(Continued from How does VoIP Work? A technical guide to functional VoIP)
Conventional voice telephony is transported in a full duplex mode on PSTN circuits optimized for voice. By the "full duplex mode" we mean that there are actually two circuits -- one for send and one for receive -- to support a normal telephone conversation between two parties. Today, once we depart the local area, all of these circuits are digital. The descriptive word "digital" may seem ambiguous to some.
Let's say that 10 years ago we looked ahead to our present time. All the circuits would consist of 8-bit words, which represent voltage samples of analog voice conversations in a PCM format. This is often characterized in the literature as G.711 service (i.e., ITU-T Rec. G.711). Data is also commonly transported in 8-bit sets called "bytes," but more properly called "octets." It is comparatively simple to replace 8-bit voltage samples of voice with 8-bit octets of data.
Essential philosophical differences remain, however, between voice in 8-bit octets and data transmission. A voice circuit is established when a subscriber desires to converse by telephone with some other telephone subscriber. The circuit between the two is set up by a signaling routine. The distant subscriber has a telephone address represented by a distinct telephone number consisting of between 7 and 12 digits. The digit sequence of the number sets up a circuit route and connectivity for conversation. The circuit is maintained in place for the duration of the conversation. It is terminated and the circuit is taken down when one or other party hangs up -- goes on-hook. The address sequence of dialed digits is sent just once, at the initiation of the connectivity. This whole process of setting up a circuit, holding the connectivity in place, and then taking down the circuit is called "signaling."
Signaling on data circuits is approached quite differently. There is the "permanent virtual circuit" (PVC), which has all the trappings of a voice circuit. The similarities stop here. Data transmission consists of frames or packets of data. A frame (or data packet) is made up of a "header" and payload. In some cases, a portion of the header may be appended at the end or on the tail of the data frame. But every data frame, at a minimum, has a header consisting of a destination address(es) and the originator's address. It will also nearly always contain some control information. This may be a word (or octet) count of the payload, a CRC sequence for error detection and/or correction, message priority, and/or some other type of control sequence(s).
Digital circuits on the PSTN have either 24-octet frames in the case of T1 (DS1) or 32-octet frames in the case of E1. Each 8-bit octet represents a voice circuit. Such a circuit may be set up using an Initial Address Message of CCITT Signaling System No. 7 or a sequence of DTMF tones where each frequency pair represents a digit in the range of 0 through 9. Once a circuit is set up, no more address messages or DTMF tones are required until the circuit is taken down.
This is not the case on a data circuit. Such a circuit also uses frames, but each frame has a standard header. A typical data frame is illustrated in Figure 1. The frame structure and how the various octets of the header (and tail) are utilized are governed by a "protocol." Various data protocols are discussed in Practical Data Communications, 2nd ed., Wiley, N.Y., 2001.
We can clearly see that there are two differing philosophies here, one for data communication,
and the other for digital voice. Digital voice is sometimes called "circuit-switched voice." A
majority in the telecommunication community saw how advantageous it would be if we could marry the
two philosophies. That is, one singular approach for both voice and data. Meanwhile, data hobbyists
were trying to transmit voice using data packets. The Internet protocol (IP) became the data
protocol of choice, but there were many drawbacks.
Return to the How does VoIP work? main page.
This was first published in June 2006