CS4450, CS5456

Introduction to Computer
Networks
Lecture 5
Rachee Singh
https://www.racheesingh.com/computernetworks/

1
 Outline of Today’s Lecture

1. What does a network consist of?

2. How is the network shared?
3. How do we evaluate a network’s performance?
4. How is the functionality of the Internet organized?
5. What key architectural principles dictate Internet design?
6. Layer 1: Physical layer

2
 Key tasks of the Internet

1. Naming, addressing: locating the destination

1. Mapping hostnames to network addresses
2. Routing: finding the path to the destination
1. Routers/switches compute paths between hosts on the Internet
3. Forwarding: sending data to the destination
1. Individual switches send packets on local links
4. Reliability: communicating despite failures
1. Handling packet loss, drops

3
 An organization of tasks on the Internet : Layers

Layer 7 (Application) Applications

Layer 4 (Transport) Reliable or unreliable transport

Layer 3 (Network) Best-e ort global packet delivery

Layer 2 (Data Link) Best-e ort local packet delivery

Layer 1 (Physical) Physical transfer of bits

Layers of Internet functionality Tasks that the Internet implements

4
ff
ff
 Layering: Pros and Cons

1. Pros:
1. Modularity
2. Reduce complexity (don’t need to know every task of the network)
3. Improves flexibility (one layer an improve in isolation)
2. Cons:
1. Cross-layer information can be useful
2. Sub-optimal performance

5
 What gets implemented where?

1. Given all this functionality is needed in the network

1. Routing, forwarding, reliable delivery, applications
2. Which components of the network should implement which layers?

6
 What functionality (layers) at the end host?

1. Bits arrive arrive on the wire

Alice
2. Have to make all the way to the application
3. What does this mean for the end host? HTTP

1. Which layers should the end host implement? TCP

2. All of them! IP

Ethernet

Optical
7
 What functionality (layers) on the switches?

1. Bits arrive on the wire —> need the Physical layer (L1)
2. Packets need to be forwarded to next router/switch —> need Data Link layer (L2)
3. Packets need to be routed globally —> need network layer (L3)
4. No reliable delivery on switches —> transport layer is not needed (L4)
1. More on this later

8
 Who implements which layer?
1. End hosts implement all layers
2. Lower 3 layers implemented everywhere

Layer 7 (Application) Layer 7 (Application)

Layer 4 (Transport) Layer 4 (Transport)
Layer 3 (Network) Layer 3 (Network) Layer 3 (Network)
Layer 2 (Data Link) Layer 2 (Data Link) Layer 2 (Data Link)
Layer 1 (Physical) Layer 1 (Physical) Layer 1 (Physical)

End host Switch End host

 Who implements which layer?
1. End hosts implement all layers
2. Lower 3 layers implemented everywhere

Layer 7 (Application) Layer 7 (Application)

Layer 4 (Transport) Layer 4 (Transport)
Layer 3 (Network) Layer 3 (Network) Layer 3 (Network)
Layer 2 (Data Link) Layer 2 (Data Link) Layer 2 (Data Link)
Layer 1 (Physical) Layer 1 (Physical) Layer 1 (Physical)

End host Switch End host

 Layer encapsulation

Alice Carol
HTTP HTTP

TCP TCP

IP IP

Ethernet Ethernet

Optical 10000010101 Optical

11
 Who implements which layer?

Layer 7 (Application) HTTP SMTP DNS {

Only end host

Layer 4 (Transport) TCP UDP

Layer 3 (Network) IP

Layer 2 (Data Link) Ethernet FDDI PPP

{In network

Layer 1 (Physical) Optical Copper Radio

12
Switches: implement physical and data link layers (L1, L2)

Switches End hosts

Links

End hosts

13
 Switches: implement physical and data link layers (L1, L2)
Routers: implement physical, data link and network layers (L1, L2, L3)

Switches End hosts

Links
Routers

End hosts
We will use terms switch/router interchangeably
14
 End-to-end principle

1. Very influential paper called “End-to-end arguments in system design” in 1984

2. Articulates the end-to-end principle
3. What does it say?
1. Some application requirements can only be correctly implemented end-to-end
2. Requirements example: security, reliability
4. Implementing them “in the network” is hard
1. Since each switch should be fool-proof

15
 Why is it hard to implement reliability “in the network”?

1. Each switch has to implement this

1. Even if one fails to, all guarantees break down
2. End hosts have to implement it anyway
3. End host complexity remains the same even if the network implements it
4. Network complexity increases
5. Generality of the network reduces

16
 End-to-end principle
“The function in question can completely and correctly be implemented only
with the knowledge and help of the application at the end points of the
communication system.

Therefore, providing that function as a feature of the communication system

itself is not possible.

Sometimes an incomplete version of the function provided by the

communication system may be useful as a performance enhancement.”

— David Clark
Chief protocol architect of the Internet form 1981
17
 Summary

1. End-to-end design + layering

1. Dictate where should network functionality be implemented (host vs. switch)
2. End-to-end design implies
1. “Smart” end hosts
2. “Dumb” network
3. Benefits of end-to-end design:
1. No application knowledge in the network (generality)
2. Less state in the network (robust to failures)

18
Hourglass structure or “narrow waist” of Internet protocols

Layer 7 (Application) HTTP SMTP DNS

Layer 4 (Transport) TCP UDP

Layer 3 (Network) IP

Layer 2 (Data Link) Ethernet FDDI PPP

Layer 1 (Physical) Optical Copper Radio

19
 Implications of the narrow waist

HTTP SMTP DNS

1. Single protocol (IP) at network layer
2. Allows different types of networks to
TCP UDP
operate on the Internet (interoperability)
1. As long as the network can talk IP IP
3. But! Everything depends on IP
1. Moving from IP to IPv6 is hard Ethernet FDDI PPP

Optical Copper Radio

20
 Fate Sharing
1. What is state?
1. Information about the communication that is happening
2. Examples: keeping track of packet losses at end hosts
2. Fate sharing means:
1. Co-locate state with entities that rely on that state (end hosts)
2. The communication shares fate with the end hosts
3. End-to-end principle relies on the concept of fate sharing
4. Why?
1. If the state is lost due to failure
2. The entity that relies on it has failed
3. .. less harm due to failure
21
 Recap: Architectural Decisions + Principles

1. How to break communication tasks into modules?

1. Layering
2. Where are modules/layers implemented?
1. End-to-end design principle
3. Where is state stored (hosts, switches, both)?
1. Fate-sharing

22
Physical Layer (L1)

23
 Physical Links

1. A link in any network: Router Router

1. A pair of routers

L3, L2 Protocols
2. Logical connection between them

??
3. Physical connection between them

24
 Physical links

A B
PACKET
PACKET
PACKET

25
 Physical connectivity: basics

PACKET
PACKET
PACKET

A B

Physical Medium
0100110011..
0100110011..
0100110011..

Bits in the packets are “encoded” on a signal in the physical medium.

26
 Physical connectivity: basics
PACKET
PACKET
PACKET

A B

0100110011.. λ1
0100110011.. λ2
λ3
0100110011..

Fiber
Fiber (glass) is an efficient (low loss) medium for transmitting signals.
27
 Wired network connectivity: optical fiber

1. Optical spectrum is the range of wavelengths in a fiber

2. Optical spectrum of fiber is in infra-red range:
• Wavelengths above 850nm
3. Why use infra-red signals?
• Lower attenuation (loss of signal) in fiber
4. What causes attenuation? Electromagnetic spectrum
• Scattering of light
• Absorption of light
28
 Wired network connectivity: optical fiber
1570 nm
1. A wavelength (λ) carrying bits on fiber is a unit of
signal
• A portion of the optical spectrum
1520 nm
2. Frequency (f) and wavelength (λ) are used
1 196.1 THz 1570 nm
interchangeably: λ ∝
f

Wavelengths
Spectrum
Frequency
3. Spacing between wavelengths to ensure signals don’t
overlap at the receiver
4. 50GHz space between wavelengths, total 4THz
bandwidth means 80 wavelengths on fiber (4000/50)
191.10 THz 1520 nm
29
 Exercise

Let’s communicate “Hey” across the classroom

30
 Exercise

(On) = 1

(Off) = 0
2 symbols (on and off) or levels
Each symbol represents 1 bit

31
 Exercise

Let’s communicate “Hey”

Binary: 01001000 01100101 01111001

32
 Signal modulation

1. Transmitter modulates light signals (wavelengths)

1. Encode bits on a wave or pulse
2. By changing the properties of the signal
2. Receiver decodes the signal to retrieve bits
3. Digital (bits on transmitter) —> analog (optical
Non-return zero (NRZ) modulation
signal) —> Digital (bits on receiver)
4. Example modulation format: NRZ

33
 Signal modulation

1. Modulating the light signal Symbol

1. Encode bits on a wave or pulse
2. By changing the properties of the signal
de
2. Types of modulations litu
p
1. Change amplitude of the signal Am
θ Phase
2. Change phase of the signal
3. ..
3. Finite set of choices for change in properties of the signal
1. Each choice is called a symbol

34
 Signal modulation

1. Modulation packs bits on a signal

• Some formats pack more bits than others
2. Types of modulations
1. Change amplitude of the signal
2. Change phase of the signal
Constellation Diagram of
3. For example: Phase shift keying (PSK) 8-PSK modulation
modulation changes the phase of the signal.
35
 Signal modulation

1. Modulation format decides:

1. Changes to the signal from a set of alternatives
(symbols)
2. Each symbol communicates a fixed number of bits
3. Number of levels in a symbol = M, number of bits
per symbol, N = log2M
Baud rate = 4, N = 2
2. Symbol rate decides:
1. Number of symbols per second (baud rate)

36
 Wired network connectivity

Hartley’s Law:
R = fplog2M
Where,
R = data rate, bit rate in bits/second
fp = symbol rate or baud rate in symbols/second
Baud rate = 4, N = 2
M = number of levels in a given symbol

37
 Revise: signal modulation

1. Simple modulation format:

1. One symbol to represent “1”
2. One symbol to represent “0”
Input bit 1 Input bit 0
2. Modify the phase of the signal to encode
1. Phase = 0 to encode input bit 0
2. Phase = 180 to encode input bit 1
3. This modulation is called binary phase shift keying (BPSK)
4. Number of bits encoded per symbol N = log2M Transmitted Symbols
1. BPSK encodes 1 bit per symbol

38
 Revise: signal modulation

1. Quadrature phase shift keying (QPSK)

10 00
1. Four symbols
2. 2 bits per symbol
11 01

Transmitted Symbols

39
 Signal modulation

10 00
Input bit 1 Input bit 0

11 01

BPSK QPSK 16-QAM

1 bit per symbol 2 bits per symbol 4 bits per symbol

Packing more bits per symbol with different modulation formats

40
 Exercise: signal modulation

QAM: quadrature amplitude modulation uses

a mix of different amplitude levels and phase
shifts to create different symbols (see right).

Constellation Diagram of 16-QAM

41
 Exercise: signal modulation

QAM: quadrature amplitude modulation uses

a mix of different amplitude levels and phase
shifts to create different symbols (see right).

Constellation Diagram of 16-QAM

Exercise: If the baud rate of the transmission is 50 Gbaud, what is the data rate of a wavelength
modulated with 16-QAM modulation?

42
 Exercise: signal modulation
QAM: quadrature amplitude modulation uses
a mix of different amplitude levels and phase
shifts to create different symbols (see right).
Hartley’s Law
R = fplog2M
Where,
R = data rate, bit rate in bits/second
fp = symbol rate or baud rate in symbols/second
Constellation Diagram of 16-QAM
M = number of levels in a given symbol

Exercise: If the baud rate of the transmission is 50 Gbaud, what is the data rate of a wavelength
modulated with 16-QAM modulation?
Hint:16-QAM has 16 levels per symbol
Answer = 50 * log216 = 200Gbps
43