Introduction to G.703

G.703, an ITU-T standard that outlines how to interface digital high-speed circuits, has become the basis of all telecommunications networks. But it didn’t come about overnight. Alexander Graham Bell invented the telephone in 1876, but it would take another 50 plus years before Bell engineer Henry Nyquist calculated digital voice transmission.

Nyquist, who had spent the 1920s working on telegraphy speeds, determined in 1933 that to digitise speech, the analogue speech pattern had to be sampled 8000 times a second. Theory, however, didn’t immediately translate into practical application. That’s because telephony has evolved over a much longer time period than data communications.

Originally, voice channels were multiplexed together by frequency-division multiplexors, which allocated a bandwidth of 3.4 kHz for each voice channel, and a guard tone between each of the mux’s channels minimised crosstalk or interference. This was the first analogue form of multiplexing, but the quality of the voice signal wasn’t superb.

Then came digitisation. With it, 8 bits of telephone voice are sampled at 8000 times a second. Its formula looks like this: 8 x 8000 = 64000 or 64 kbps. This digitising method is called Pulse Code Modulation (PCM) and is defined in detail by the G.711 standard. Companding, a process in which a signal’s amplitude range is compressed before transmission then expanded when received, minimises the number of bits that the PCM must sample in the analogue voice signal before it’s encoded, or “quantised,” in digital form. Companding is available in two encoding varieties: A-law, which is used in Europe, and Mew-law (or µ-Law), common in North America and Japan.

G.703’s physical interface.
G.703 is a specific standard covering physical and electrical characteristics of a digital E1 interface. Under it, data can be travel over balanced (120-ohm) pairs or unbalanced (dual 75-ohm) coax wires; RJ-45 connectors are used for the balanced version, and a pair of BNC connectors is used for the unbalanced interface. Both operate within Layer 1 (the Data-Link Layer) of the seven-layer OSI model.

In the U.K., BT®, the initial telephony provider, built up its G.703 infrastructure using 75-ohm BNC connectors. Most PTTs in the U.K. have followed the same standard and type of interface closely. That’s because when the marketplace was opened up to competitive forces, these providers were allowed to use BT’s existing infrastructure. European PTTs, however, have adopted the 120-ohm interface with RJ-45 connectors.

There are also two types of logical presentation: unstructured and structured. Here’s how they differ:

• Unstructured/Unframed/Clear Channel™ G.703 provides the user with the full 2.048-Mbps bandwidth when run over European E1 lines or 1.544-Mbps bandwidth when supplied via North American T1 lines.
• Structured/Framed G.703 gives a user between 64 kbps and 1.984 Mbps of bandwidth in 64-kbps steps and is also called “framed” service (the G.704 specification details how G.703 operates in frame mode). With structured G.703, you also have the option of running Cyclic Redundancy Check-4 (CRC-4) for bit error monitoring within the first 64K timeslot, which is labelled Timeslot 0 (zero).

Line encoding with G.703.
Line encoding is the method of physically putting all the 1’s and 0’s (that is, the actual data) onto the physical wires. G.703 uses encoding systems, including High-Density Bipolar 3 (HDB3) in Europe and Alternate Mark Inversion (AMI) and Bipolar 8-Zero Substitution (B8ZS) in North America. They all operate on the Transport Layer (Layer 2) of the seven-layer OSI model.

All these line encoding techniques are three-level encoding schemes. In contrast to most data communication protocols in which only two-level schemes typically represent a mark “1” and a space “0,” the three-level system allows an extra change of state, (i.e., a clock to be included). This scheme is used to balance out voltages across wires and, more importantly, encode a clock along with the data structure. The signal is a 1-volt peak-to-peak signal.

Framed vs. unframed services.
As noted earlier, a G.703 service between two sites can be framed or unframed. An unframed service will run at 2.048 Mbps and doesn’t split the data rate in any way. A framed service, however, divides the 2.048-Mbps data stream into 32 64K timeslots. The first timeslot is used not only to initially set up the framing, but also to carry additional information from one end of the line to the other—often when services go international. This leaves 1.984 Mbps for user data.

The type of service provided by a PTT is usually an unframed one. And although it’s possible to order an end-to-end framed service in the rest of Europe, you can’t order one in the U.K.

If you require several devices to be connected at each end, you’ll have to use multiplexors that support G.704 framing. This is the only way multiplexors can divide the bandwidth into blocks that can be allocated to individual end-user equipment. If the equipment doesn’t support G.704 framing, it can’t divide the bandwidth in a way that makes it suitable for support of multiple devices. But if you only need to support a single end-user device, then support for G.704 framing isn't required, and setup is much easier. In fact, the setup only involves choosing a clock source, whether you want to generate the clock internally (in Master Mode) or receive it across across the network externally (Slave Mode).

PBX signalling techniques
There are two main types of Private Branch Exchange or Public Branch Exchange (PBX) signalling: Common Channel Signalling (CCS) and Channel Associated Signalling (CAS).

CCS uses Timeslot 16 to carry a protocol (a defined set of messages or instructions common to connecting devices) between the PBXs. Within that protocol, messages are exchanged relating to information—such a handset lifted, number dialled, ringing tone, engaged tone—for each of the 30 voice channels.

Here are the CCS protocols:

• QSIG, which is used between two ISDN PBXs.
• Q.931, which is used between an ISDN PBX and the outside world.
• Digital Private Network Signalling System (DPNSS), which is used between two non-ISDN PBXs.
• Digital Access Network Signalling System 2 (DASS2), which is used between a single non-ISDN PBX and the outside world.
• CCITT7, which is another example of a protocol that has widespread use; it’s also called a Number 7, System 7 or Common Channel Signalling #7.

G.704 Timeslot Allocation

The diagram shows the allocation of timeslots for a G.704 frame over a 2.048-Mbps G.703 link. Each timeslot is 64 kbps (64K), so in any 1 second of data, there would be 2.048 Megabits of data (1’s and 0’s) split into 32 equal timeslots of 64K. This is called G.704 framing. The first timeslot (Timeslot 0) is used by the end devices to initiate G.704 framing.

It has three synchronisation bits, a Cyclic Redundancy Check (CRC-4) and then national or international bits (depending on the use of the circuit). Some manufacturers use these bits for sending remote management information across the link. Following Timeslot 0 are another 31 timeslots, each of which can be used for user data (voice, data or video). The only other special timeslot is Timeslot 16, which when used between PBXs, carries signalling information such as handset lifted, number dialled, ringing tone and other functions. Although this information can be carried in any timeslot (other than Timeslot 0), it has traditionally used Timeslot 16. Common Channel Signalling (CCS) and Channel Associated Signalling (CAS) and are the two main types of PBX/exchange signalling.

CAS uses bits within Timeslot 16 of the 32 timeslots (0–31) to represent the status of each of the 30 voice channels (see the diagram above). There are 8 bits present in Timeslot 16 (as is the case with the other 31 timeslots), and each complete E1/T1 frame carries information relating to 2 timeslots. That is, within the 8 bits, the first 4 bits represent the first timeslot, and the last 4 bits represent the second. The first frame represents timeslots (voice calls) 1 and 17, the next frame would represent timeslots 2 and 18, and so on up to 15 and 31. This means that 16 frames, together called a “super frame” (or “multiframe”), are needed to provide the information on all channels.