This document runs through some information transmission basics, it is worth referring also to
which details various methods of bit encoding over carrier signals.
Communication over large distances have led the human race to develop ingenious methods. Fire beacons have been used
for centuries whether in 800 BC by the ancient Greeks, 150BC as coding stations by the Greeks or
by the British in 1588 as a warning of the approaching Spanish Armada. The tom tom (Polynesia), hitting stones (Northern Spain),
Whistling (La Gomera) are all examples of early communication methods.
The French began using Semaphore (movement-based) and the Heliograph (reflected light)
in 1791 and in the 19th and 20th centuries, the signal lamp and morse code was developed to provided
more sophisticated messaging. Much of the early communication methods were dependent on line of sight
or weather conditions.
Electrical telegraphy was invented in 1837 and then Wheatstone, in 1840, came up with a means to represent letters and figures.
Graham Bell in 1876 invented the telephone - a complementary method of communication.
Things then started to move fast with telegraph services, telegram and dialled telex.
Although weather was no longer such a limiting factor, all of these technologies were speed limited to typing speed.
The Move from Analogue to Digital Transmission
Analogue transmission carries information signals on a continuously varying wave.
These may be transmitted via air, water or cable. There are two main types of analogue transmission,
both based on how to modulate data to combine an input signal with a carrier signal. Usually, this carrier
signal is a specific frequency, and data is transmitted through its variations. The two techniques are
Amplitude Modulation (AM), which varies the amplitude of the carrier signal, and Frequency Modulation (FM),
which modulates the frequency of the carrier signal.
Analogue transmission has been the traditional way to convey voice, data and video. If the information was destined
for digital devices such as computers
then modulator-demodulators (modems) were required to extract the digital information from the analogue signal.
In situations where a signal often has high signal-to-noise ratio and is unable to achieve source linearity,
or for long distance, high output systems, analogue signalling becomes a problem due to attenuation problems.
When the signal traverses a communications path (cable, air etc.), the condition of the signal can rapidly deteriorate due to
losses created by the medium. Usually the longer the propagation distance, the greater the distortion.
A cable has resistance, capacitance and induction both within itself and from EMF around it. These
all add to cable impedance. Be aware that a loss of 3dB is equivalent to losing half of the power and
look at these typical cable attenuation figures:
- Coaxial cable (50Ω)
- 1.6dB per 10m at 100MHz
- 5.3dB per 10m at 1000MHz.
- Low loss air spaced (75Ω) coaxial cable which is used for TV, (utilises 5 cell air spaced polyethylene insulation)
- 0.78dB per 10m at 100MHz
- 2.44dB per 10m at 860MHz
- Satellite TV coaxial cable (5 cell semi-air spaced polyethylene insulation)
- 1.9dB per 100m at 10MHz
- 6.1dB per 100m at 100MHz
- 20.0dB per 100m at 1000MHz
As a wave signal using modulation techniques
traverses a medium it deteriorates and there is noticeable signal distortion. This can
manifest itself by white noise from interference or heat, and can be seen as distorted pictures or hissing.
This deterioration is difficult to rectify without complex electronics. Simple amplification just
amplifies the noise that has been introduced. If digital encoding
techniques have been used on this wave signal then even if there is substantial noise, provided that
the encoded binary is recognisable, then it can be reproduced with 100% accuracy.
The benefits associated with digital are as follows:
- Initially the USA telephony system operated at +80 and -80 volts levels which, for accurate transmission, was
difficult to maintain over long distances, systems now use lower power standard 5V TTL logic levels.
- The state difference can be set at a level to suit the requirement, or within the technology available.
- Provided the two high and low thresholds can be determined, then the form of the pulse can be whatever you want
and transmitted on whatever medium that you want.
- It can be readily reproduced from information of the two state conditions.
- Data can be passed quickly, determined by the ability to recognise the two state conditions.
Rather than use frequency and amplitude modulation on wave signals, digital modulation techniques are used.
Refer to Digitisation of Voice
for detail on how voice is digitised and
digital modulation methods such as Pulse Code Modulation (PCM).
The speed of a transmission path may be expressed in three different ways:
- Modulation Rate - the rate in which the circuit changes in a given
time e.g. a wave signal may take 25ms
to complete one state change i.e. T = 0.025s. This means that in 1 second there are
1 / 0.025 = 40 changes, or 40 baud.
- Data Signalling Rate - the rate at which information can be transmitted which is expressed in bits/second. The formula
used to determine this is [1/T]log2n where n is the number of signalling conditions in a given cycle e.g.
if there are two possible conditions within a wave and the wave takes 25ms to cycle, then the data rate is given by
[1/0.025]log22 = 40 * 1 = 40 bits/sec
- Data Rate - the rate at which data actually arrives to its destination. The CCITT define it as 'the average
number of bits, characters or blocks per unit time passing between corresponding equipment in a data transmission system'.
The number of bits does not correspond to the amount of data being transmitted and received by the data sink.
This is due to framing.
Note that for coding systems designed to offer more than 2 signalling conditions within a wave cycle, the data rate
is greater for that wave e.g. if there are four possible conditions within a wave and the wave takes 25ms to cycle, then
the data rate is given by [1/0.025]log2
4 = 40 * 2 = 80 bits/sec for the same 40 baud.
(see Data Encoding
for examples of data coding systems).
Another consideration is that depending on the transport system, a
varying number of bits are used for framing, and these bits have nothing to do with the actual data bits
being transmitted or received by the data sink (receiving station).
An example is the modem.
If a modem operates asynchronously in stop/start mode
at 9,600 bits per second i.e. 1 bit is used to delimit the start of a block of 8 bits of data
and 1 bit is used to delimit the end of the block of 8 bits of data, then the amount of data
received by the data sink will be 8 bits out of every 10. For our example, instead
of 9,600 bits of data being received it will be 0.8 * 9600 = 7,680 bits in one second.
As the complexity of transmission systems increases, so does the framing of the data.
The actual Data Transfer Rate (DTR) is given by the formula DTR = N/T
where N is the number of bits received by the data sink
and T is the time taken to transmit those 'N' bits of data.
This is assuming that that the data sink has received no errors and has requested resends.
In the analogue world Bandwidth is defined as the range of frequencies a bandwidth channel
is able to carry. This range has no frequencies below the low cutoff frequency and no
frequencies above the high cutoff frequency. Electronic filters provide the
means of isolating the required frequencies within the bandwidth range.
In reality a particular signal is composed of a range of frequencies (Harmonics)
that can be identified with Fourier Analysis. If a particular frequency or range of frequencies
is used to carry data bits then it is important to ensure that other frequencies do not
impinge on the carrier signal frequencies.
How do we work out what Digital Bandwidth we have available on
a given signal?
If we take a simple square wave representing voltage over time and this
wave completes one cycle every 1/25th of a second, then
this signal can run at a maximum frequency of 25Hz. It can also run at 0Hz if no state
changes occur over time e.g. perhaps to indicate a series of '0' bits.
In the case of alternating '1's and '0's we have a maximum frequency of 25Hz where
there are two state changes in the signal, we therefore have a 50 baud signal. If each
state change represents one bit of information ('1' or '0') then we have a signal of
frequency 25Hz carrying 50bps (bits per second).
In the above example we have two possible states given by high voltage and low voltage.
There is nothing to stop us having multiple voltage levels. For instance,
we could represent '10' with 5v, '11' with 10v, '01' with -5v, '00' with -10v.
We now are able to represent 4 bits of information in any given cycle which gives
us a digital bandwidth of 100bps.
Transmission can take a number of forms. Simplex Transmission is where the data flows in only one direction
from source to sink e.g. logging; Half-Duplex Transmission where the data can flow in both directions but
not at the same time; Duplex Transmission where the data can flow in both directions at the same time.
When transmitting bits the bit patterns need to be understood by both sender and receiver. Both ends
need to know when the actual bit stream begins and ends. Also, during the bit stream both ends
need to know where they are within the bit stream itself, in other words there has to be some form
of synchronisation between the sender and receiver. This synchronisation can take a number
- Bit (Clock) Synchronisation - the start of each bit cell
- Byte (Character) Synchronisation - the start and end of each character or byte
- Frame (Block) Synchronisation - the start and end of each complete message block
If the sender and receiver's clocks are the same i.e. dependent on each other, then the transmission is said to
be Synchronous and the receiver keeps in step with the clock that it receives as it accepts the continuous
bitstream. If the clocks are independent of each other then the transmission is Asynchronous
and the receiver re-synchronises at the beginning of each byte or block that it receives.
This uses mainly PSTN dialup lines of varying quality and modems
to provide digital connections. Circuit switching is used to provide a link for a particular call
and levels of quality line types can be offered by the telephone companies e.g. Line Type 1 - Basic
voice, Line Type 5 - Basic data, Line Type 7 - voice and data over private lines. In addition,
you can obtain conditioned lines with improved communications. There is D Conditioning
and levels 1 to 8 of C Conditioning. A more expensive dedicated analogue line can be
bought where the circuit is fixed and is not different every time that you dial up, thereby
giving you a more consistent service.
Circuit Switched Path
When making an analogue phone call you first obtain a dial tone, then you dial a number.
This number is sent to the local switch containing a D-Channel Bank using touch tone Dual Tone
Multifrequency (DTMF) signals. That is as far as DTMF gets. The voice call is converted
to the digital Pulse Code Modulation format and analogue signalling to digital signalling by the
D-channel bank. The switch routes the call
from this point through the digital switch network using the Management (M)-Plane protocol
called Signalling System Number 7 (SS7) which is a form of CCS. SS7 sends messages to the switch
which is connected to the destination phone and this far end switch sends a Control (C)-Plane
message that rings the far end phone. When the phone is picked up the C-Plane mechanisms
send the message that the path is available. The digitised voice is the User (U)-Plane
The Modern Telco digitizes speech using Pulse Code Modulation (PCM) on 64K (DS0) channels.
64 Kbps is considered to be Digital Signal Level 0.
Each channel is sampled 8000 times/second according to Nyquist's Theorem,
and incorporates 8 bits per sample (hence 8bits x 8000 giving 64,000 bits/sec).
This figure of 8000 comes from the fact that the valuable range of telephone signals is
100Hz to 4kHz, and the sampling rate is twice that of the highest signal.
The standard G.711 defines the Pulse Code Modulation (PCM) 64Kb/s voice channel.
DS0 trunks make up the trunks around the digital network that can carry data or voice.
For voice the conversion to 2-wire analogue occurs at the switch closest to the user.
The call handling in the 'external' network is dealt with by SS7.
Straight digital signals (bipolar) are used across these lines so no modem is required. A
Channel Service Unit/Data Service Unit (CSU/DSU) provides the interface for the end user and
converts the DTE's digital signals into the Synchronous digital signals used over the WAN.
Kilostream services (BT's version of Digital Data Services) offer from 2.4Kb/s to 64Kb/s (56Kb/s
in the USA) whilst Kilostream N allows multiples of between 2 times and 16 times 64Kb/s to
give a range of bandwidths from 128Kb/s to 1024Kb/s. BT's Megastream offers 2, 8, 34, 45,
140, 155Mb/s bandwidths for really bandwidth intensive traffic.
For the DS-1, also called T1,
Time Division Multiplexing (TDM) is used to transport multiple channels over one line.
Clocking of the serial transmission needs to occur at one end of the link or the other, sometimes you will
see the clocking options as internal i.e. provided by the local device, or line meaning
that the clock is provided by the remote device.
Two-pairs are used in a T1 link. The T1 link can operate in full-duplex mode where one pair transmits
and the other pair receives. 24 channels are available for transmission and these are grouped together
to form a Frame i.e. the 24 time slots (8 bits each) plus one framing bit form one T1 frame
(193 bits, the 193rd bit being the synchronisation/framing bit).
For 8000 samples a second, a T1 frame must be transmitted every 125 usecs, we can therefore
calculate the T1 line rate as 193 x 8000 = 1.544 Mbps (A DS0 line rate is 8 bits x 8000 = 64 Kbps).
The frames can also be grouped into 12 sequenced frames to form a Superframe (SF)
(also called a D4) which means that 12 framing bits are used per SF. These 12 framing bits
are also called F bits. They
form the sequence 100011011100 and are used to sequence the SF within 4 frames. In one
second 8000 'F' bits are used for framing. This is encapsulated in the G.704 framing standard.
A D4 contains 288 channels.
The frames could also be grouped into 24 to form the newer framing format called
the Extended Superframe (ESF). The 8000 'F' bits are used differently in ESF where 2000 'F'
bits are used for framing, 2000 are used for CRC-6 error checking and 4000 are used as a supervisory
channel for things such as loopback and error reporting. An ESF contains 576 channels.
Channel Associated Signalling (CAS)
T1 signalling can take the form of CAS using Robbed Bit Signalling where bits are 'robbed' from
the channels carrying the voice. This is called In-band Signalling. In the SF, the LSB is
'robbed' from each of the 24 x 8-bit timeslots in the 6th and the 12th frames. The A bit comes
from the 6th frame timeslots whereas the B bit comes from the 12th frame timeslots.
These 'robbed' bits are used for call supervision and trunk signalling in the voice environment e.g.
the 'A' bit is commonly used in the same way that the 'M' lead is used in E&M signalling i.e. signalling
by pulsing the 'A' bit. This Bit Robbing is fine if the channels are used for voice because the 8 bit
samples being reduced to 7 bits every 6 frames does not significantly impact on voice quality.
Data is of course not so forgiving with the lowered quality line so each channel is reduced to 56kbps
for data (In the US a type of ISDN called Switched Services uses bit-robbing technology that
results in a 56kbps B-channel). The problem with using CAS is that these robbed bits are really only
used when setting up and establishing a call, the rest of the time the bandwidth is wasted. The only
messages used are Wink, Ringing, Hang up and Pulse Digit Dialling.
The ESF operates a similar manner to the SF other than bits are robbed from the 18th frame (C bits) and the 24th frame
The main difference between channelised lines (analogue) and non-channelised lines (ISDN) is that they
do not have a built-in D-channel. For example, all 24 channels on a T1 line only carry data. The
signalling is in-band or associated to the data channels (Channel Associated Signalling (CAS)).
Traditional channelised lines do not support digitized data calls (for example, BRI with 2B+D).
Channelised lines support a variety of in-band signal types, such as ground start, loop start, wink start, immediate start, E&M and R2.
Common Channel Signalling (CCS)
T1 signalling can also take the form of CCS which is normally Common Channel Signalling Number 7
(SS7) or Primary Rate
ISDN where one channel (D-channel, channel 24) is used for Q.931 signalling. This is called
Out-of-band Signalling since the signalling is in a channel that is separate from the voice
channels. This speeds up call setup by up to a factor of 5, to 1-3 seconds. One signalling channel
can handle up to 1500 calls. SS7 is a protocol in its own right, very akin to X.25 where switches
exchange billing, switching and signalling information.
With CCS, PRI does not operate Bit Robbing but takes one of the channels and uses that for signalling
(D-channel) instead leaving 23 channels for the data. The line encoding coding scheme used to allow both data and
voice is ususally based on a pseudo-ternary bipolar code called Bipolar with 8-Zeros Substitution
(B8ZS). This is called Clear Channel. Another coding scheme called B7 exists for
voice only applications and yet another called Alternate Mark Inversion (AMI) is commonly used.
The European standard E1 is another way of using TDM to transport multiple channels over a single
line. The E1 interface has 32 channels or time slots. Framing is carried out in time slot 17,
the 32 x 8-bit slots are grouped into a 256 bit Frame and then 16 frames are grouped into a
Multiframe. Time slot 17 (channel 16) in the first frame of the Multiframe indicates the
beginning of the Multiframe. In the first frame there are the 4 signalling bits (ABCD) for channel 1
and 4 bits (ABCD) for channel 17. In the second frame there are 4 signalling bits (ABCD) for channel 2
and 4 bits (ABCD) for channel 18. This carries on 16 times to form the Multiframe. G.704 covers this
framing for E1 (as well as T1).
The E1 digital line operates at 2048Kb/s and is made up of 32 x 64Kb/s. Often, the E1 is broken up into 64Kb/s channels
and is known as Fractional E1 (or F-T1 in the States) if the customer is presented with a group, for example a 384Kb/s circuit
(made up of 6 x 64Kb/s channels). In this instance, the carrier determines the
grouping of the channels, one E1 may server a number of different clients.
A channelised T1/E1 (CT1/CE1) line is an analogue line that was originally intended to support
analogue voice calls, but has evolved to support analogue data calls. ISDN does not transmit across channelised T1/E1 lines.
Alternate Mark Inversion (AMI) can be used with E1 but the most common
E1 line-encoding scheme used is called High-Density Bipolar with 3-zeros (HDB3). The error checking used is called
Cyclic Redundancy Check with level 4 checking (CRC-4), although this can be turned off (no-CRC4) with some providers.
Australia has a different way of E1 framing from the rest of the world.
Channel Associated Signalling (CAS)
In CAS, Time slot 17 (E0 channel 16) is used for signalling and time slot 1 (E0 channel 0) is used for
framing synchronisation and alarms, the other 30 are used for voice and data.
The CAS signalling is considered in-band and is very simple as it just identifies four states:
- 00 - Idle
- 01 - Seizure
- 10 - Disconnect
- 11 - Busy
Common Channel Signalling (CCS)
E1 can also be set up for CCS where channel 16 carries signalling such as Q.931 for Primary Rate ISDN (I.421).
This signalling, based on HDLC type protocols can often be vendor proprietory signalling that needs to be transparently passed
by network equipment. Examples include BT's Digital Private Network Signalling System (DPNSS),
Nortel's Meridian Customer Defined Networking (MCDN),
QSIG and Signalling System 7 (SS7) which is a standard for CO to CO signalling used throughout the
One channel can carry a different data signal from another and therefore allows multiplexing to occur
to give 32 simultaneous data transmissions. The 64Kb/s data rate is known as
Digital Signal Level 0 (DS-0) and the 1.544Mb/s rate
is known as DS-1 (T1), the following table details some of the transmission rates:
||No. of T1 channels
||No. of Voice channels
||Data Rate (Mb/s
Multiplexers combines several channels into a single bit stream.
Synchronous data transmission differs from Asynchronous transmission in that data is sent
as a continuous stream of data packets separated by start and stop bits conforming to precise clocking
which is governed either by one end of a link, or the other.
Asynchronous transmission has no central precise clocking and packets can come and go at
undefined times and in different orders with different frequencies. Each end of an Asynchronous link is
responsible for its own clocking.
For interest sake, Isynchronous transmission is when asynchronous data is sent over
synchronous transmission. Plesiochronous transmission is when digital signals with different,
but reliable, clock signals are being used.
Up until this point we have looked at Circuit Switched Networks, however, there are
also Packet Switched Networks where data is split into individual packets and each packet
is switched separately through the WAN. These means that packets can end up at the destination in a
different order from when they left. For this reason, each packet is tagged with a destination
address and order number. Often called Any-to-Any connections, Packet switched networks
are only switching small packets, so resends are fast and the whole switching process is a fast one.
Virtual circuits are set up to provide either a temporary path Switched Virtual Circuit (SVC)
or a Permanent path Permanent Virtual Circuit (PVC).
Broadband ISDN (B-ISDN)
The key component of
B-ISDN technology is the Asynchronous Transfer Mode (ATM)
of transmission. ATM was adopted
as a switching technology for twisted pair wires and other LAN transmission media such as optical.
At one point the use of the synchronous
optical network SDH (SONET)
was initially advocated for B-ISDN due to the low bit error rates
and high bandwidth potential of optical transmission.
Sometimes there’s a bit of confusion between ATM
and Broadband ISDN (B-ISDN)
. The two are
related because ATM evolved from the standardisation efforts for B-ISDN. The fact
is that ATM is the technology upon which B-ISDN is based.
Usually the term 'B-ISDN' is applied to wide area carrier services. Even though the technology
foundation is ATM. The term B-ISDN is not usually applied to local area or campus networks.
The QSIG CCS protocol is significant because it is based on Q.931 and Q.933 and is being developed to cover the first
three layers of the OSI model with layer 3 being the messaging layer. The messaging facilities are more complex allowing
for features such as Call Forwarding, CTI and ACD to be extended across the digital trunks between different
Signalling System 7 (SS7)
SS7 is a form of CCS called CCS number 7 and has been defined by the ITU as ITU-T#7.
SS7 uses the signalling channel on T1 (24) or E1 (17)
to carry signalling, billing and switching information between SS7 capable switches. This is carried
out-of-band in parallel to the calls.
Terminals and Modems
Various types of Terminal access over the Wide Area network use different
methods to mimic being directly connected to the host:
- Telnet/rlogin - provides a virtual terminal connection to a computer. The
protocol rlogin is specific to Unix systems.
- Local-area Transport (LAT) - is provided by DEC to allow terminals to connect
to a number of hosts at one time.
- TCP/IP Telnet 3270 (TN3270) - is the virtual terminal protocol that allows
one to access 3270 applications.
- X.25 Packet Assembler/Disassembler (PAD) - The PAD translates the character-based
terminal output into X.25 packets that can be switched on the Wide Area Network. X.25
has versatility due to its addressing schemes.
Dial-in Modems are used when there are Asynchronous connections to the Wide Area Network.
Protocols that run over the asynchronous connections include Serial Line Protocol (SLIP),
Point to Point Protocol (PPP) and AppleTalk Remote Access Protocol (ARAP).
Traditionally, modems converted digital signals to analogue
to traverse the local loop where they are converted to digital as they are routed across
the network before being converted back into analogue for the local loop at the remote end
where the modem at the remote client converted from analogue to digital again. The theoretical
speed limit for data throughput using this scenario is 33kbps.
With the V.90 56kbps standard the the digital-to-analogue conversion still occurs
at the client end but instead of converting back to analogue at the remote end, say
at an ISP, the signal remains digital and can do so provided that the service provider
has a digital link to the telco. This means that although the data speed in one direction
from the client to the ISP is still limited to a maximum of 33kbps, the data speed from
the ISP to the client (download) can theoretically reach 56kbps. If both ends have to convert
back to analogue signalling then the maximum speed in either direction will only be
33kbps, for example modem to modem connection between two private individuals.
The rate between a computer and a modem can be up to 57.6 kbps for a V.32bisPlus modem
or 115.2 kbps for a V.34 modem. These higher speeds are achieved using V.42bis compression
(level 5 of the Microcom Networking Protocol (MNP)).
Modems commonly monitor the line quality and apply
Adaptive Speed Levelling (ASL) and lower the speed if the line quality deteriorates
thereby keeping the modem connection open.