Dedicated Digital Service (DDS) is a class of digital services ranging from 56Kbps lease lines to T3 trunks and beyond. It encompasses ISDN and the various incarnations of T. DDS represents the infrastructure of the telco, which, apart from the last copper mile, is almost entirely digital.
Typically, DDS runs over two sets of twisted-pair copper cable. There are two types of DDS: short-haul cables, which can run about 660 feet, and long-haul cables, which can run about 6,000 feet. After these distances, attenuation takes its toll and the signal must be repeated with an aptly named T1 repeater. Higher-speed DDS, such as SONET/SDH, is available via fiber only.
There are several telecommunication-specific terms that we will use in following sections. We'll pause to define them here.
Signaling is the process of exchanging information between points that set up, control, and eventually terminate a call. There are two common methods of embedding the signaling: in-band and out-of-band.
In-band signaling sends control signals embedded in the same channel as the telephone call itself. In telecom parlance, a channel is just a path, real or virtual, through which signals can flow. Out-of-band signaling uses a dedicated channel just for control of the telephone call.
Signal encoding is the process of electronically transmitting the signal on the wire in a specific format. The telecom industry generally uses return to zero bipolar signaling or alternate mark inversion (AMI). The signaling is synchronous and clocked. The term synchronous, in telecom, means that data can always be found in the same position at the same time. This is yet another reason for the all-important clocking and synchronization.
A 0 is signaled by no voltage on the line. Deviation, either positive or negative, indicates a 1 (hence the need for clocking). Typically, pulses alternate in polarity to maintain a 0-volt direct current on the line. If the signal didn't spend an equal amount of time above or below 0, it would eventually saturate the transformer. The alternating positive and negative voltage also serves as a form of error detection.
A bipolar violation occurs when two consecutive pulses in the same polarity are received; this is an indication of one or more bit errors. Bipolar violations are sometimes created intentionally because the bipolar error can be used as a third signal of sorts.
Figure 2 The signal is clocked, so every hash mark represents a number. Either positive or negative deviation from 0 represents a 1 bit. No voltage is a 0. A bipolar error (B) occurs when two like-signed 1s occur in a row. This signals a transmission error. Sometimes, however, bipolar errors are performed intentionally for signaling purposes.
To maintain the 0 volt average on the line, return to zero bipolar signaling has a requirement for "ones density"at a minimum, there must be 12.5 percent (1 bit in every 8) 1s to 0s ratio.
In other words, each 8-bit time slot must contain at least a single 1. So what do you do when you need to send a large block of 0s? There are several answers.
The first method is affectionately known as B7ZCS, bipolar 7 with zero code suppression. This is a form of bit-stuffing, adding bits to the data stream to ensure synchronization. The equipment sets a known bit (the least significant bit) to 1. A voice listener would never be able to tell the difference, but this would really bug a computer. This encoding standard reduces the overall bandwidth to 56Kbps rather than 64Kbps.
ZBTSI, zero byte time slot interchange, is a complicated method of encoding extra 0s by buffering part of them and moving them to a different channel. This method isn't commonly used because of the expense of the buffering hardware.
B8ZS, bipolar with 8-zero substitution, encodes a string of eight consecutive 0s into a series of 1s, 0s, and a specific pair of bipolar violations. B8ZS does not reduce your overall bandwidth and, thus, is required for all "clear channel" 64Kbps transmission rates (see Figure 3).
Figure 3 To code a string of eight 0s, B8ZS sends three 0s, two 1s, a 0, a bipolar error, and a 1.
In the basic DDS definition, because there's no separate signaling data channel, all signaling is done in-band; other line conditions such as idle, out of service, and loopback are also encoded using nonaccidental bipolar violations.