ECE544: Communication Networks-II

Spring 2015
D. Raychaudhuri
Lecture II
Includes teaching materials from L. Peterson, J. Kurose

Todays Lecture
Recap of network architecture &
top-down design
architecture paper discussion

Shared media (MAC) protocols

Ethernet
Token ring
IEEE 802.11

Link Layer: Introduction

Some terminology:
hub/repeater (layer 1), bridge/LAN switch (layer 2),
router (layer 3), host (layers 1-3 + app)
Links are communication channels that connect
adjacent nodes along communication path (point-topoint, shared, wired, wireless)
Layer-2 frame: encapsulates payload/datagram/IP
packet/service unit
LAN 1
LAN 2
Link
Lin
k
Host

Switch

Switch
Link
Router
Host

Link Layer Services

Data-link layer: transfer datagram from
one node to adjacent node over a link
Framing: encapsulate datagram into frame,
adding header, trailer.

Identify what set of bits constitute a frame, that is,

determining the beginning and the end of a frame

channel access if shared medium

MAC addresses used in frame headers to identify

source, destination
different from IP address!

Reliable delivery between adjacent nodes

Error detection
Error recovery: forward error correction code,
retransmission (ARQ)

Link Layer Communication

Link layer implemented in adaptor (NIC) and driver (Ethernet
card, WLAN card)
Sending side: encapsulates higher layer payload in a frame, adds
error checking bits, flow control, etc.
Receiving side: error detection, flow control, extracts payload,
passes to the receiving node
Link layer protocol

Sending Datagram
node

Datagram
Frame

Frame

Adaptor

Adaptor

Host
CPU
Control status
register

Cache

Memory

I/O
bus

Bus
interface

NIC
Link Network
interface

Recv
node

Layer 2 vs. Layer 3

Layer 2 switching
Based on MAC
address
Self configuring and
plug & play
Transparent to
protocols above the
MAC layer
Fast and inexpensive
Does not limit the
scope of broadcasts
Does not scale to
extremely large
networks

Layer 3 routing
Based on IP address
Must get IP address
(DHCP or manual
assign)
Easily connect LANs
that uses different link
protocols
Scalable to large
network by subnet
routing
Broadcast limited only
in a subnet

 Link Layer Techniques

Encoding (more Physical Layer stuff)
Framing & PPP Protocol
Error Detection & Correction
ARQ

Self study topics (see Ch2 & slides)

Binary Encoding
Binary Encoding: turn the binary data (bits) into
signals to transmit on cable or optical fiber link
(physical layer stuff, but better to know)
Baseband, not modulate to high frequency
Nonreturn To Zero (NRZ): 1=high signal, 0=low
signal
May stay on high or low signal too long for a long strings
of consecutive 1s or 0s => baseline wander, clock
recovery problems.

Nonreturn to Zero Inverted (NRZI): 1 = signal

transition (low to high, or high to low), 0=no
change.
Solve the problem of consecutive 1s, but not consecutive
0s

Manchester Encoding

Manchester Encoding: NRZ_encode data XOR clock

Clock cycle (a low/high pair) = 2 x signal interval
Baud rate (the signal change rate) = 2 x bitrate
0 =high-to-low transition, 1 = low-to-high transition
Clock recovery

Variation: Differential Manchester

1 = the first half of the signal equal to the last half of the previous
bits signal
0 = the first half of the signal opposite to the last half of the
previous bits signal
Bits
NRZ
Clock
Manchester
NRZI

Point-to-Point Data Link Protocol

Two types of links
point-to-point link (easier than broadcast link)
one sender, one receiver on the link, NO Media
Access Control
no need for explicit MAC addressing
e.g., dialup link, ISDN line

Broadcast (shared wire or medium)

popular point-to-point DLC protocols:

PPP (point-to-point protocol): byte-oriented
PPP for dial-up access
PPP over Ethernet (DSL)
HDLC (High level data link control): bit-oriented
Modem

PPP

PPP Functions
Framing: encapsulation of network-layer datagram in
data link frame
Identify what set of bits constitute a frame, that is,
determining the beginning and the end of a frame
carry data of any network layer protocol (not just IP) at
same time
ability to demultiplex upwards

bit transparency: must carry any bit pattern in the

data field
error detection (no correction)
connection liveness: detect, signal link failure to
network layer
network layer address negotiation: endpoint can
learn/configure each others network address
PPP

no error correction/recovery
no flow control
out of order delivery OK
no need to support multipoint links (e.g., polling)

PPP Data Frame

Flag: delimiter (framing)

Address:
Control:
Protocol: upper layer protocol to which frame carried (e.g. IP)
Info: upper layer data
Check: CRC
Octet: 1

1 or 2
protocol

01111110

11111111

00000011

flag

address

control

variable
info

2 or 4
CRC

1
01111110

Byte Stuff
data transparencyrequirement: data field
must be allowed to include flag pattern
<01111110>
Q: is received <01111110> data or flag?

Sender: adds (stuffs) extra < 01111110>

byte after each < 01111110> data byte
Receiver:
two 01111110 bytes in a row: discard first byte,
continue data reception
single 01111110: flag byte

PPP Link Control Protocol (LCP)

Before exchanging network-layer
data, data link peers must
configure PPP link (max. frame length,
authentication)

learn/configure network
layer information
for IP: carry IP Control Protocol (IPCP)
msgs (protocol field: 8021) to
configure/learn IP address

High-Level Data Link Control (HDLC)

Bit oriented protocol: view the frame as a collection of
bits, does not care byte boundaries.
Sentinel characters 01111110 transmitted as the link is
idle for synchronization
Bit stuffing: to distinguish the data pattern 01111110 in
the body from the special beginning/end sequence
after transmitting any 5 consecutive 1s in body, insert a 0
011111xxxx => 0111110xxx

Bits: 8
01111110
Beginning
sequence

16
Header

variable
body

16
CRC

8
01111110
ending
sequence

Error Detection
EDC= Error Detection and Correction bits (redundancy)
D = Data protected by error checking, may include header fields
Error detection not 100% reliable!

protocol may miss some errors, but rarely

larger EDC field yields better detection and correction

Parity Checking
Single Bit Parity:
Detect single
bit errors

Two Dimensional Bit Parity:

Detect and correct single
bit error

Internet Checksum
Goal: detect errors (e.g., flipped bits) in
transmitted segment (note: used at transport layer)

Sender:
treat segment
contents as sequence
of 16-bit integers
checksum: addition
(1s complement
sum) of segment
contents, and take
the ones complement
of the result
sender puts
checksum value into
UDP checksum field

Receiver:
compute checksum
of received segment
check if computed
checksum equals
checksum field value:
NO -error detected
YES -no error
detected. But maybe
errors (internet
checksum not very
strong for error
detection, but simple)

Cyclic Redundancy Check (CRC)

A (n+1)-bit message M can be represented as a
polynomial of degree n. For example,
X = 10011010;
M(X) = X7 + X4 + X3 + X
Choose k+1 bit pattern (divisor), C(X), a polyn of degree
k
goal: get k CRC bits, Y, such that
P=<M,Y> exactly divisible by C (modulo 2)
receiver knows C, divides <M,Y> by C. If non-zero
remainder: error detected!
can detect all nburst
errors less than
k+1 bits
bits
k bits
M: data bits to be sent

M x 2k XOR R

Y: CRC

CRC Example
Goal: design P(X) such that it
is exactly divisible by C(X)
T(X) = M(X) Xk (add k zeros to

the end of the message)

R remainder [

T(X )
]
C( X )

Subtract the remainder from T(X) to get P(X).

P(X) is now exactly divisible by C(X).
Corresponding to the complete
transmitted message

(Remember all addition/subtract use modulo-2

arithmetic)

Automatic Repeat reQuest(ARQ)

Stop-and-wait ARQ

Transmit a frame and wait for acknowledge

If positive acknowledge (ACK) from receiver, send next frame
If ACK does not arrive after a certain period of time (Timeout), retransmits the
frame
Simple, low efficiency

Go-back-N ARQ

Transmit frames continuously, no waiting

The receiver only acks the highest-numbered frames received in sequence
ACK comes back after a round-trip delay
If timeout, the sender retransmits the frames that are not acked and N-1
succeeding frames that were transmitted during the round-trip delay (N
frames transmitted during a round-trip delay)
Need buffer at transmitter, does not have to buffer the frames at the receiver,
moderate efficiency and complexity. Less efficient when the round-trip delay
is large and data transmission rate is high

Selective-repeat

Transmit continuously, no waiting

The receiver acks all successfully received frames
The sender only retransmits (repeats) the unacked frames when their timers
expire
Most efficient, but most complex, buffer needed at both transmitter and
receiver, need per frame timer

Sliding Window
Reliable delivery: retransmission
Ordered delivery: preserve the order in
which the frames are transmitted
Receiver does not pass along (buffer) out-oforder frames

Flow control: feedback mechanism by which

the receiver is able to throttle the sender
Inform the sender of how much frames the
receiver has room to receive

Sliding Window (Cont)

Send window size (SWS): the upper bound on the
number of unacked frames that the sender can
transmit,

set according to the round-trip delay to keep the pipe full

(recall: bandwidth x delay product represents the amount
of data that could be in transit)

LAR: the sequence # of the last ack received

LFS: the sequence # of the last frame sent
Receiver window size (RWS): the upper bound on
the number of out-of-order frames that the
receiver is willing to accept
LAF: the sequence # of the largest acceptable
frame
LFR: the sequence # of the last frame received
SeqNumToAck: the largest sequence # not yet
acked, such that all frames with seq # <=
SeqnumToAck have been received

Sliding Window
LAR

Sender

LFS

SWS

Ack 1
1

Receiver
LFR

Error

Ack 2

Ack 2
3

RWS

10 11 12

Ack 4 Ack 5
Ack 2 Ack 3
Ack 6

Ack 2
7

10 11 12

LAF

SeqNumToAck

LFS-LAR<=SWS, LAF-LFR<=RWS
Finite Seq. # wraps around:
SWS < (MaxSeqNum+1)/2 when RWS=SWS to distinguish
between different incarnations of the same seq. #

Shared Media Networks

MAC (medium access control)

ALOHA, Slotted ALOHA

CSMA/CD, CSMA/CA
Token Ring
TDMA, Dynamic TDMA
FDMA, CDMA

LAN Technologies

IEEE 802.3 Ethernet

IEEE 802.5 Token Ring
IEEE 802.11 Wireless LAN

Medium Access Sublayer

End host
Application

Presentation

Applications

Session

Transport

TCP/UDP

Network

IP

Data link

Physical

LLC
MAC

Subnet
Physical

Medium access control (MAC)

sublayer is not relevant on
point-to-point links
The MAC sublayer is only
used in broadcast or shared
medium/channel networks
All communication entities
share a common channel
Wired networks: Ethernet LAN
Wireless & Mobile Networks:
Satellite, Cellular, Wireless LAN,

Media Access Protocol

Shared broadcast channel
two or more simultaneous transmissions by nodes:
interference
Collision if node receives two or more signals at the same
time

MAC protocol
Determines how nodes share channel, i.e., determine
when node can transmit

Ideally, if broadcast channel of rate R bps

When one node wants to transmit, it can send at rate R.
When M nodes want to transmit, each can send at
average rate R/M (fairness)

MAC Classification
Channel Partitioning

divide channel into smaller pieces (time slots, frequency,

code)
allocate piece to node for exclusive use
TDMA, CDMA, FDMA

Random Access

channel not divided

When node has frame to send, transmit with the total channel
bandwidth

No coordination between nodes, control is completely

distributed
two or more nodes transmit simultaneously collision
random access MAC protocol should specify:

how to detect collisions

how to recover from collisions (e.g., via delayed retransmissions)
Examples: ALOHA, Slotted ALOHA, CSMA/CD, CSMA/CA

Taking turns

Nodes take turns

Token ring

Hybrid

Combine two or more techniques together

Pure (Unslotted) ALOHA

Early packet radio network created at the
U. of Hawaii in 1970
Uplink channel (clients->hub) and
downlink channel (hub->clients) uses
different frequencies

Hub

Client nodes send data frames to the

central hub using the shared uplink
channel.
The hub immediately re-send the
received frames, allowing clients to
determine whether or not their data had
been received properly.

Simplest form of random access,

provides basis for more advance
contention MAC

Client

Aloha Algorithm
Aloha Algorithm:
Nodes transmit immediately whenever they have a frame to
send
No synchronization among nodes
If collision, retransmit after random delay
random delay prevents the same frames from
colliding over and over again
collision window or vulnerable period:
frame sent at t0 collides with other frames sent in [t0-1,t0+1]

Pure Aloha efficiency

Assume that the aggregate frame arrival is Poisson Process
P [k arrivals in a time-interval] =

e GG k
k!

G: the mean number of aggregate arrivals (all nodes in network) in

the time interval
time-interval = one frame transmission time
Conditional successful probability for one frame transmission
attempt is

P0 = P [0 other attempts in 2 time-intervals] =e-2G

The probability of successful transmission

S = GP0 = Ge-2G
S is optimum at G=1/2
S=1/2e = 0.184

Slotted Aloha

Assumptions
all frames same size
time is divided into equal size
slots, time to transmit 1 frame
nodes start to transmit frames
only at beginning of slots
nodes are synchronized
if 2 or more nodes transmit in
slot, detect collision
Feedback channel about
whether packet is received or
not (half-duplex)

Operation

when node obtains fresh

frame, it transmits at the
beginning of next slot
no collision, node can send
new frame in next slot
if collision, wait a random
number of slots and try to
send again

Efficiency of Slotted ALOHA

Aggregate frame arrival is Poisson Process
e GG k
P [k arrivals in a time-interval] =
k!

G: the mean number of aggregate arrivals (for all nodes in

network) in this interval

time-interval = slot (one frame transmission time)

Successful probability for each slot is :

S= P [1 attempt in a slot] =Ge-G
S is optimum at G=1
S=1/e = 0.368
(slotted aloha reduce the potential collision period from 2t to t
by node synchronization)

Performance of ALOHA

Throughput versus offered traffic for ALOHA systems

The main reason for poor channel utilization of ALOHA (pure or
slotted) is that all stations can transmit at will, without paying
attention to what the other stations are doing.

Carrier Sense Multiple Access (CSMA)

CSMA: listen before transmit:
If channel sensed idle: transmit
If channel sensed busy, defer transmission
Human analogy: dont interrupt others!
Can collisions occur in this scheme?
Two nodes might attempt to transmit a frame at the same
time
Propagation delay means two nodes may not hear each
others transmission immediately

Several variants of CSMA protocols:

Non-Persistent CSMA
1-Persistent CSMA
P-Persistent CSMA

Non-persistent CSMA
To send data, a node first listens to the
channel to see if anyone else is transmitting.
If so, the node waits a random period of time
(instead of keeping sensing until the end of
the transmission) and repeats the algorithm.
Otherwise, it transmits a frame.
If a collision occurs, the node waits a random
amount of time and starts all over again.

1-persistent CSMA
Algorithm:
1. To send data, a node first listens to the channel to see if anyone else is
transmitting.
2. If so, the node waits (keeps sensing it) until the channel becomes idle.
Otherwise, it transmits a frame.
3. If a collision occurs, the node waits a random amount of time and starts
all over again.

It is called 1-persistent because the station transmits with a

probability of 1 whenever it starts sensing the channel and finds
the channel idle. (Greedy)

P-persistent CSMA

Assume channels are slotted

One slot = contention period (i.e., one round trip propagation
delay)
Algorithm:
1. Sense the channel

2.

If channel is idle, transmit a packet with probability p

if a packet was transmitted, go to step 2
if a packet was not transmitted, wait one slot and go to step 1
If channel is busy, wait one slot and go to step 1.
In other words, wait until idle and then transmit with probability
p

Detect collisions

If a collision occurs, wait a random amount of time and go to

step 1

Propagation Delay
A

D only sense As transmission after a

propagation delay
If is larger than packet transmission time,
too much time wasted.
CSMA in satellite communication? No.
Distance & propagation delay determine
collision probability
The size (length) of the network must be limited!

CSMA Performance Analysis

Assumptions
Constant length packets
No errors, except those caused by
collisions
Collision: entire packet transmission time
wasted
Each host can sense the transmissions of
all other hosts
The propagation delay is small compared
to the transmission time

Analysis of Non-persistent CSMA

Unsuccessful
transmission
period

Successful
transmission
period

Normalized Time

a
Y
a

1
Busy period

Idle period

1
Busy period

Poisson arrival, P(k arrivals in time duration t) =

Prob. of success transmission S= U x I/

(B+I)
Mean B = Y + 1 + a , mean I = 1/G
U = Ge-Ga
FY(y)=P{no packet occur in an duration of
1
a-y } = e-G(a-y)
E (Y ) (CDF)
a (1 e aG )
G

a: the ratio of
propagation
delay to packet
transmission
time

e Gt G kt
k!

Ge aG
S
G (1 2a ) e aG

Comparison of the channel utilization versus load for

various random access protocols

CSMA with Collision Detection

CSMA/CD (Carrier Sense Multiple Access with Collision
Detection) protocol further improves ALOHA by
aborting transmissions as soon as a collision is
detected.
Operation:

To send data, a node first listens to the channel to see if

anyone else is transmitting.

If not, it transmits a frame

If channel busy, deferral as in CSMA

the node wait a random period of time and repeats the

algorithm (non-persistent), or waits until the end of the
transmission (1-persistent)

The node will detect the collision, if collision detected,

abort its transmission (reducing channel wastage),
waits a random amount of time, and starts all over
again.

How to Detect Collision

Prerequisite: A node can listen while

talking
Easy in wired LANs: measure signal
strength, compare Tx and Rx signals
Difficult in wireless LANs: receiver shut
off while transmitting

Tx

Rx

CSMA/CA
Wireless LANs
How can a node detect collision if it cannot listen while
talking?
Collision Avoidance
Random Backoff (instead of 1-persistent)
Request-to-send (RTS)/clear-to-send (CTS)
CS no longer works well
Rules:
carrier
==> do not transmit
no carrier ==> OK to transmit
But the above rules do not always apply to wireless.

Problems with carrier sensing

Hidden terminal problem

Z
Y

W
W finds that medium is free
and it transmits a packet to Z

no carrier ===>
/ OK to transmit

Problems with carrier sensing

Exposed terminal problem

W
Z is transmitting
to W

Z
X

Y will not transmit to X

even though it cannot interfere
Presence of carrier ===>
/ hold off transmission

Solving Hidden Node problem with RTS/CTS

- listen RTS
- wait long enough
for the requested
station to respond
with CTS
- if (timeout) then
ready to transmit

CTS

RTS

X
Y

- listen CTS
- wait long enough
for the transmitter
to send its data

listen RTS ==> transmitter is close

listen CTS ==> receiver is close

Note: RTS/CTS does not solve exposed terminal problem. In the example above,
X can send RTS, but CTS from the responder will collide with Ys data.

RTS/CTS exchange example

SIFS
DIFS

Transmitter

Receiver

Other

Frame

RTS

ACK

CTS
352
304
s 10 s
s

8192 s
10
s

304
10 s
s

NAV (RTS)
NAV (CTS)

RTS + CTS + Frame + ACK exchange invoked when frame

size is large
NAV (Network Allocation Vector)
NAV maintains prediction of future traffic on the medium
based on duration information that is announced in
RTS/CTS frames prior to actual exchange of data

Taking Turns MAC protocols

Token passing:
Polling:
control token passed from one
master node
node to next sequentially.
invites slave
nodes to transmit token message
concerns:
in turn
token overhead
concerns:
complexity
polling
overhead
single point of failure (token)
single point of
failure (master)

TDMA
Time Division Multiple Access (TDMA)

Fixed TDMA
access to channel in "rounds"
each station gets fixed length slot (length =
packet transmission time) in each round
unused slots go idle Not efficient
example: 6-station LAN, 1,3,4 have pkt, slots
2,5,6 idle

Dynamic TDMA
In dynamic TDMA, a scheduling algorithm dynamically
reserves a variable number of timeslots in each frame
to variable user data streams, based on the traffic
demand of each user data stream.
Negotiations (beforehand) to determine how to allocate
slots dynamically.

Modem
preamble

TDMDownlink

TDDTDMAFrame
DTDMAUplink
SALOHA
control

BurstfromAccessPoint>Mobiles

BurstfromUserA
ToAccessPoint

Frame header and schedule

UserB

UserC

FDMA

Frequency bands

FDMA: frequency division multiple access

channel spectrum divided into frequency bands
each station assigned a frequency band
unused transmission time in frequency bands go
idle if assignment fixed
Inefficient => make it dynamically assigned to
different stations based on traffic demand
OFDMA

Spread Spectrum and CDMA

What if we dony not divide up the channel by time (as in
TDMA), or frequency (as in FDMA)? Is collision inevitable?
Not if collision is no longer damaging!
Is there any way to decode bits garbled by other
overlapping frames?
Code Division Multiple Access (CDMA) based on Spread Spectrum
Another perspective to solve multiple access problems
Spread Spectrum is a PHY innovation, not a MAC technique.
CDMA encodes data with a special code associated with each
user and uses the constructive interference properties of the
special codes to perform the multiplexing.

Spread Spectrum
Idea
spread signal over wider frequency band than
required
originally deigned to thwart jamming
Frequency Hopping
transmit over random sequence of frequencies
sender and receiver share
pseudorandom number generator
seed

Spread Spectrum (cont)

Direct Sequence
for each bit, send XOR of that bit and n
random bits
random sequence known to both sender
and receiver
called n-bit chipping code
1
0

Data stream: 1010

1
0

Random sequence: 0100101101011001

1
0

XOR of the two: 1011101110101001

Code Division Multiple Access (CDMA)

Multiplexing Technique used with spread spectrum
Start with data signal rate D
Called bit data rate
Break each bit into k chips according to fixed
pattern specific to each user
Users code
New channel has chip data rate kD chips per
second
E.g. k=6, three users (A,B,C) communicating with
base station R
Code for A = <1,-1,-1,1,-1,1>
Code for B = <1,1,-1,-1,1,1>
Code for C = <1,1,-1,1,1,-1>

LAN technologies
Ethernet
Token Ring
Wireless LAN

Ethernet Overview
History
developed by Xerox PARC in mid-1970s
roots in Aloha packet-radio network
standardized by Xerox, DEC, and Intel in 1978
similar to IEEE 802.3 standard
CSMA/CD
Evolution: Bus topology (90s) Star topology (now)
Most successful access network technology
Hub or switch

Advance

Ethernet Frame

Preamble: 8 bytes
7 bytes with pattern 10101010 followed by one byte with
pattern 10101011
used to synchronize receiver, sender clock rates
Addresses:6 bytes
if adapter receives frame with matching destination address,
or with broadcast address, it passes data in frame to net-layer
protocol, otherwise, adapter discards frame
Type: 2 bytes
indicates the higher layer protocol (mostly IP but others also
supported)

CRC: 4 bytes

checked at receiver, if error is detected, the frame is simply dropped

Body: 46-1500 bytes

Sending adapter encapsulates IP datagram (or other network layer

protocol packet) in Ethernet frame

Octets

Preamble

Dest
addr

Src
addr

Type
64-1518

72-1526

46-1500
Body

4
CRC

MAC Address
MAC Addresses
unique, 48-bit unicast address assigned to each adapter
example: 38:10:2b:e4:b1:02
broadcast: all 1s, ff:ff:ff:ff:ff:ff
multicast: multicast flag (the lowest bit of the 1st octet)= 1
01-00-5E-00-00-00 to 01-00-5E-7F-FF-FF for IP multicast
IP multicast group address mapped to the lower
order 23 bits of MAC address (not one-to-one
mapping)
Unique MAC address allocation administered by IEEE
manufacturer buys portion of MAC address space
the first three octets as vendor-specific

MAC Address vs. IP Address

48-bit MAC address
32-bit IP address
Layer 2
Network layer
Used to get packet from
Used to get packet to
one interface to
destination IP subnet
another within the
Hierarchical
same LAN/subnet
Change when moving
(Ethernet, token ring)
Flat
Depending on IP subnet
to which node is
Unique
attached
No change when
moving

IP to MAC address translation: ARP (more later)

Different Flavors of Ethernet Format

Ethernet version II
Octets

6
2
Src
Dest
Type
MAC
MAC
Data link header

Preamble

IEEE 802.3
Octets 8
Preamble

Dest
MAC

Src
MAC

Datalink Header

46-1500

Length DSAP SSAP

Body

4
FCS

Data & CRC (FCS)

1
Control

Logical Link Control

43-1497
Body

4
FCS

Data & CRC (FCS)

Length: the length of the data in the frame (excluding preamble, CRC, DLC
addresses, and the Length field itself)
Destination Service Access Point (DSAP): a pointer to a memory buffer in the
receiving station. It tells the receiving NIC in which buffer to put this
information. useful in situations where users are running multiple protocol
stacks, etc...
Source Service Access Point (SSAP)
Control: the type of LLC frame
Distinguish Ethertypes and Control field
Ethertypes value > 0x05DC (1500), Length <= 1500

Unreliable, connectionless service

Connectionless: No handshaking
between sending and receiving adapter.
Unreliable: receiving adapter doesnt
send acks or nacks to sending adapter
stream of datagrams passed to network
layer can have gaps
gaps will be filled if app is using TCP
otherwise, app will see the gaps

Ethernet CSMA/CD
1. If sender senses channel idle, it starts to transmit
frame. If it senses channel busy, waits until
channel idle and then transmits (1-persistent
CSMA)

Inter-frame gap: time to send 96 bits (9.6 s for 10Mbps)

2. If adapter transmits entire frame without

detecting another transmission, the adapter is
done with frame !
3. If adapter detects another transmission while
transmitting, aborts and sends 32-bit jam signal
(collision detection)
4. After aborting, sender enters exponential backoff

after the mth collision, adapter chooses a K at random

from {0,1,2,,2m-1}. Then waits K512 bit times (k x
51.2 us in 10 Mbps Ethernet) and returns to Step 1
give up after several tries (usually 16)

Ethernet CSMA/CD (Cont)

Exponential Backoff:

Jam Signal:
Goal: adapt retransmission
make sure all other
attempts to the estimated
transmitters are aware of
current # of active stations or
collision;
load
heavy load: random wait will be
32 bits
longer
Frame: 64 (preamble) +
32 (jamming sequence) = first collision: choose K from
{0,1}; delay is K512 bit (51.2
96 bits Runt Frame
s in 10 Mbps) transmission
times
after second collision: choose K
from {0,1,2,3}
after ten collisions, choose K
from {0,1,2,3,4,,1023}

Collisions
A

The longer the propagation delay,

the higher probability of collision.

Worst case:
A sends at t, As frame
arrives B at t+d
B begins transmitting at t+d
and collides with As frame
B sends runt frame, the runt
frame arrives A at t+2d
To detect collision, A must
continue transmit until t+2d.
A must transmit for 2d.
Round-trip delay about 51.2
us for 2500m long Ethernet
with 4 repeater
Corresponds to 512 bits for
10 Mbps Ethernet
So min frame size 512 bits

10BaseT and 100BaseT

10/100 Mbps rate; latter called fast ethernet

T stands for Twisted Pair
Star toplogy, max 100m between node and hub
Hubs: physical-layer repeaters
bits coming from one link go out all other links at the
same rate
no frame buffering
no CSMA/CD at hub: adapters detect collisions
provides net management functionality
Hub

Legacy Ethernet
10Base5

Bus topology with coaxial cable

10 Mbps, Up to 500m each segment
No more than 4 repeaters between any pair of stations
Max 2500 m
Max 1024 hosts

10Base2

Daisy chain
Up to 200m
Repeater
Terminator

Terminator

Transceiver

Adaptor

10 Base5 Ethernet

Gbit Ethernet
uses standard Ethernet frame format
allows for point-to-point links and shared
broadcast channels
in shared mode, CSMA/CD is used; short
distances between nodes required for
efficiency
uses hubs
Full-Duplex at 1 Gbps for point-to-point links
10 Gbps now

Ethernet Performance
Max throughput <1 as a function of span

As propagation delay increases, efficiency decreases

instability can occur unless load is reduced under congestion
conditions
retransmission backoff policy for stability

~0.8
Thru

stable policy
(retx backoff)

Capacity Limit
Traffic
margin

Overload
region

unstable policy
(no backoff)

load lines
Normal operating
point

stable policy
(backoff too high)

Offered Traffic

Wireless LANs
802.11 a/b/g different Phy technologies
802.11 b/g: 20 MHz channel in 2.4 GHz, up
to 11 Mbps (802.11b), 54 Mbps (802.11g)
phy data rate
802.11a: 20 MHz channel in 5GHz, up to 54
Mbps phy data rate

802.11n:
130 Mbps phy data rate on 20 MHz channel
(2 x 2 MIMO)
300 Mbps phy data rate on 40 MHz channel
(channel bonding with 2 x 2 MIMO)

See supplementary WLAN tutorial slides

Todays Homework
Peterson & Davie, Chap 2, 4th ed
2.6
2.18
2.23
2.33
2.44
2.42
Download and review Ethernet and 802.11 MAC
specs, and study IEEE 802.11 Wireless LAN
Overview slides
Due 2/6
74