Read pages 236 to 254 and this lesson.  Then, complete the hw8 before the lecture.  Bring your homework notes to class.

Multiplexing
Chapter 7 describes multiplexing, the practice of combining several signals together so that they may be sent simultaneously over the same medium.  In this lesson, we will investigate the different kinds of multiplexing including one specific implementation used for voice and data networks, the T1 line.

Let's begin by defining some simple terms.

Multiplexing:  combining several communications channels together so they may be transmitted as a single signal on a medium.

Demultiplexing:  separating multiplexed signals.

Why would you want to do this?  One reason is to save money.  For example, the telephone company combines several telephone conversations into a single signal that it transmits over a single cable.

Way back in an earlier chapter you studied how to draw a signal in the frequency domain.  We need to return to the frequency domain and take a look at a typical voice signal.  The human voice ranges from about 100 Hz to about 4000 Hz.  Voice has many different frequency components.  The most powerful components are centered around 1000 Hz.  The frequency components of voice at higher frequencies are much less powerful.  Below 1000 Hz, the power drops off rapidly, too.  Take a look at the rough sketch of the average voice power at different frequencies in the graph below.

We also noted in an earlier chapter that a typical cable supports a much greater range of frequencies than is required for one voice signal.  If there were some way to use those frequencies we could make more efficient use of the bandwidth. 

Frequency Division Multiplexing
Dividing up a range of frequencies and allocating those parts to different signals for simultaneous transmission.  This can be simply explained as users taking a small portion of the bandwidth all the time.  Here is a simple example... five cellular phone users share 1 MHz of bandwidth, from 100 MHz up to 101 MHz.  The first user gets 100 to 100.2 MHz, the second user gets 100.2 to 100.4 MHz, and so on on up to the fifth user who gets 100.8 to 101.0 MHz.  In simple terms FDM gives a user "part of the bandwidth all of the time."

Note that the frequency axis is vertical in the above graphic.

Voice signals can be lined up in a row for transmission.   Up to about 13,000 channels have been combined onto a single coaxial cable for simultaneous transmission.  A few voice signals combined in this way might look like the following diagram.

Time Division Multiplexing
Dividing up the available time and allocating those parts to different signals for simulated simultaneous transmission.  This can be simply explained as users taking turns to access the full bandwidth of the channel for a portion of the available time.  Our cellular phone users might get all the bandwidth  for 1/5 of the time, each taking their turn in order, over and over again.

Note that the frequency axis is vertical in the above graphic.

Space Division Multiplexing
Dividing up a channel by location so that simultaneous transmissions may occur.  This might be best explained with an example.  Radio stations are assigned a particular frequency, in a particular area.  That same frequency may be assigned to another radio station in another area.  Another example is for cellular telephones.  Within a cell, a phone using a particular frequency encounters no interference from another phone in a nearby cell using the same frequency. 

Summary

FDM:  user gets part of the frequency all the time
TDM:  user gets all the frequency part of the time
SDM:  user gets all the frequency all the time in a certain location (usually applies to transmissions of radio, TV or mobile phone signals)

The T-System
The T-System is a collection of time division standards.  Each one has slightly different rules of operation and each operates at a different total data rate.  We will look at the two most common T-System protocols, the DS0 frame and the DS1 frame.  DS0 means "Digital Signal" level "Zero".  The DS1 would of course be "Digital Signal" level "One".  Both of these operate on ISDN lines.  ISDN stands for Integrated Services Digital Network.

The DS0 frame was created to hold one voice sample which is eight bits.  A standard digital telephone line will transmit DS0 frames synchronously (one after another with no pauses) to support one telephone conversation.

For a quality voice signal, 8000 DS0 frames must be sent per second.  This means that to have a quality voice signal, 8000 samples of the voice must be made in one second. 

Memorize the magic number:  8000

If you order a digital phone line from your local carrier (the phone company), you'll be offered ISDN service at the basic rate.  BRI, or basic rate interface, is the standard ISDN offering for the residential market.  It is actually two phone lines and a control channel.  The lines are called B channels and the control channel is the D channel. 

Digitized voice samples are sent over the B channel 8000 times per second.  Because there are 8 bits per DS0 frame, the data rate on the B channel is exactly 64,000 bits per second. 

Once your voice samples travel all the way to the carrier's central office, they are multiplexed with other voice samples that are heading off in a similar direction. 

A T1 line is a physical standard that supports sending DS1 frames.  DS1 frames represent Time Division Multiplexed voice samples.  Your voice samples must be sent at the (magic number) rate of 8000 per second.  Mixed with other voice channels, a very high capacity link is required to support the combination.

A DS1 frame is a collection of 24 DS0 frames with one additional bit at the front, called a framing bit.  The framing bit is used to keep track of the beginning of the frames as they come in.  The framing bit alternates for each DS1 frame.

One voice sample from each of the 24 voice channels is placed in the DS1 frame.  The whole frame is then sent out and the next frame is constructed.  This must happen without delay.  If a particular voice channel has no activity, zeros are placed in the channel and it is sent anyway.

There are 24 channels in a DS1 and each channel is an 8-bit voice sample.  That means there are 192 bits for the channels.  There is also a single framing bit at the beginning of the frame.  Total everything up and you get 193 bits for a DS1 frame.  The DS1 frame contains only one sample from each channel.  The samples of a channel have to be sent 8000 times per second.  Therefore, 8000 DS1 frames have to be sent in one second.

8000 DS1 frames/sec X 193 bits/frame = 1,544,000 bits/sec = 1.544 Mbps

A T1 may also be used to send data.  Sometimes the data is simply placed in the voice channels and multiplexed with the voice.  This can be a problem because some systems actually steal a bit from the voice channel so that control signals can be sent.  Control signals are used to set up, monitor and tear down voice connections.  At other times signals are sent on a separate channel to control the connection.  In this case, data transmissions are not affected.

Bit robbing is a term used to describe using the least significant bit in a single voice channel for signaling.  The voice sample bit with the least importance (least significant bit, on the right hand side in our DS0 diagram) is overwritten with the control information.  Half the time the bit is, by chance, correct.  Of course, the other half the time, the bit is incorrect.  For a voice conversation, you'll never notice any difference.  However, for a data connection, having a bit in error is fatal. 

How does bit robbing work and what can we do to send data over a T1 system that uses bit robbing?

To send control information along with the voice samples, the least significant bit in every channel of a DS1 frame is overwritten with a signaling bit.  This happens once in six frames.  The control information is used on a voice link to open the connection, send the dialed numbers from switch to switch, monitor the links state, and disconnect at the end of the call.  For data, the control channel might not be used.

To avoid problems when sending data on this link, data is never put in the least significant bit.  That means that only 7 bits of a channel are used for data.  The eighth bit is reserved for signaling information.

So, why not put data into all eight bits for five frames and then use just seven bits for the frame that has bit robbing applied?  It turns out that the logic is too expensive to implement.  It is cheaper to just set the connection to use 7 bits out of 8 in the channel.

Example:  What is the data rate on a DS1 in a single channel when the line uses bit robbing?

Only seven bits are available for use in sending data.

7 bits/channel X 1 channel/frame X 8000 frames/sec = 56,000 bps

Another example:  What is the throughput on a T1 line if all channels are used for data and bit robbing is used.

Only seven bits are used for data in each channel, but all 24 channels are used. 

7 bits/ch x 24 ch/frame = 168 bits are used for data

168 bits of info / 193 bits total = 0.87 --> 87% throughput

How is the framing bit used?
The framing bit is a single bit at the beginning of the DS1 frame.  This bit is important as it marks the beginning of a frame.  We need to be able to distinguish which bits belong to which channels.  The framing bit helps us keep track of the beginning of the frame.

For DS1 framing, the framing bit alternates between 1 and 0 on each frame.  The receiver scans the incoming data until is locates a very long string of these alternating bits.  By storing the frames in memory and searching for the pattern, 0101010..., the frames can be "aligned".  That is the beginning of each frame can be determined.

Example:  What is the overhead of a DS1 frame?

overhead = overhead bits / total bits  = 1 / 193 = 0.00518
overhead = 0.52%

What is the overhead data rate?

overhead rate = overhead bits in a frame * number of frames per second
overhead rate = 1 bit/frame * 8000 frames/sec = 8000 bits/sec

There were some problems with this type of framing.  The frame alignment was easy to lose.  Then the link would need to be re-syncrhonized (i.e. search for that 1010... pattern and find the beginning of the frames).  Doing this wastes time and forces retransmission of lost data, further slowing the connection. 

An improved version of framing came about.  The standard DS1 frame was used (one frame bit and 24 channels that are 8 bits each), but a group of 12 frames were bundled together and the framing bits were defined in a specific patter for the 12 frames.  This permitted more accurate tracking of the beginning of the frames and reduced the number of times that the link had to be re-syncrhonized.  The package of 12 frames was called a super frame.

Yet, this was still not enough to keep the link up and running for long periods of time without losing track of the beginning of the frames.  Extended Super Frame (ESF) was developed to cure this problem.  ESF bundled 24 frames together.  The frames were still DS1 frames, each with a single framing bit.  However, the framing bit was used for several different functions. 

DS1 framing and Super Frame had a problem in that no statistics were kept by the equipment to show how the health of the link.  ESF was developed to help solve this. 

The framing bits are used in this fashion

Frame Number	Purpose of framing bit
4, 8, 12, 16, 20, 24	Traditional framing bits used to locate and keep track of the beginning of frames.  Bits have this repeating pattern:  001011
2, 6, 10, 14, 18, 22	CRC error check. It checks a block of 24 frames.  The CRC remainder is actually sent with the next set of 24 frames.  Error detection is only used to monitor the health of the link.
Odd numbered frames	Facility Data Link.  This is actually a communication channel that the equipment uses to share its statistics.  Performance reports are sent on this low data rate connection.

 The main benefits of ESF include:

• Don't have to stop the customer's data transmissions to run tests
• Can collect performance information
• Improved reliability

Complete hw 8 according to the schedule.

Used with permission of the Author; Copyright (C) Kevin A. Shaffer 1998 - 2018, all rights reserved.