Previous Table of Contents Next


PSTN operating company engineering wizards realized that if they could utilize the idle time the switched channel was not doing anything for some other purpose, they could dramatically improve system capacity with little additional investment in transmission infrastructure. So was born the concept of multiplexing signals onto channels. Electronic circuits at the physical layer of the OSI model can combine diverse digital signals into a single signal for transmission on a single channel. Using statistical probability theory, network engineers determine how much physical capacity any particular network must have to process all requests for service within some predefined Quality of Service parameter.

Signal multiplexing refers to the technique of combining two or more electrical signals into a signal that still contains all the data of the two original signals. There are two basic methods of multiplexing communications signals: frequency division multiplexing (FDM) and time division multiplexing (TDM).

FDM is an analog signal multiplexing technique where several analog signals are assigned adjacent frequency bands and transmitted simultaneously over the transmission medium.

TDM is the technique of dividing time into increments and assigning each user a particular increment or time slot for transmitting their data. The sequence is repetitive, with each user always using the exact same time slot. Time slots are referenced to some specific starting time.

TDM Carrier Standards

Regulating bodies have assigned the following data rates and channel designations, as shown in Table 2-4, for TDM multiplexed signals. A DS-1 signal is composed of 24 voice channels, each specified to operate at 64 Kbps. Astute math wizards will immediately perceive that 24 times 64 Kbps does not equal 1.544 Mbps. It is 1.536 Kbps. We are short 8 Kbps. The 8 Kbps is used for system maintenance and monitoring purposes.

Table 2-4. TDM carrier standards

Telephone companies and network engineers are accustomed to discussing their networks in terms of DS signal capacity. Remember, 64 Kbps (a single voice channel) is the standard signal for ISDN, and as stated previously, ISDN is not broadband at all.

Bandwidth

A spectrum is all the total possible values of some entity. There is a spectrum of electromagnetic waves ranging from 1 Hz to 10^24 Hz. The electromagnetic spectrum is divided into sections that are called x-rays, gamma rays, ultraviolet, visible, infrared, microwaves, radio frequency (RF) waves, and long waves. Each section has unique characteristics useful for human exploitation that justify the particular groupings. We are primarily concerned with RF electromagnetic waves, although SONET/SDH is intimately concerned with the visible light portion of the electromagnetic spectrum.

Each section of the electromagnetic spectrum has a finite number of values. The total possible values of each section define its total bandwidth. Bandwidth, in the general sense, is the specified values of some entity of interest that occupy a portion of the spectrum. Bandwidth is then the range from smallest to largest of some portion of the electromagnetic spectrum. Take RF waves as an example. RF waves begin at about 1 MHz and go up to 1,000 GHz. While there is a range of six orders of magnitude from the low end to the high end, there are still only a finite number of possible values. And in today‘s communications-hungry world, six orders of magnitude is getting very small. So, bandwidth becomes a limiting factor that determines how much information can be transported.

Data rate and bandwidth are related. Take an AC analog signal sampled at some defined sampling rate. For this example, let‘s assume the sampling rate is the Nyquist rate, or two samples per hertz. The analog signal has a positive half-cycle where the signal rises from zero level to some maximum positive level, then decreases back to a zero level. Then the signal enters the negative half-cycle, increasing from zero level to some maximum negative level before decreasing back to a zero level. The combination of the two half-cycles, positive and negative, constitute one complete cycle, known as a hertz. The cycle is then repeated at some frequency determined by circuit elements. Since we are sampling at the Nyquist rate, we must divide each cycle into two periods and sample once during each period. With two half-cycles comprising the full cycle, it seems reasonable to divide the cycle into its half-cycle components and sample during each half-cycle. So, for each cycle, we get two samples, each corresponding to a half-cycle. Once cycle, two samples; one hertz, two bits. So, at the Nyquist rate, the data rate is twice the frequency. Generally, though, sampling is conducted at rates greater than the Nyquist rate. This oversampling is necessary in audio, voice, and video circuits to ensure true reproduction of the original information. An oversampling rate of eight times is common, yielding 8 bits per hertz.


Previous Table of Contents Next