. The compressed voice is sent across the IP link to the remote site

Protocol Breakdown
VoIP phone calls are broken down into packets and sent across the data network. Usually voice packets are tagged with a Virtual LAN (VLAN) tag so that the routers can tell the difference between a voice packet and a data packet. Then the router will redirect the voice traffic to the proper place. For more information visit the Wikipedia page].

Sampling Theorem
Sampling is a process of converting Analog signals into a set of numeric value that can be further transformed into bits for the communication systems to understand. It is basically the conversion or representation of voltages into a digitised format.

Sampling theory is based on Nyquist-Shannon's theorem which states that the sampling rate should be twice the rate of highest frequency of Audible voice so that a perfect reconstruction is possible. The maximum frequency of Human voice is approximately 9000 Hz, although the majority of normal human speech falls in the 3500 hz range. Industry standard value for sampling is considered as 4000 Hz (4KHz). So the sampling rate would be 2 times of 4000 Hz = 8000 samples per second.

Public VoIP Services
Public VoIP service consist of vendors such as [Skype] or [Vonage]. AT&T and TimeWarner are also providing VoIP services through their broadband services. As with public services the customer does not have the phone switch at their place of residence.

Private VoIP Services
Private VoIP Services such as Cisco's CallManager] are private to a company. The company owns all the equipment from phone to phone switch. This kind of implementation is very expensive but also some what more secure because the company has control over every end point before the phone call exits to the PSTN.