Time-Division Multiplexing

Analog Signal Processing
An analog signal varies continuously over time. It could vary in amplitude, frequency, or phase. These components define the sound wave an analog signal represents. The amplitude, frequency, and phase shift are three characteristics of the analog signal that can be varied to convey information. Analog signals are inherently susceptible to attenuation as they progress along the transmission medium. Analog signals are also susceptible to electromagnetic interference (EMI), radio frequency interference (RFI), and other noise sources. This results in signal distortion with changes in frequency characteristics.

Analog telephony signals span the 200-Hz to 3.4-kHz frequency band. Such analog signals are referred to as narrowband due to their narrow frequency response.

Analog video signals operate in a frequency band from flat response (0 Hz) up to 60 MHz. Such analog signals are referred to as broadband due to their wide frequency response. The National Television System Committee (NTSC) and PAL broadcast (radio frequency [RF] transmission) standards impose a limit on the bandwidth of the video signal of about 6 to 10 MHz. Video bandwidth is, effectively, the highest-frequency analog signal a monitor can handle without distortion. Amplification can be used to compensate for signal attenuation. However, narrowband repeaters cannot distinguish between the signal and distortion components of the analog signal. The repeater amplifies the entire input signal, thereby amplifying the noise along with the original signal. The effects of noise and distortion are cumulative along the analog transmission system.

Analog Signal Generation and Reception
The generation of an analog telephony signal takes place when a person speaks into the transmitter of a telephone set. Changes in the air pressure result in sound waves that are sensed by the diaphragm. The diaphragm responds to changes in air pressure and varies circuit resistance by compressing or decompressing carbon in the transmitter. The change in resistance causes a variation in the output voltage, thereby creating an electrical signal analogous to the sound wave. The phone connects to a central office (CO) in the caller's neighborhood through a subscriber line interface circuit (SLIC) that executes functions, such as powering the phone, detecting when the caller picks up or hangs up the receiver, and ringing the phone when required. A codec at the CO converts the analog voice signals to digital data for easy routing through the voice network and delivery to the CO located in the recipient's neighborhood. At the recipient's CO, the digital data stream is converted back into an electrical analog signal. During reception, a varying current flows through the coil and vibrates the receiver diaphragm that reproduces the sound wave. Digital transmission systems overcome the basic analog issue of the cumulative effects of noise and distortion by regenerating rather than amplifying the transmitted signal. The regenerative repeater detects the presence of a pulse (signal) and creates a new signal based on a sample of the existing signal. The regenerated signal duplicates the original signal and eliminates the cumulative effects of noise and distortion inherent in analog facilities.

Analog-To-Digital Conversion
Converting an analog telephony signal to a digital signal involves filtering, sampling, quantization, and encoding. The following example involves an audio frequency (AF) signal.

Filtering
Audio frequencies range from 20 Hz to 20,000 Hz. Telephone transmission systems are designed to transmit analog signals between 200 Hz and 3400 Hz. End frequencies below 200 Hz and above 3400 Hz are removed by a process called filtering.

As indicated in Figure 2-4, a band pass filter (BPF) is used to filter the audio telephony band for analog-to-digital (A/D) conversion. BPFs are constructed using analog electronic components, such as capacitors and inductors.


Figure 2-4. Filtering of the Analog Telephony Waveform

[View full size image]





Sampling
In the sampling process, portions of a signal are used to represent the whole signal. Each time the signal is sampled, a pulse amplitude modulation (PAM) signal is generated. According to the Nyquist theorem, to accurately reproduce the analog signal (speech), a sampling rate of at least twice the highest frequency to be reproduced is required. Because the majority of telephony voice frequencies (200 to 3400 Hz) are less than 4 kHz, an 8-kHz sampling rate has been established as the standard. As illustrated in Figure 2-5, the PAM sampler measures the filtered analog signal 8000 times per second, or once every 125 microseconds. The value of each of these samples is directly proportional to the amplitude of the analog signal at the time of the sample (PAM, as mentioned previously).


Figure 2-5. Pulse Amplitude Modulation (PAM)





Quantization
Quantization represents the original analog signal by a discrete and limited number of digital signals. When the original signal is in a quantized state, it can be safely relayed for any distance without further loss in quality. To obtain the digital signal, the PAM signal is measured and coded. As shown in Figure 2-5, the amplitude or height of the PAM is measured to derive a number that represents its amplitude level. Quantization essentially matches the PAM signals to one of 255 values on a segmented scale. The quantizer measures the amplitude or height of each PAM signal coming from the sampler and assigns it a value from –127 to +127. In telephony systems, each amplitude value (sample) is expressed as a 13-bit code word. Comparing the sample to a companding characteristic, which is a nonlinear formula, forms an 8-bit byte.

Encoding
The decimal (base 10) number derived via quantization is then converted to its equivalent 8-bit binary number. As illustrated in Figure 2-6, the output is an 8-bit "word" in which each bit can be either a 1 (pulse) or a 0 (no pulse). This process is repeated 8000 times a second for a telephony voice channel service. The output (8000 samples/second * 8 bits/sample) is a 64-kbps PCM signal. This 64-kbps channel is called a DS0, which forms the fundamental building block of the digital signal level (DS level) hierarchy.


Figure 2-6. Pulse Code Modulation (PCM)

[View full size image]





µ-law and A-Law Coding
Voice signals are not uniform, and some signals are weaker than others. The dynamic range is the difference in decibels (dB) between weaker (softer) and stronger (louder) signals. The dynamic range of speech can be as high as 60 dB. This does not lend itself well to efficient linear digital encoding. G.711 µ-law and A-law encoding effectively reduce the dynamic range of the signal, thereby increasing the coding efficiency and resulting in a signal-to-noise ratio (SNR) superior to that obtained by linear encoding for a given number of bits. The µ-law and A-law algorithms are standard compression algorithms used in digital communications systems to optimize and modify the dynamic range of an analog signal for digitizing. The µ-law is typically used on T1 facilities, whereas the A-law is used on E1 facilities.

Companding (compression and expansion) is a method commonly used in telephony applications to increase dynamic range while keeping the number of bits used for quantization constant. The compression is lossy, but provides lower quantization errors at smaller amplitude values than at larger values. Basically, the voice is sampled at 8000 samples per second and converted into a 14-bit word (µ-law) or 13-bit word (A-law) that goes into the compander. The samples are processed using a nonlinear formula to transform them into 8-bit words. The compander also inverts all even bits in the word. In A-law companding, for instance, the 13-bit word 1111111111111 is converted to 11111111 (+127) using compression, resulting in the PCM word 10101010 (AA hex). Telephony PCM words use a polarity, chord, and step makeup. Nonlinear coding uses more values to represent lower-volume levels and fewer values for higher-volume levels. This way, µ-law and A-law companding algorithms permit subtleties of a voice conversation to be captured.

Echo Cancellation
Line echo is created when a signal encounters an impedance mismatch in the telephone network, such as that typically caused by a two- to four-wire (hybrid) conversion in an analog system. The hybrid is a transformer located at the facility that connects the two-wire local loop coming from homes or businesses to the four-wire trunk at the CO for inter-exchange carrier (IXC) interconnectivity. The echo is intensified by distance and impedance-mismatched network equipment. In circuit-switched long-distance networks, echo cancellers reside in the metropolitan COs that connect to the long-distance network. These echo cancellers remove electrical echoes made noticeable by delay in the long-distance network. To eliminate echo, echo cancellation devices use adaptive digital filters, nonlinear processors, and tone detectors.

The adaptive filter is made up of an echo estimator and a subtractor. The echo estimator monitors the receive path and dynamically builds a mathematical model of the line that creates the returning echo. The echo estimate is then fed to the subtractor, which subtracts the linear part of the echo from the line in the send path. The nonlinear processor evaluates the residual echo, removes all signals below a certain threshold, and replaces them with simulated background noise that sounds like the original background noise without the echo. Echo cancellers also include tone detectors that disable echo cancellation by user equipment upon receipt of certain tones during data and fax transmission. As an example, the echo-cancellation function is turned off upon receipt of the high-frequency tone that precedes a modem connection.



Circuit-Switched Networks
Figure 2-7 shows an example of a circuit-switched network from a customer's perspective. Such a topology is also referred to as a point-to-point line or nailed-up circuit. Typically such lines are leased from a local exchange carrier (LEC) or IXC and are also referred to as leased lines. One leased line is required for each of the remote sites to connect to the headquarters at the central site.


Figure 2-7. Leased Lines from a Customer Perspective

[View full size image]





The private nature of leased line networks provides inherent privacy and control benefits. Leased lines are dedicated, so there are no statistical availability issues associated with oversubscription, as there are in public packet-switched networks. This is both a strength and weakness. The strength is that the circuit is available on a permanent basis and does not require that a connection be set up before traffic is passed. The weakness is that the bandwidth is paid for even if is not used, which is typically about 40 to 70 percent of the time. In addition to the inefficient use of bandwidth, a major disadvantage of leased lines is their mileage-sensitive nature, which makes it a very expensive alternative for networks spanning long distances or requiring extensive connectivity between sites.

Leased lines also lack flexibility in terms of changes to the network when compared to alternatives, such as Frame Relay. For example, adding a new site to the network requires a new circuit to be provisioned end to end for every site with which the new location must communicate. If there are a number of sites, the costs can mount quickly. Leased lines are priced on a mileage and bandwidth basis by a carrier, which results in customers incurring large monthly costs for long-haul leased circuits.

In comparison, public networks (such as Frame Relay) require only an access line to the nearest CO and the provisioning of virtual circuits (VCs) for each new site with which it needs to communicate. In many cases, existing sites will require only the addition of a new VC definition for the new site.

From the carrier perspective, the circuit assigned to the customer (also known as the local loop) is provisioned on the digital access and cross-connect system (DACS) or channel bank. The individual T1 circuits are multiplexed onto a T3 and trunked over terrestrial, microwave, or satellite links to its destination, where it is demultiplexed and fanned out into individual T1 lines. Figure 2-8 shows this scheme. The T-carrier hierarchy, DS1, and DS3 are covered later in this chapter.


Figure 2-8. Leased Lines from a Carrier Perspective

[View full size image]





TDM Signaling
Signaling in the TDM telephony world provides functions such as supervising and advertising line status, alerting devices when a call is trying to connect, and routing and addressing information. Two different types of signaling information are within the T1/E1 system:

Channel-associated signaling (CAS)

Common channel signaling (CCS)

Channel-Associated Signaling (CAS)
CAS is the transmission of signaling information within the information band, or in-band signaling. This means that voice or data signals travel on the same circuits as line status, address, and alerting signals. Because there are 24 channels on a full T1 line, CAS interleaves signaling packets within voice packets; therefore, there are 24 channels to use for voice. Various types of CAS signaling are available in the T1 world. The most common forms of CAS signaling are loopstart, groundstart, and ear and mouth (E&M) signaling. CAS signaling is often referred to as robbed-bit signaling because signaling bits are robbed from every 6th and 12th frame in a D4 superframe (SF) or 6th, 12th, 18th, 24th frame, and extended superframe (ESF). This is explained in greater detail in a later section.

Common Channel Signaling (CCS)
CCS is the transmission of signaling information out of the information band. The most notable and widely used form of this signaling type is ISDN. One disadvantage to using an ISDN primary rate interface (PRI) is the removal of one DS0, or voice channel (in this case, for signaling use). Therefore, one T1 would have 23 DS0s, or bearer B channels for user data, and one DS0, or D channel for signaling. It is possible to control multiple PRIs with a single D channel, each using non-facility-associated signaling (NFAS). This enables you to configure the other PRIs in the NFAS group to use all 24 DS0s as B channels.