[Back to Lecture Notes page]

Some Example Systems at the Physical Layer

Sub-topic Outline:

• Conventional Telephone System
  • Modulation - converting analog to digital and vice-versa
  • Multiplexing - having multiple channels on one signal
• T and E Carriers
• Integrated Services Digital Network (ISDN)
• Asynchronous Transfer Mode (ATM) networks
• Communication Satellites

The Telephone System

• For computer networks, more cost effective to make use of existing transmission media than install new lines.
• Make use of Public Switched Telephone Network (PSTN)
  • Eventhough PSTN doesn't have the performance we would like
  • Normal digital computer
    • transmission at 10⁷ to 10⁸ bps
    • error rate about 1 in 10¹² to 10¹³ bits
  • Phone line transmission
    • Max 10⁴ bps
    • error rate about 1 in 10⁵ bits

The Structure of the Telephone System

Fig 2-15 Tanenbaum textbook p105

The telephone system consist of three main components:

• The Local Loop which connects up homes/offices to end offices. End offices usually cover on 1 to 10 km. Local loops are usually cabled using twisted pair, and signals are analog signals.>
• Trunks which connects the end offices to toll offices, and connect toll offices to switching offices, They are usually fiber optics or microwave, and signals are usually in digital.
• Switching Offices which are the main switching centers in the telephone system.

Analog-Digital Conversion on the Local Loop

• Most local loops still use analog signals, but computers, and most major trunks are digital. - so we need to convert the the signals from digital to analog and back to digital for communications throught the telephone system.
• To do the conversion, we use:
  • A Modem (modulator/demodulator) at the user end - which converts the digital signals of a computer to analog signals that can be sent from the home/office onto the local loop.
  • A Codec (Coder/decoder) at the end offices - which converts the analog signal on the local loop to digital signals to be transmitted through the major trunks.

Modulation Techniques

Modulation techniques refers to the way we convert an digital signal to an analog wave. Converting the analog wave back to digital is called demodulation.

Fig 2-18 Tanenbaum textbook p110

In the telephone system, we analog signal comes in the form of a tone over the telephone line. To use that tone to (which can be represented by a sine wave) to send our digital signal, we can use one of 3 ways:

1. Amplitude modulation - we change the amplitude of the wave, which means we vary the voltage of the line.
2. Frequency modulation - we change the frequency of the wave, which means we use different tones.
3. Phase modulation - we ofset the original wave by a certain amount.

Modulation on Standard Modems

• The modulation techniques are negotiated between the modem and the end office at connection time - Negotiation happens when the modem connects to the telephone service
• Modulation techniques are usually a combination of all amplitude, frequency and phase - but the limit for frequncy have been reached (3000 Hz) for conventional analog telephone lines, so no point going higher.
• Development for modem standards are now for new combinations of amplitude and phase modulation, and for better compression and error correction.
• Some past ITU standards:
  • V.32, V.32 bis, V.34, MNP5, V.42 bis
• Latest standards:
  • V.90 at 56K bps

  Before 56Kbps modems, the believe limit of modems were 33.6Kbps. The 56Kbps rate was achieved by requiring direct digital connection from ISP to phone company. Previously there was an extra set of modulation/demodulation when the ISP needs to do the digital-analog-digital conversion as well because the ISP was using a modem. So for an ISP to support V.90 modems, they must have direct digital lines to the telephone system - most do.

  Note that the above are maximum possible rates. What rate you actually get using these modems depend more than the conditions of the telephone lines, network traffic, etc.

Multiplexing

• Multiplexing means to have multiple conversations over the same line - ie. to have different channels

  • more cost-effective to install high-bandwidth lines (ie. fiber optics) than low bandwidth lines - eventhough the cost of fiber optic cables is higher than copper wire cables, most of the expenses is actually in the installation process and not medium. So it becomes more expensive to install a large amount of copper wire cables than to install a few fiber optic cables.
  • with such high-bandwidth lines, we can have multiple conversations over the same lines
  • multiplexing useful in end offices and toll and switching offices, where multiple lower bandwith lines comes together and needs to be transmitted out in higher bandwidth lines.

• Two types of multiplexing:

  1. Frequency Division Multiplexing (FDM), and a variation called Wavelength Division Multiplexing (WDM)
  2. Time Division Multiplexing (TDM)

Frequency Division Multiplexing

Fig 2.24 Tanenbaum textbook p119.

• Each channel allocated a frequency band

  • we put the waves through a filter so that the resulting waves are in the given frequency band
  • standard about 3000Hz per channel

• Interchannel bands

  • separate the different channels by having interchannel bands which no one should use
  • standard about 500Hz above and below each channel

• Multiplex all the channels together

  • combine all the filtered waves
  • should have very little overlap because of the interchannel bands
• Receiver de-multiplex the signal - receiver must know which band belongs to to which channel

Wavelength Division Multiplexing

Fig 2.25 Tanenbaum textbook p120.

• A variation of FDM for Fiber Optics
• the concept is the same, but instead of frequency of sound, we use combine and extract different frequencies (or wavelength) of light
• uses prisms to do the combining and extracting
• remember that wavelength is related to frequencies (see previous lectures).

Time Division Multiplexing

• FDM uses analog concepts, and requires analog circuitry - TDM does not.
• We use TDM for sending multiple channels over digital lines - so that circuits at sender and receiver (and all intermediates) do not need analog electronics.

• Basically,

  • split the transmission into time intervals
  • for each interval transmit the bits for a single channel
  • different channels will transmit at their allocated interval slots
• Analog signals can be sampled (measured at regular intervals) and converted to 7 or 8 bit numbers using Pulse Code Modulation (PCM)

Pulse Code Modulation

Basically,

• we have a wave which has an amplitude value at different times
• we sample (ie. detect the values) at regular interval, and represent the sampled values as a binary numbers in 7 or 8 bits
- therefore we can have 2⁷ (128) or 2⁸ (256) different values
• we then transmit that sequence of numbers
• for the bandwidth of normal telephone lines (4KHz), there is no point detecting at faster than 125 microsec intervals - so time intervals are usually split in 125 microsec blocks

All modern telephone systems uses PCM to convert to analog sound signals to digital.

My description in the final part of the lecture notes on signals, with the title "An Example of Sending Digital Signals through an Analog Medium", is based on PCM.

The T1 Carrier

• A protocol based on TDM.

• Wide spread in North America and Japan - not commonly used outside.

  We use T1 to refer to the multiplexing method, as well as the hysical network (usually in fiber optics) which carries the multiplexed data. Some literature only refer to T1 only as the carrier.

Fig 2.26 Tanenbaum textbook p122.

• uses PCM and TDM to put 24 voice channels together - can transmit at 1.544 Mbps
  • sample all signals from all 24 channels at during one 125 microsec time slot - get 24 8-bit numbers (ie 24x8=192 bits).
  • add one bit for synchronization to make 193 bits.
  • transmit 193 bits at that 125 microsec time slot (at that rate, we get 1.544 Mbits in one second)

• The synchronization bit

  • In two successive frames of 193 bits (between two 125 microsec slots), the first bit is always for synchronization
  • The pattern for the first bit is 01010101... (repeated pattern of 01's). This means the first frame will have 0 as the first bit, the second frame will have 1 as the first bit, the third frame will have 0 as the first bit, etc.
  • The receiver continually check this first bit to make sure the bits stream is in synch. If the bit all of a sudden no longer follows the 010101... patterns, the receiver knows the stream is no longer in synch (perhaps some bits have been lost somewhere)

An important note here about a change: up till now, we have talked about transmission media and their properties in transmitting signals. They offer services (ie. transmitting signals) used by other higher level services. T1 (and later on E1, T2, ATM, etc) involves higher level definitions (eg. number of channels, synchronisation of bits, etc) than just sending/receiving signals, and so are not completely dependant on the transmission media. So for example we can built a T1 carrier using any media, as long as we can transmit 24 channels in 125 microsec, etc. Obviously if we want the connection to be over a reasonable distance, not all media will be able to do this, but the point is still that T1 DOESN'T define you must use fiber optics. Even though we are still talking about physical layer services this is now a higher level protocol. The main point here is that although we still put the protocol T1 in the physical layer, it is one level higher than the physical medium.

The E1 Carrier

• Used widely outside of North America and Japan - not used much inside N America and Japan.
• In T1, for each of the 8-bit numbers, only 7 bits represent the values we sample. The 8th is used for control.
• CCITT came up with the definition for E1, which uses all 8 bits for sampled value (so we can have 256 possible values instead of 128) - use separate channels for control (rather than 1 bit in every channel as in T1).

• In E1,

  • we have 32 channels instead of 224 in one 125 microsec slot
  • 30 channels are sampled values, 2 for control.
  • Using this, we can get 2.048 Mbps

Higher-Order Carriers

• Putting mutiple carrier lines to be put together into one higher bandwidth line using TDM.
• One possible way is using the method shown in Fig 2-28 of the textbook
• The T series of carriers (T1, T2, T3) moves up in ratios of 4, 6 and 7.
• CCITT for the E series (E1, E2, E3, etc) moves up in ratios of 4 for each step (E1 = 32 channels, E2 = 128 channels, ...)

Switching

We have discussed what happens when a signal goes from digital-analog-digital (modulation), and how to put multiple signals into a single high-bandwidth line (multiplexing). But how do we actually get a connection from one end of a network to another. This involves knowing what happens to the data at each node in the network.

• Circuit Switching vs Packet Switching

  • Fig 2-34
  • Fig 2-36

In circuit switching, when a connection (eg. telephone call) is made, a physical path is found from the one end of the connection to the other. This physical 'path' is allocated to this connection, and no one else can make use of. This doesn't mean you can't have more than one physical path going through a single real link (eg. like a copper wire). A few physical paths can go through a single copper wire by using multiplexing.

In packet switching, there are no allocated paths. A single connection (eg. exchanging text messages on a chat line) involves having all transmission broken up into blocks, and each block sent to the other side independently. How each block is routed is up to the individual nodes which receives it. Internet communications occurs by packet switching. We will study routing when we get to the topic on the Network Layer.

Message Switching is a variation of packet switching, where a whole message (could be voice) is transmitted as a block, and not broken up into smaller pieces. It is easier to synchronize, but is not as flexible and efficient in terms of bandwidth.

Note that 2-34(a) is simplified because we might have more than one circuit going through a link (through multiplexing)

Integrated Services Digital Network (ISDN)

• A system developed by phone and telegraph companies under CCITT

  • to have a fully digital circuit-switched telephone network
  • to replace the current analog-digital mixed network
  • to integrate voice and non-voice services
  • voice will still be the primary service, but with more

Example Services ISDN wants to Support

• Instant calls through single buttons
• Transmitting sender info (phone number, name, address etc) with call connect - may be displayed on the receivers phone display.
• Connect phones to databases
• Call forwarding
• World-wide conference calls
• Automated alarm calls to emergency services.
• etc.

The ISDN Architecture

Fig 2-41

• The digital bit pipe - a "conceptual" pipe between the ISDN carrier and the customer, for bits to flow
• NT1 - the main "hub" for all ISDN equipment on a customer's site - has a limited number of connectors to it.
• NT2 (also called a Private Branch eXchange, PBX) - a switch in the large customer's site which involves multiple ISDN equipment
• T, S, R, U Reference points

What the ISDN Bit Pipe Can Support

• ISDN defines the following types of channels can can go through it's "conceptual" bit pipe:

  A - 4KHz analog telephone
  B - 64Kbps digital PCM channel
  C - 8 or 16 Kbps digital channel
  D - 16 Kbps digital channel for out-of-band signalling
  E - 64 Kbps digital channel for internal ISDN signalling
  H1 - 384, 1536 and 1920 Kbps digital channel
  

  H2 - 34 and 54 Mbps
  H4 - 135 Mbps
  

• Different bit pipes support different combinations of the above channels.
• Eg. Fig 2-42
• Developers of early ISDN mostly focused on the 64Kbps B channel (which is why "normal" ISDN is usually called Narrowband ISDN)

Popularity of ISDN

• Narrowband ISDN lacks the necessary rate for most of todays services (like video on demand)
• It is still widely use because it is a faster alternative to 56Kbps modems for Internet access, and at an affordable price (especially for small offices and organizations).

Broadband ISDN and ATM Networks

• As an aletrnative to Narrowband ISDN
• Network at 155Mbps

• Use ATM (Asynchronous Transfer Mode) networks

  • virtual circuits as oppose to circuit switching

    Fig 2-43

    In virtual circuits, we use a packet switch network to 'simulate' circuit switching. When we want a connection, we find a path from one end of the connection to another. We then we ask each node in this path to keep this path in their routing tables, and to reserve some resources (eg. priority in using a channel in one outgoing line) for this connection. Any packets belonging to this connection will be routed the same way, and given priority use of some lines if possible.

  • the cells transmitted are asynchronous as oppose to synchronous (eg T1) - Fig 2-44

    In the TDM in T1 carriers, we have the signals from 24 incoming lines put in one outgoing line. Each of the signals in the incoming line has a particular spot allocated in a particular time block in the outgoing signal. So for a particular 125 microsec block, if no signal was received by some of those incoming lines, their block will remain empty in the outgoing signal. This is called synchronous - all 24 lines are syncronised to be multiplexed and sent out. In asynchronous transmmission, any signal that comes in are transmitted out in their own 'cells' without having a rigid order on which signal belongs where in the multiplexed block.

• Requires large replacement cost for equipment (including physical lines) to implement

Communication Satellites

• the first satellite: the moon
  • US Navy bounced signals off the moon for communications
• artificial satellites
  • signals amplified before sent back to Earth
  • contains transponders
  • each transponder listens for signals on some parts of the electromagnetic spectrum, then rebroadcast back to Earth in another frequency to avoid interfering with the signal - if they broadcast in the same frequency as the incoming signal, the incoming and outgoing signals will interfer with each other.

Geosynchronous Satellites

• Geosynchronous: moving at the same speed as the speed of Earth rotation - so on Earth they appear stationary - very useful for users of a satellite.
• Need to be about 36,000 km above the equator - near the Earth surface, satellites needs to move at a much higher speed to maintain an orbit because the gravitational pull is stronger. At about 36,000 km, the satellites can move at about the same speed as the Earch is rotating.
• International agreements exists for allocating Geosynchronous Satellites orbits (who goes where) and frequencies (what bands on the electromagnetic spectrum they listen and rebroadcast on).
• Because most are very high frequency microwave tranmissions, the signals can be absorbed by rain - solve this by having receiving ground station spread widely
• Typical satellite has 12-20 transponders, each allocated 36-50 MHz bandwidth.
• Transmitting down to ground stations can be wide beams or spot beams
  • wide-beams can cover all parts of Earth visible to the satellite
  • spot beams cover only a few hundred km areas

Very Small Aperture Terminals (VSATs)

Fig 2-56 Tanenbaum textbook p166

• Small, low-cost ground stations - makes use of geosynchronous satellites

  Each VSAT doesn't have enough power to communicate with each other (they only have 1 meter antenna and give out 1 watt of power), but they transmit a 'hub' through a satellite, and the hub amplifies it resends it to the intended VSAT (again using the satellite).

• Low data rate
  • uplink (the data from ground to satellite) is about the rate of 19.2 Kbps
  • downlink (the data from satellite to ground) is about the rate of 512 Kbps (it is so much more because the VSAT are low powered and the satellites are usually a lot more high powered)

Low-Orbit Satellites

• At low altitudes, satellites needs to move at faster speed to maintain orbit - so they go out of range for a ground station very quicky.

Fig. 2-57

In 1990, Motorola developed and launched a system of 77 satellites in a chain so that as one satellite goes out of range, another comes into range to continue the transmission. This was called the Iridium Project (Iridium is the 77th element in the periodic table). This led to a huge amount of such statellite chains being launched by different companies.

Satellite vs Fiber

20 years ago, it looked like satellite communication was the future of telecommunications. Only after the introduction of fiber optics and the broadband (ie. high bandwidth) networks that it looks like satellites are not necessary. Still it is unique since:

1. Fibers only available to where the cables are installed. Satellites (potentially) broadcast everywhere.
2. Some parts of ground-based networks still are very slow (eg. local loops using twisted pairs).
3. It can supply connect mobile users
4. Good for broadcasting, rather than point-to-point.
5. Good for places where laying ground lines are not feasible.

[Back to Lecture Notes page]