Introduction to Communication Networks

Unit 5
Switching Principles

EECS 122 SPRING 2007

Spring 2007

Acknowledgements slides comming from:

Data and Computer Communication by Wiliam Stallings (our

supplementary textbook) numerous slides!

Data Communications and Networking by B. Forouzan, Mc Graw

Hill, 2004

Some figures have been used form the earlier issues of the
EECS 122 tought by Prof Jean Walrand.

Introduction to Telephones & Telephone Systems by A. Michael

Noll, Artech House, 1986

Prof. Adam Wolisz

EECS 122 SPRING 2007

2 of 54

Switching

It is NOT efficient to build a physically separate path for

each pair of communicating end systems (left picture).

There is a set of path sections (e.g. electrical cables) and

switches (right picture).
Prof. Adam Wolisz

EECS 122 SPRING 2007

3 of 54

Circuit Switching Principle

A connection between the ingoing and outgoing segments of the
transmission path is established on demand, for the exclusive use
of a pair of end users - until explicitly released

Prof. Adam Wolisz

EECS 122 SPRING 2007

4 of 54

Circuit Switching features

The whole path is reserved for the single pair of end users inefficient use of the path if there is bursty traffic

The delay in transmission of user data is constant,

determined mainly by the signal propagation

Explicit setup and release of a switch setting is needed, the

intention to setup/release a connection must be conveyed
from the initiator to each switch (signaling)
Time

is needed for both: propagation of the signaling

information and operation of each switch, as well as
propagation of the confirmation of path setting

During the establishing/releasing the path segments can not

be used but later the resources are assured!!!
Perfect for LONG transmissions of flows with pretty
constant bit rate!!!
Prof. Adam Wolisz

EECS 122 SPRING 2007

5 of 54

Circuit Switching

[Fairhurst]

Duplex connectivity
might be set up...
The disconnect might be
Issued by either side!!!
Prof. Adam Wolisz

EECS 122 SPRING 2007

6 of 54

Comparison of Switching Techniques- again...

Burstiness of data

In many data communication applications, data occur in bursts

separated by idle periods
This type of intermittent data can often be transmitted more
economically by assembling the data into packets (or messages) and
interspersing packets from several channels on one physical
communication path

Prof. Adam Wolisz

EECS 122 SPRING 2007

7 of 54

Message Switching Principle!

There does exist a permanent connection between each input and

each output of the switch (like a fully connected matrix or shared
memory).
The user generated data units (messages) have to carry information
uniquely defining the route to be chosen (header)

Prof. Adam Wolisz

EECS 122 SPRING 2007

8 of 54

Message Switching - features

Path capacity is used more efficiently- multiplexing of

different flows (pairs of users) on a single path segment
takes place
Each message must carry a header with routing information.
This header must be processed upon arrival of the message
to the switch.

Usually the store and forward principle is used. Cut-through principle

fairly difficult to implement.

Buffering is needed in the switches in order to avoid

overloading of the output segments - thus variable queuing
delay is enforced in addition to the propagation delay.

The transmission can start immediately

Variable data length makes the control (e.g. memory

management) difficult
Prof. Adam Wolisz

EECS 122 SPRING 2007

9 of 54

Message switching

Prof. Adam Wolisz

[Fairhurst]

EECS 122 SPRING 2007

10 of 54

Serialization

Small message caught behind big message

Prof. Adam Wolisz

EECS 122 SPRING 2007

11 of 54

Packet Switching Principle and features

A similar approach as above, but the length of the
transmitted information is limited to a certain, user
independent block- called the packet.

The above mentioned effect is removed...

Packets have a unique maximal size, making the control (e.g.

memory management, header processing) easier.

There is a pipelining effect, increasing the path utilization

There is an overhead for dividing messages into packets and

putting them together within the end systems

The processing overhead within the switch (e.g. routing) is

pro packet rather than pro message

The information overhead is pro packet rather than pro

message
Prof. Adam Wolisz

EECS 122 SPRING 2007

12 of 54

Packet switching

Prof. Adam Wolisz

[Fairhurst]

EECS 122 SPRING 2007

13 of 54

Delay on the way summary

Prof. Adam Wolisz

[ Source: Dr. Cheng Lehigh Univ.]

EECS 122 SPRING 2007

14 of 54

More features:

Line efficiency (+)

Single node to node link can be shared by many packets over time

Packets queued and transmitted as fast as possible

Data rate conversion (+)

Each station connects to the local node at its own speed

Nodes buffer data if required to equalize rates

Packets are accepted even when network is busy (-)

Delivery may slow down

Priorities can be used per packet

Prof. Adam Wolisz

EECS 122 SPRING 2007

15 of 54

Comparison of switching approaches

(a) Circuit switching

Prof. Adam Wolisz

(b) Message switching

EECS 122 SPRING 2007

[Tannenbaum]

(c) Packet switching

16 of 54

Effect of Packet Size on Transmission Time

Prof. Adam Wolisz

EECS 122 SPRING 2007

17 of 54

Store and forward packet switching

Prof. Adam Wolisz

EECS 122 SPRING 2007

[eecs122, walrand]

18 of 54

Cut-thorugh packet switching

Prof. Adam Wolisz

EECS 122 SPRING 2007

[eecs122, walrand]

19 of 54

Bottleneck...

[eecs122, walrand]

It is possible to estimate the bottelneck trhoughput sending back toback packets of constant length and observing the time difference
between their Arrival times...
Prof. Adam Wolisz

EECS 122 SPRING 2007

20 of 54

Details of Packet Switching Techniques

Packets can be handled in two ways

Datagram

Virtual circuit

DATAGRAMS

Each packet treated independently

Packets can take any practical route

Packets may arrive out of order

Packets may go missing

Up to receiver to re-order packets and recover from missing packets

Prof. Adam Wolisz

EECS 122 SPRING 2007

21 of 54

Datagrams

Simple idea:

what is this? Complete destination address or complete

description of the route (we will discuss this later!)
Different priorities per packet might be used..

Processing of a datagram:

dont set up a connection, just make sure each packet contains

enough information to get it to destination

switch creates a table, mapping destinations to output port (ignores

input ports)
when a packet with a destination address in the table arrives, it
pushes it out on the appropriate output port
when a packet with a destination address not in the table arrives,
something clever has to be done ( a different problem!)

Where does the content of this tables come from? This is

again a separate issue (Routing! _ will discuss it later!)
Prof. Adam Wolisz

EECS 122 SPRING 2007

22 of 54

Tables for datagram processing

Each destination has to be listed in this tables (well at least
the region in which the destination is... This is hierarchy..)

Prof. Adam Wolisz

EECS 122 SPRING 2007

23 of 54

Why Virtual Circuit Packet Switching

The idea is to combine the advantages of circuit switching

with the advantages of datagram switching

Virtual circuit packet switching:

After a small connection setup phase only short (compared to full

addresses) connection identifier are used per packet; this reduces the
addressing overhead per packet
During the setup phase, a table is created stating how to process a
packet with the corresponding connection identifier; this reduces the
per packet processing! very important for high speed links

Prof. Adam Wolisz

EECS 122 SPRING 2007

24 of 54

Event Timing

Prof. Adam Wolisz

EECS 122 SPRING 2007

25 of 54

How das forwarding work in VCs?...

Prof. Adam Wolisz

EECS 122 SPRING 2007

(Garcia, Ch7)

26 of 54

Virtual circuit issues

Good: easy to associate resources with flows

can guarantee buffering and delay, as well as care for Sequencing and
Luck of errors. This makes quality of service guarantees (QoS) easy
to provide
Also good: VCI small, making per-packet overhead small.

Bad: not good in the face of crashes

doesnt handle host crashes well: each connection has state stored
throughout network. to close connection, host must explicitly issue a
tear down.

In general, to survive failure, want to make stuff as stateless as

possible, trivially eliminating any storage management problems.

Doesnt handle switch crashes well: have to teardown and reinitiate a

new circuit

Prof. Adam Wolisz

EECS 122 SPRING 2007

27 of 54

ATM
example of virtual circuit usage

Prof. Adam Wolisz

EECS 122 SPRING 2007

28 of 54

Introduction to ATM

ATM is based on some important concepts:

Usage of small and fixed sized packets simplifies the

processing inside a switch and thus enables very high data
rates (155,52 Mbps and 622,08 Mbps are common; higher
rates are possible)
Two protocol layer relate to ATM functions:

virtual circuits
fixed-size packets or cells
small cell size
statistical multiplexing
integrated services

the ATM layer for all services that provide fixed-size packet transfer
capabilities and
the ATM adaptation layer (AAL) that is service dependent (e.g. not
ATM based protocols)

These concepts build a network that can carry multiple

classes of traffic with quality-of-service guarantees
Prof. Adam Wolisz

EECS 122 SPRING 2007

29 of 54

Extension of the flow identifier: VCCs and VPCs

Virtual circuits are referred to as virtual channel connections

(VCC)
A second sublayer has been introduced: the concept of
virtual path connections (VPC); a VPC is a bundle of VCC that
have the same endpoint; this concept is used to decrease the
control costs (esp. in high speed networks like ATM) for
connections that share common paths

Prof. Adam Wolisz

EECS 122 SPRING 2007

30 of 54

ATM Concepts: Small packets

(RPI)

An ATM Layer packet has 53 bytes, including 5 Bytes header

(8 bit VPI + 16 bit VCI) and 8 bit header error control

At 8KHz, each a byte comes every 125 microseconds

The smaller the cell, the less an endpoint has to wait to fill
it Low packetization delay, but

The smaller the packet, the larger the relative header

overhead

Prof. Adam Wolisz

EECS 122 SPRING 2007

31 of 54

ATM Cell Structure (for later discussion)

Prof. Adam Wolisz

EECS 122 SPRING 2007

[RPI]

32 of 54

ATM Adaptation Layer

The AAL may enhance the service provided by the ATM
Layer to the requirements of a specific service (user,
control, management).

Acts on ATM Layer data streams (mapping for the next

higher layer)

Different requirements of the protocols on top of the

AAL
several AAL protocols are required

AAL protocols are characterised by a common set of

functions

required by several protocols to be run over an ATM network

specific adaptation requirements of protocols (originally

designed for other network types)

Prof. Adam Wolisz

EECS 122 SPRING 2007

33 of 54

ATM Service Categories

(Stallings)

ABR (Available bit rate): Source follows network

feedback. Max throughput with minimum loss.
UBR (Unspecified bit rate): User sends whenever it wants.
No feedback. No guarantee. Cells may be dropped.
Prof. Adam Wolisz

EECS 122 SPRING 2007

34 of 54

Telephone Network
POTS: Plain Old Telephone Network
(a classical Circuit switching Network)

Prof. Adam Wolisz

EECS 122 SPRING 2007

35 of 54

Components of the Telephone Network

A telephone call involves much switching and transmission equipment. Each telephone
is connected to a central office by a pair of wires called the local loop. The first stage
of switching occurs at the serving central office. Calls to another office within the
local access and transport area (LATA) are carried over interoffice trunks. Calls
outside the LATA are handled by interexchange carriers over their own transmission
and switching facilities. The point of presence is the place where the IXC connects to
the facilities of the local exchange carrier.
Prof. Adam Wolisz

EECS 122 SPRING 2007

36 of 54

US structure after 1984 divestiture

Prof. Adam Wolisz

EECS 122 SPRING 2007

37 of 54

Telephone System: analog and digital

Tannenbaum

The local loop is analog.

The toll connecting trunks are recently usually digital
The chain: Analog - digital Analog.
But: Telephony is MUCH more than VOICE Transmission!
Be aware of SIGNALLING
Prof. Adam Wolisz

EECS 122 SPRING 2007

38 of 54

Signaling- Basics

Signaling is needed in networks to control their operation

and to indicate the status.

Subscriber-Loop Signaling
Audible

communication with the subscriber (dial tone, ringing tone,

busy signal and so on).

signal to make the telephone ring

Transmission

of the number dialed to central office

Interoffice Signaling
Transmission

of information between switches (e.g.: setting up a

call, indicating that a call establishment is completed, can not be
completed, or has ended)

Transmission

of information for billing purposes

Transmission

of diagnosis-relevant information

Prof. Adam Wolisz

EECS 122 SPRING 2007

39 of 54

Control Signals (origin-destination)

Prof. Adam Wolisz

EECS 122 SPRING 2007

40 of 54

The Local Loop Signalling

The 48-volt common battery is located at the central office. The circuitry in
the phone instruments draws direct current from the local loop. The flow of
dc over the local loop is sensed by a line relay at the central office. A
transformer connects the local loop to the switching equipment so that only
the ac speech signal continues. The ringer in the telephone instrument is
always connected across the line, and a capacitor prevents direct current
from flowing through it.
The ringing voltage consists of bursts of a pure tone, or sine wave, at a
frequency of 20 Hz and with a rms electromotive force of 75 volts. The
bursts are on for 2 seconds and off for 4 seconds.
Prof. Adam Wolisz

EECS 122 SPRING 2007

41 of 54

Pulse Dialing

Off-Hook

Dialing

Inter-Digit Next Digit

Make
(Circuit Closed)

700 ms

Break
(Circuit Open)
Pulse Period
(100 ms)
Prof. Adam Wolisz

EECS 122 SPRING 2007

US:60/40 Break/Make

42 of 54

Tone Dialing

Dual Tone Multifrequency (DTMF)

697

Prof. Adam Wolisz

1209

1336

1477

1633

770

852

941

EECS 122 SPRING 2007

Timing:
60 ms Break
40 ms Make

43 of 54

Voice Channel Bandwidth

Voice Channel
Output
Voltage
or
Energy

Voice Signal

.2

Tone Dialing
Signals

Systems Control
Signals

Frequency
(K-Hertz)

Signalling on digital trunks is much more complex....

Prof. Adam Wolisz

EECS 122 SPRING 2007

44 of 54

Signaling Variants
Signaling Techniques for Circuit-Switched Networks

Prof. Adam Wolisz

EECS 122 SPRING 2007

45 of 54

In Channel Signaling

Use same channel for signaling and call

Requires no additional transmission facilities

Inband

Uses same frequencies as voice signal (SF,MF,DTMF)

Can go anywhere a voice signal can

Impossible to set up a call on a faulty speech path

Out of band

Voice signals do not use full 4kHz bandwidth

Narrow signal band within 4kHz used for control

Can be sent whether or not voice signals are present

Need extra electronics

Slower signal rate (narrow bandwidth)

Prof. Adam Wolisz

EECS 122 SPRING 2007

46 of 54

Common Channel Signaling

Control signals carried over paths independent of voice

channel

One control signal channel can carry signals for a number

of subscriber channels

Common control channel for these subscriber lines

Associated Mode

Common channel closely tracks inter-switch trunks

Disassociated Mode

Additional nodes (signal transfer points)

Effectively two separate networks

Prof. Adam Wolisz

EECS 122 SPRING 2007

47 of 54

Signaling: In Channel vs. Common Channel

Prof. Adam Wolisz

EECS 122 SPRING 2007

48 of 54

Signaling Modes

Modern signaling runs

over a packet swithced
dissassiciated mode...
Prof. Adam Wolisz

EECS 122 SPRING 2007

49 of 54

Data communication over local loop...

The telephone local loop is the mostly deployed

communication channel: home, office.... etc...

It has always been tempting to use it for data transmission.

Features: Engineered to support transmisison in the spectrum

300Hz 3400 Hz. Strongly Nonlinear Atenuation
characteristics in frequencies beyond that...
Options:

modem: modulate digital data onto analog voice channel

turn local loop completly into digital (ISDN)
TDM access supporting 2B + D channels = (2 *64+ 16) kbits/s duplex
DSL introduce in frequency division a new channel parallel to the
voice channel... Dealing with the Unpleasant attenuation

Prof. Adam Wolisz

EECS 122 SPRING 2007

50 of 54

Modem

[Noll]

Transmission complex multilevel modulation, and more..

Interface between computer and modem ...(RS232,,,)
Control commands Hayes Command Set...
Prof. Adam Wolisz

EECS 122 SPRING 2007

51 of 54

DSL

Prof. Adam Wolisz

[Tannenebaum]

EECS 122 SPRING 2007

52 of 54

DSL general usage of frequencies

Prof. Adam Wolisz

EECS 122 SPRING 2007

[Tannenbaum]

53 of 54

DSL: Modulation adjusted to sub-carrier quality

Prof. Adam Wolisz

EECS 122 SPRING 2007

[Halsall]

54 of 54