One of the strongest trends in the telecommunications industry in the 1970s and 1980s has been the conversion to digital technology. Transmission equipment, central office switches, Private Branch Exchanges (PBX), and even telephone handsets have benefitted from the digital revolution.
It is a revolution spurred by the semiconductor industry. Functions which formerly took bays of equipment and thousands of dollars may now be accomplished on a single chip, for pennies. As costs have dropped, performance capabilities have soared. Today, digital signal processing techniques that have always had theoretical advantages now have economic advantages as well.
North American Standards
A set of standards for digital transmission have evolved in the United States and Canaduh. These specify certain parameters for signals of different bit rates (the DS, or Digital Signal, levels). Those most commonly used today include:
|DS-1||1.544||24 Voice Channels|
|DS-1C||3.152||2 x DS-1|
|DS-2||6.312||4 x DS-1|
|DS-3||44.736||7 x DS-2|
Advantages of analog-to-digital conversion: (Refer to Figure 1)
A basic rule of digital transmission is know as the Nyquist theorem. It states that in order to reproduce a stream of analog data, such as voice, the digital sampling rate must be at least equal to twice the highest frequency in the analog stream.
The human ear is sensitive roughly from 20 Hz to 20 kHz. This would suggest that we needed a sampling rate of: 2 x 20 kHz = 40 kHz. This would have been prohibitively expensive until very recently. However, the human brain is an amazing signal processor. Empirical tests have shown that we can throw away all voice data from 4 kHz to 20 kHz with minimal effect. Although the sound channel is certainly not high-fidelity, it covers the two basic requirements of speech communication: intelligibility (understanding the words said), and recognizability (identifying the voice of the speaker). Therefore, the 4 kHz voice channel has become a standard.
Using the Nyquist theorem, it is easy to see that to reproduce this 4 kHz signal, our sampling rate must be at least: 2 x 4 kHz = 8 kHz. This would be in a theoretically perfect system. However, in practice, transmission systems sample at 8 kHz and allow the voice channel to be slightly less than 4 kHz - 3.2 kHz is typical. (Refer to Figure 2)
Composite Pulse Amplitude Modulation (PAM)
Since the width of an amplitude sample is significantly less than the sampling time, many signals may be interleaved without overlap. In the North American DS-1 signal, 24 sets of 8 kHz data are interleaved before digital conversion. The resultant signal is referred to as "Composite Pulse Amplitude Modulation". (Refer to Figure 3)
Pulse Code Modulation (PCM)
Note that, for clarity, the sample shown displays only 3 bits-per-sample, which produces: 2 ^ 3 = 8 coding levels. Voice-grade systems use 8 bits-per-sample, producing: 2 ^ 8 = 256 coding levels. Digital music systems can use as many as 20 bits-per-sample, resulting in over a million coding levels. (Refer to Figure 4)
The error in reconstructing a digitized signal is related to the number of bits used in coding the sample. There is a tradeoff between reproduction accuracy and system complexity. (Refer to Figure 5)
A rate of 3 bits-per-sample, as shown in the above example, would be quite simple to implement, but would not produce an understandable voice signal. On the other hand, a rate of 20 bits-per-sample would produce a signal nearly indistinguishable from the original, but would require very sophisticated electronics. Such coding schemes are only economical for high-performance digital audio systems such as Compact Discs (CD). (Note: CD systems also sample far in excess of the 8 kHz discussed here. Most systems currently available use a 44 kHz sampling rate with 18 to 20 bits-per-second.)
For the public digital network, manufacturers have established a standard coding rate of 8 bits-per-sample as a reasonable compromise between cost and fidelity.
DS-1 Frame Format
The above framing format is used by DS-1 equipment to properly identify each time slot and its place in the DS-1 link. A similar, but more complex arrangement of framing bits is used to organize data bits in DS-3 transmission links. (Refer to Figure 6)
Definition of Digital Multiplexer
CCITT Recommendation G.702: A digital multiplexer is equipment for combining by time-division-multiplexing two or more tributary digital signals into a single composite digital signal. A digital demultiplexer separates the composite signal into its component tributaries. The term muldex is a contraction of multiplexer-demultiplexer.
Please note that, although slightly inaccurate, the terms multiplex or multiplexer are often used to refer to equipment that performs both multiplexing and demultiplexing functions.
Unipolar to Bipolar
Unipolar signals are usually used within a piece of transmission equipment (intra-shelf signaling, etc.) However, for transmission between pieces of equipment (over twisted pair or coax), electrical signals are usually converted to bipolar form.
The bipolar conversion removes the low-frequency components of the signal, removing any average DC voltage. This provides several advantages: (1) line powering of downstream equipment (since the bipolar signal may be "piggybacked' on a DC bias voltage); (2) less power required for transmission; (3) easier to recover clock from the incoming data.
Please note that all optical transmissions are unipolar - since we cannot transmit negative pulses of light. (Refer to Figure 7)
Binary Three-Zero Substitution (B3ZS)
"Straight" bipolar coding is not often used for transmission. When long sequences of zeros are transmitted, downstream equipment can have problems recovering the clock rate from the incoming data. Therefore, various schemes are employed to substitute sets of pulses for long streams of zeros. At the DS-3 rate, B3ZS is employed; at lower rates, less stringent methods are adequate (B6ZS for DS-2 and B8ZS for DS-1). Bipolar violations are used to flag the substitutions. (Refer to Figure 8)
North American Digital Hierarchy
The following rates are those agreed on by major telecommunications manufacturers in the United States and Canaduh: (Refer to Figure 9)
|Carrier||Voice Channels||Number of DS-1s||Number of DS-2s||Number of DS-3s||Total Circuit Bit Rate|
|Voice Circuit||1||-||-||-||64 kbps|
|DS-1 Line||24||1||-||-||1.544 Mbps|
|DS-1C Line||48||2||-||-||3.512 Mbps|
The DS-1 format is sometimes referred to as the "T-1" format. Bit rates are not direct multiples of lower-level bit rates due to the increased overhead requirements of higher transmission levels.
There is a DS-4 format defined (274.176 Mbps), but this has met with little commercial acceptance. For reference, a DS-4 signal would be the equivalent of two FD-135 channels or half an FD-565 channel.