For over a hundred years, telephone networks were exclusively circuit-switched. What this meant was that for every telephone call made, a dedicated connection was established between the two endpoints, with a fixed amount of bandwidth allocated to that circuit. Creating such a network was costly, and where distance was concerned, using that network was costly as well. Although we are all predicting the end of the circuit-switched network, many people still use it every day, and it really does work rather well.
In the PSTN, there are many different sizes of circuits serving the various needs of the network. Between the central office and a subscriber, one or more analog circuits, or a few dozen channels delivered over a digital circuit, generally suffice. Between PSTN offices (and with larger customers), fiber-optic circuits are generally used.
Since the standard method of digitizing a telephone call is to record an 8-bit sample 8,000 times per second, we can see that a PCM-encoded telephone circuit will need a bandwidth of eight times 8,000 bits per second, or 64,000 bps. This 64 Kbps channel is referred to as a DS-0 (that’s “Dee-Ess-Zero”). The DS-0 is the fundamental building block of all digital telecommunications circuits.
Even the ubiquitous analog circuit is sampled into a DS-0 as soon as possible. Sometimes this happens where your circuit terminates at the central office, and sometimes well before.[97]
The venerable T1 is one of the more recognized digital telephony terms. A T1 is a digital circuit consisting of 24 DS-0s multiplexed together into a 1.544 Mbps bitstream.[98] This bit stream is properly defined as a DS-1. Voice is encoded on a T1 using the μlaw companding algorithm.
The European version of the T1 was developed by the European Conference of Postal and Telecommunications Administrations[99] (CEPT), and was first referred to as a CEPT-1. It is now called an E1.
The E1 is comprised of 32 DS-0s, but the method of PCM encoding is different: E1s use alaw companding. This means that connecting between an E1-based network and a T1-based network will always require a transcoding step. Note that an E1, although it has 32 channels, is also considered a DS-1. It is likely that E1 is far more widely deployed, as it is used everywhere in the world except North American and Japan.
The various other T-carriers (T2, T3, and T4) are multiples of the T1, each based on the humble DS-0. Table 7.2, “T-carrier circuits” illustrates the relationships between the different T-carrier circuits.
Table 7.2. T-carrier circuits
|
Carrier |
Equivalent data bitrate |
Number of DS-0s |
Data bitrate |
|---|---|---|---|
|
T1 |
24 DS-0s |
24 |
1.544 Mbps |
|
T2 |
4 T1s |
96 |
6.312 Mbps |
|
T3 |
7 T2s |
672 |
44.736 Mbps |
|
T4 |
6 T3s |
4,032 |
274.176 Mbps |
At densities above T3, it is very uncommon to see a T-carrier circuit. For these speeds, optical carrier (OC) circuits may be used.
The Synchronous Optical Network (SONET) was developed out of a desire to take the T-carrier system to the next technological level: fiber optics. SONET is based on the bandwidth of a T3 (44.736 Mbps), with a slight overhead making it 51.84 Mbps. This is referred to as an OC-1 or STS-1. As Table 7.3, “OC circuits” shows, all higher-speed OC circuits are multiples of this base rate.
Table 7.3. OC circuits
|
Carrier |
Equivalent data bitrate |
Number of DS-0s |
Data bitrate |
|---|---|---|---|
|
OC-1 |
1 DS-3 (plus overhead) |
672 |
51.840 Mbps |
|
OC-3 |
3 DS-3s |
2,016 |
155.520 Mbps |
|
OC-12 |
12 DS-3s |
8,064 |
622.080 Mbps |
|
OC-48 |
48 DS-3s |
32,256 |
2488.320 Mbps |
|
OC-192 |
192 DS-3s |
129,024 |
9953.280 Mbps |
SONET was created in an effort to standardize optical circuits, but due to its high cost, coupled with the value offered by many newer schemes, such as Dense Wave Division Multiplexing (DWDM), there is some controversy surrounding its future.
As with any circuit, it is not enough for the circuits used in the PSTN to just carry (voice) data between endpoints. Mechanisms must also be provided to pass information about the state of the channel between each endpoint. (Disconnect and answer supervision are two examples of basic signaling that might need to take place; Caller ID is an example of a more complex form of signaling.)
Also known as robbed-bit signaling, CAS is what you will use to transmit voice on a T1 when ISDN is not available. Rather than taking advantage of the power of the digital circuit, CAS simulates analog channels. CAS works by stealing bits from the audio stream for signaling purposes. Although the effect on audio quality is not really noticeable, the lack of a powerful signaling channel limits your flexibility.
When configuring a CAS T1, the signaling options at each end must match. E&M (Ear & Mouth or recEive & transMit) signaling is generally preferred, as it offers the best supervision. Having said that, in an Asterisk environment the most likely reason for you to use CAS would be for a channel bank, which means you are most likely going to have to use FXS signaling.
CAS is very rarely used on PSTN circuits anymore, due to the superiority of ISDN-PRI. One of the limitations of CAS is that it does not allow the dynamic assignment of channels to different functions. Also, Caller ID information (which may not even be supported) has to be sent as part of the audio stream. CAS is commonly used on the T1 link in channel banks.
The Integrated Services Digital Network (ISDN) has been around for more than 20 years. Because it separates the channels that carry the traffic (the bearer channels, or B-channels) from the channel that carries the signaling information (the D-channel), ISDN allows for the delivery of a much richer set of features than CAS. In the beginning, ISDN promised to deliver much the same sort of functionality that the Internet has given us, including advanced capabilities for voice, video, and data transfer.
Unfortunately, rather than ratifying a standard and sticking to it, the respective telecommunications manufacturers all decided to add their own tweaks to the protocol, in the belief that their versions were superior and would eventually dominate the market. As a result, getting two ISDN-compliant systems to connect to each other was often a painful and expensive task. The carriers who had to implement and support this expensive technology, in turn, priced it so that it was not rapidly adopted. Currently, ISDN is rarely used for much more than basic trunking—in fact, the acronym ISDN has become a joke in the industry: “It Still Does Nothing.”
Having said that, ISDN has become quite popular for trunking, and it is now (mostly) standards-compliant. If you have a PBX with more than a dozen lines connected to the PSTN, there’s a very good chance that you’ll be running an ISDN-PRI (Primary Rate Interface) circuit. Also, in places where DSL and cable access to the Internet are not available (or are too expensive), an ISDN-BRI (Basic Rate Interface) circuit might provide you with an affordable 128 Kbps connection. In much of North America, the use of BRI for Internet connectivity has been deprecated in favor of DSL and cable modems (and it is never used for voice), but in many European countries it has almost totally replaced analog circuits.
Basic Rate Interface (or Basic Rate Access) is the flavor of ISDN, and is designed to service small endpoints such as workstations.
The BRI flavor of the ISDN specification is often referred to simply as “ISDN,” but this can be a source of confusion, as ISDN is a protocol, not a type of circuit (not to mention that PRI circuits are also correctly referred to as ISDN!).
A Basic Rate ISDN circuit consists of two 64 Kbps B-channels controlled by a 16-Kbps D-channel, for a total of 144 Kbps.
Basic Rate ISDN has been a source of much confusion during its life, due to problems with standards compliance, technical complexity, and poor documentation. Still, many European telecos have widely implemented ISDN-BRI, and thus it is more popular in Europe than in North America.
The Primary Rate Interface (or Primary Rate Access) flavor of ISDN is used to provide ISDN service over larger network connections. A Primary Rate ISDN circuit uses a single DS-0 channel as a signaling link (the D-channel); the remaining channels serve as B-channels.
In North America, Primary Rate ISDN is commonly carried on one or more T1 circuits. Since a T1 has 24 channels, a North American PRI circuit typically consists of 23 B-channels and 1 D-channel. For this reason, PRI is often referred to as 23B+D.[100]
In Europe, a 32-channel E1 circuit is used, so a Primary Rate ISDN circuit is referred to as 30B+D (the final channel is used for synchronization).
Primary Rate ISDN is very popular, due to its technical benefits and generally competitive pricing at higher densities. If you believe you will require more than a dozen or so PSTN lines, you should look into Primary Rate ISDN pricing.
From a technical perspective, ISDN-PRI is always preferable to CAS.
Signaling System 7 (SS7) is the signaling system used by carriers. It is conceptually similar to ISDN, and it is instrumental in providing a mechanism for the carriers to transmit the additional information ISDN endpoints typically need to pass. However, the technology of SS7 is different from that of ISDN; one big difference is that SS7 runs on a completely separate network from the actual trunks that carry the calls.
SS7 support in Asterisk is on the horizon, as there is much interest in making Asterisk compatible with the carrier networks. An open source version of SS7 (http://www.openss7.org) exists, but work is still needed for full SS7 compliance, and as of this writing it is not known whether this version will be integrated with Asterisk. Another promising source of SS7 support comes from Sangoma Technologies, which offers SS7 functionality in many of its products.
It should be noted that adding support for SS7 in Asterisk is not going to be as simple as writing a proper driver. Connecting equipment to an SS7 network will not be possible without that equipment having passed extremely rigorous certification processes. Even then, it seems doubtful that any traditional carrier is going to be in a hurry to allow such a thing to happen, mostly for strategic and political reasons.
[97] Digital telephone sets (including IP sets) do the analog-to-digital conversion right at the point where the handset plugs into the phone, so the DS-0 is created right at the phone set.
[98] The 24 DS-0s use 1.536 Mbps, and the remaining .008 Mbps is used by framing bits.
[99] Conférence Européenne des Administrations des Postes et des Télécommunications.
[100] PRI is actually quite a bit more flexible than that, as it is possible to span a single PRI circuit across multiple T1 spans. This can give rise, for example, to a 47B+D circuit (where a single D-channel serves two T1s) or a 46B+2D circuit (where primary and backup D-channels serve a pair of T1s). You will sometimes see PRI described as nB+nD, because the number of B- and D-channels is, in fact, quite variable. For this reason, you should never refer to a T1 carrying PRI as “a PRI.” For all you know, the PRI circuit spans multiple T1s, as is common in larger PBX deployments.