Chapter 3

Transport Layer

A note on the use of these Powerpoint slides:

We’re making these slides freely available to all (faculty, students, readers).
They’re in PowerPoint form so you see the animations; and can add, modify,
and delete slides (including this one) and slide content to suit your needs.
They obviously represent a lot of work on our part. In return for use, we only
ask the following: Computer
▪ If you use these slides (e.g., in a class) that you mention their source
(after all, we’d like people to use our book!)
Networking: A Top
Down Approach
▪ If you post any slides on a www site, that you note that they are adapted
from (or perhaps identical to) our slides, and note our copyright of this
material.
7 th edition
Thanks and enjoy! JFK/KWR
Jim Kurose, Keith Ross
All material copyright 1996-2016 Pearson/Addison Wesley
J.F Kurose and K.W. Ross, All Rights Reserved April 2016
Transport Layer 2-1
Chapter 3: Transport Layer
our goals:
▪ understand ▪ learn about Internet
principles behind transport layer protocols:
transport layer • UDP: connectionless
services: transport
• multiplexing, • TCP: connection-oriented
demultiplexing reliable transport
• reliable data transfer • TCP congestion control
• flow control
• congestion control

Transport Layer 3-2

Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-3

Transport services and protocols
application

▪ provide logical
transport
network

communication between
data link
physical

app processes running on

different hosts

lo tra
▪ transport protocols run in

gi
ca nsp
end systems

l e or
nd t
• send side: breaks app

-e
nd
messages into segments,
passes to network layer
• rcv side: reassembles
application
transport

segments into messages,

network
data link
physical
passes to app layer
▪ more than one transport
protocol available to apps
• Internet: TCP and UDP
Transport Layer 3-4
Transport vs. network layer
▪ network layer: household
logical
communication 12 kidsanalogy:
in Ann’s house
sending letters to 12 kids
between hosts in Bill’s house:
▪ transport layer: ▪ hosts = houses
logical ▪ processes = kids
communication ▪ app messages = letters in
envelopes
between processes ▪ transport protocol = Ann
• relies on, enhances, and Bill who demux to in-
network layer house siblings
services ▪ network-layer protocol =
postal service

Transport Layer 3-5

Internet transport-layer protocols
▪ reliable, in-order
application
transport
network

delivery (TCP) data link

physical
network

• congestion control network

data link
data link
physical

• flow control
physical
network

lo
gi ran
data link

• connection setup

ca sp
physical

t
l e or
nd t
▪ unreliable, unordered
network

-e
data link

nd
physical

delivery: UDP network

data link

• no-frills extension of
physical
network
data link
“best-effort” IP
application
physical transport
network
data link network

▪ services not available: physical data link

physical

• delay guarantees
• bandwidth guarantees

Transport Layer 3-6

Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-7

Multiplexing/demultiplexing
multiplexing at
handle data from multiple
sender: demultiplexing at
use header info to deliver
sockets, add transport header received receiver:
segments to
(later used for demultiplexing) correct
socket
application

application P1 P2 application socket

P3 transport P4
proces
transport network transport s
network link network
link physical link
physical physical

Transport Layer 3-8

How demultiplexing works
▪ host receives IP datagrams 32
bits
• each datagram has source source port dest port
IP address, destination IP # #
address
other header
• each datagram carries one fields
transport-layer segment
• each segment has source, applicatio
destination port number n
▪ host uses IP addresses & data
port numbers to direct (payload)
segment to appropriate
TCP/UDP segment
socket format

Transport Layer 3-9

 Connectionless demultiplexing
▪ recall: created socket has ▪ recall: when creating
host-local port #: datagram to send into
DatagramSocket mySocket1 UDP socket, must specify
= new DatagramSocket(12534);
• destination IP address
• destination port #

▪ when host receives IP datagrams with same

UDP segment: dest. port #, but different
• checks destination port source IP addresses and/
# in segment or source port numbers
will be directed to same
• directs UDP segment to socket at dest
socket with that port #

Transport Layer 3-10

Connectionless demux: example
DatagramSocket
DatagramSocket serverSocket =
new DatagramSocket DatagramSocket
mySocket2 = new mySocket1 = new
DatagramSocket (6428); DatagramSocket
(9157); application (5775);
application P application
P 1 P
3 4
transport
transport transport
network
network link network
link physical link
physical physical

source port: source port:

6428 ?
dest port: 9157 dest port: ?

source port: source port:

9157 ?
dest port: 6428 dest port: ?
Transport Layer 3-11
 Connection-oriented demux
▪ TCP socket identified ▪ server host may support
by 4-tuple: many simultaneous TCP
• source IP address sockets:
• source port number • each socket identified by
• dest IP address its own 4-tuple
• dest port number ▪ web servers have
▪ demux: receiver uses different sockets for
all four values to direct each connecting client
segment to appropriate • non-persistent HTTP will
socket have different socket for
each request

Transport Layer 3-12

 Connection-oriented demux: example

application
application P P P application
P 4 5 6 P P
3 2 3
transport
transport transport
network
network link network
link physical link
physical physical
server: IP
address B

host: IP source IP,port: B,80 host: IP

address A dest IP,port: source IP,port: address C
A,9157 C,5775
dest IP,port: B,80
source IP,port:
A,9157
dest IP, port: B,80 source IP,port:
C,9157
dest IP,port: B,80
three segments, all destined to IP address: B,
dest port: 80 are demultiplexed to different Transport Layer 3-13
sockets
 Connection-oriented demux: example
threaded
server
application
application application
P P4 P P
3 2 3
transport
transport transport
network
network link network
link physical link
physical physical
server: IP
address B

host: IP source IP,port: B,80 host: IP

address A dest IP,port: source IP,port: address C
A,9157 C,5775
dest IP,port: B,80
source IP,port:
A,9157
dest IP, port: B,80 source IP,port:
C,9157
dest IP,port: B,80

Transport Layer 3-14

Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-15

UDP: User Datagram Protocol [RFC 768]
▪ “no frills,” “bare bones” ▪ UDP use:
Internet transport ▪ streaming multimedia
protocol apps (loss tolerant, rate
▪ “best effort” service, sensitive)
UDP segments may be: ▪ DNS
• lost ▪ SNMP
• delivered out-of-order ▪ reliable transfer over
to app
UDP:
▪ connectionless:
▪ add reliability at
• no handshaking application layer
between UDP sender,
receiver ▪ application-specific error
recovery!
• each UDP segment
handled independently
of others
Transport Layer 3-16
UDP: segment header
length, in bytes of
32 UDP segment,
bits including header
source port dest port
#
lengt #
checksu
h m why is there a UDP?
▪ no connection
applicatio establishment (which can
n add delay)
data
▪ simple: no connection
(payload)
state at sender, receiver
▪ small header size
UDP segment ▪ no congestion control:
format UDP can blast away as
fast as desired

Transport Layer 3-17

UDP checksum
Goal: detect “errors” (e.g., flipped bits) in
transmitted segment
sender: receiver:
▪ treat segment contents, ▪ compute checksum of
including header fields, received segment
as sequence of 16-bit ▪ check if computed
integers checksum equals checksum
▪ checksum: addition field value:
(one’s complement sum)
of segment contents • NO - error detected
▪ sender puts checksum • YES - no error detected.
value into UDP But maybe errors
checksum field nonetheless? More later
….
Transport Layer 3-18
Internet checksum: example
example: add two 16-bit integers
1 1 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0
1 1 1 0 1 0 1 0 1 0 1 0 1 0 1 0 1

wraparound 1 1 0 1 1 1 0 1 1 1 0 1 1 1 0 1 1

su 1 1 0 1 1 1 0 1 1 1 0 1 1 1 1 0 0
m
checksu 1 0 1 0 0 0 1 0 0 0 1 0 0 0 0 1 1
m

Note: when adding numbers, a carryout from the most

significant bit needs to be added to the result

* Check out the online interactive exercises for more

examples: h ttp://gaia.cs.umass.edu/kurose_ross/interactive/ Transport Layer 3-19
Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-20

Principles of reliable data transfer
▪ important in application, transport, link layers
• top-10 list of important networking topics!

▪ characteristics of unreliable channel will determine

complexity of reliable data transfer protocol (rdt)
Transport Layer 3-21
Principles of reliable data transfer
▪ important in application, transport, link layers
• top-10 list of important networking topics!

▪ characteristics of unreliable channel will determine

complexity of reliable data transfer protocol (rdt)
Transport Layer 3-22
Principles of reliable data transfer
▪ important in application, transport, link layers
• top-10 list of important networking topics!

▪ characteristics of unreliable channel will determine

complexity of reliable data transfer protocol (rdt)
Transport Layer 3-23
Reliable data transfer: getting started

rdt_send(): called from above, deliver_data(): called by

(e.g., by app.). Passed data to rdt to deliver data to upper
deliver to receiver upper layer

sen receiv
d e
side side

udt_send(): called by rdt, rdt_rcv(): called when packet

to transfer packet over arrives on rcv-side of channel
unreliable channel to receiver

Transport Layer 3-24

Reliable data transfer: getting started
we’ll:
▪ incrementally develop sender, receiver sides of
reliable data transfer protocol (rdt)
▪ consider only unidirectional data transfer
• but control info will flow on both directions!
▪ use finite state machines (FSM) to specify sender,
receiver
event causing state
transition
actions taken on state
state: when in this transition
“state” next state stat stat
uniquely determined e even
by next event t
e
1 action
s
2

Transport Layer 3-25

rdt1.0: reliable transfer over a reliable channel
▪ underlying channel perfectly reliable
• no bit errors
• no loss of packets
▪ separate FSMs for sender, receiver:
• sender sends data into underlying channel
• receiver reads data from underlying channel

Wait for rdt_send(data) Wait for rdt_rcv(packet

call from call from extract )(packet,data)
above packet = make_pkt(data) below deliver_data(data)
udt_send(packet)

sende receive
r r
Transport Layer 3-26
rdt2.0: channel with bit errors
▪ underlying channel may flip bits in packet
• checksum to detect bit errors
▪ the question: how to recover from errors:
• acknowledgements (ACKs): receiver explicitly tells
sender that pkt received OK
• negative acknowledgements (NAKs): receiver explicitly
tells sender that pkt had errors
• sender retransmits pkt on receipt of NAK
How do humans recover from
▪ new mechanisms in rdt2.0 (beyond rdt1.0):
• error detection “errors”
>sender
during
• receiver feedback: conversation?
control msgs (ACK,NAK) rcvr-

Transport Layer 3-27

rdt2.0: channel with bit errors
▪ underlying channel may flip bits in packet
• checksum to detect bit errors
▪ the question: how to recover from errors:
• acknowledgements (ACKs): receiver explicitly tells
sender that pkt received OK
• negative acknowledgements (NAKs): receiver explicitly
tells sender that pkt had errors
• sender retransmits pkt on receipt of NAK
▪ new mechanisms in rdt2.0 (beyond rdt1.0):
• error detection
• feedback: control msgs (ACK,NAK) from receiver to
sender

Transport Layer 3-28

rdt2.0: FSM specification
rdt_send(data)
sndpkt = make_pkt(data, checksum) receive
udt_send(sndpkt)
rdt_rcv(rcvpkt) && r
isNAK(rcvpkt)
Wait for Wait for rdt_rcv(rcvpkt) &&
call from ACK or udt_send(sndpkt) corrupt(rcvpkt)
above NAK
udt_send(NAK)

rdt_rcv(rcvpkt) && isACK(rcvpkt)

Wait for
Λ
call from
sende below

r
rdt_rcv(rcvpkt) &&
notcorrupt(rcvpkt)
extract(rcvpkt,data)
deliver_data(data)
udt_send(ACK)

Transport Layer 3-29

rdt2.0: operation with no errors
rdt_send(data)
snkpkt = make_pkt(data, checksum)
udt_send(sndpkt)
rdt_rcv(rcvpkt) &&
isNAK(rcvpkt)
Wait for Wait for rdt_rcv(rcvpkt) &&
call from ACK or udt_send(sndpkt) corrupt(rcvpkt)
above NAK
udt_send(NAK)

rdt_rcv(rcvpkt) && isACK(rcvpkt)

Wait for
Λ call from
below

rdt_rcv(rcvpkt) &&
notcorrupt(rcvpkt)
extract(rcvpkt,data)
deliver_data(data)
udt_send(ACK)

Transport Layer 3-30

rdt2.0: error scenario
rdt_send(data)
snkpkt = make_pkt(data, checksum)
udt_send(sndpkt)
rdt_rcv(rcvpkt) &&
isNAK(rcvpkt)
Wait for Wait for rdt_rcv(rcvpkt) &&
call from ACK or udt_send(sndpkt) corrupt(rcvpkt)
above NAK
udt_send(NAK)

rdt_rcv(rcvpkt) && isACK(rcvpkt)

Wait for
Λ call from
below

rdt_rcv(rcvpkt) &&
notcorrupt(rcvpkt)
extract(rcvpkt,data)
deliver_data(data)
udt_send(ACK)

Transport Layer 3-31

rdt2.0 has a fatal flaw!
what happens if ACK/ handling duplicates:
NAK corrupted? ▪ sender retransmits
▪ sender doesn’t know current pkt if ACK/NAK
what happened at corrupted
receiver!
▪ sender adds sequence
▪ can’t just retransmit: number to each pkt
possible duplicate
▪ receiver discards (doesn’t
deliver up) duplicate pkt
stop and wait
sender sends one packet,

then waits for receiver

response
Transport Layer 3-32
rdt2.1: sender, handles garbled ACK/NAKs
rdt_send(data)
sndpkt = make_pkt(0, data, checksum)
udt_send(sndpkt) rdt_rcv(rcvpkt) &&
( corrupt(rcvpkt) ||
Wait for Wait for
ACK or
isNAK(rcvpkt) )
call 0 from
NAK 0 udt_send(sndpkt)
above
rdt_rcv(rcvpkt)
&& rdt_rcv(rcvpkt)
notcorrupt(rcvpkt) && notcorrupt(rcvpkt)
&& isACK(rcvpkt) && isACK(rcvpkt)
Λ
Λ
Wait for Wait for
ACK or call 1 from
rdt_rcv(rcvpkt) && NAK 1 above
( corrupt(rcvpkt) ||
isNAK(rcvpkt) ) rdt_send(data)

udt_send(sndpkt) sndpkt = make_pkt(1, data, checksum)

udt_send(sndpkt)

Transport Layer 3-33

 rdt2.1: receiver, handles garbled ACK/NAKs
rdt_rcv(rcvpkt) && notcorrupt(rcvpkt)
&& has_seq0(rcvpkt)
extract(rcvpkt,data)
deliver_data(data)
sndpkt = make_pkt(ACK, chksum)
udt_send(sndpkt)
rdt_rcv(rcvpkt) && (corrupt(rcvpkt) rdt_rcv(rcvpkt) && (corrupt(rcvpkt)
sndpkt = make_pkt(NAK, chksum) sndpkt = make_pkt(NAK, chksum)
udt_send(sndpkt) udt_send(sndpkt)
Wait Wait
rdt_rcv(rcvpkt) && for for rdt_rcv(rcvpkt) &&
not corrupt(rcvpkt) && 0 from 1 from not corrupt(rcvpkt) &&
has_seq1(rcvpkt) below below has_seq0(rcvpkt)
sndpkt = make_pkt(ACK, chksum) sndpkt = make_pkt(ACK, chksum)
udt_send(sndpkt) udt_send(sndpkt)
rdt_rcv(rcvpkt) && notcorrupt(rcvpkt)
&& has_seq1(rcvpkt)

extract(rcvpkt,data)
deliver_data(data)
sndpkt = make_pkt(ACK, chksum)
udt_send(sndpkt)

Transport Layer 3-34

rdt2.1: discussion
sender: receiver:
▪ seq # added to pkt ▪ must check if received
▪ two seq. #’s (0,1) will packet is duplicate
suffice. Why? • state indicates
whether 0 or 1 is
▪ must check if received expected pkt seq #
ACK/NAK corrupted
▪ note: receiver can not
▪ twice as many states know if its last ACK/
• state must NAK received OK at
“remember” whether sender
“expected” pkt should
have seq # of 0 or 1

Transport Layer 3-35

 rdt2.2: a NAK-free protocol
▪ same functionality as rdt2.1, using ACKs only
▪ instead of NAK, receiver sends ACK for last pkt
received OK
• receiver must explicitly include seq # of pkt being ACKed
▪ duplicate ACK at sender results in same action as
NAK: retransmit current pkt

Transport Layer 3-36

 rdt2.2: sender, receiver fragments
rdt_send(data)
sndpkt = make_pkt(0, data, checksum)
udt_send(sndpkt) rdt_rcv(rcvpkt) &&
( corrupt(rcvpkt) ||
Wait for Wait for
ACK isACK(rcvpkt,1) )
call 0 from
above 0 udt_send(sndpkt)
sender
FSM rdt_rcv(rcvpkt)
fragment && notcorrupt(rcvpkt)
rdt_rcv(rcvpkt) && && isACK(rcvpkt,0)
(corrupt(rcvpkt) || Λ
has_seq1(rcvpkt)) Wait receiver
for
udt_send(sndpkt) 0 from FSM
below fragment
rdt_rcv(rcvpkt) && notcorrupt(rcvpkt)
&& has_seq1(rcvpkt)
extract(rcvpkt,data)
deliver_data(data)
sndpkt = make_pkt(ACK1, chksum)
udt_send(sndpkt) Transport Layer 3-37
rdt3.0: channels with errors and loss

new assumption: approach: sender waits

underlying channel “reasonable” amount of
can also lose packets time for ACK
(data, ACKs) ▪ retransmits if no ACK
• checksum, seq. #, received in this time
ACKs, retransmissions ▪ if pkt (or ACK) just delayed
(not lost):
will be of help … but
• retransmission will be
not enough duplicate, but seq. #’s
already handles this
• receiver must specify seq
# of pkt being ACKed
▪ requires countdown timer

Transport Layer 3-38

rdt3.0 sender
rdt_send(data)
rdt_rcv(rcvpkt) &&
sndpkt = make_pkt(0, data, checksum)
udt_send(sndpkt) ( corrupt(rcvpkt) ||
rdt_rcv(rcvpkt) start_timer isACK(rcvpkt,1)
Λ )
Λ Wait for Wait
for timeout
call 0from
ACK0 udt_send(sndpkt)
above
start_timer
rdt_rcv(rcvpkt)
&& rdt_rcv(rcvpkt)
notcorrupt(rcvpkt) && notcorrupt(rcvpkt)
&& isACK(rcvpkt,1)
stop_timer && isACK(rcvpkt,0)
stop_timer
Wait Wait for
timeout for call 1 from
udt_send(sndpkt) ACK1 above
start_timer rdt_rcv(rcvpkt)
rdt_send(data) Λ
rdt_rcv(rcvpkt)
&& sndpkt = make_pkt(1, data, checksum)
( corrupt(rcvpkt) || udt_send(sndpkt)
isACK(rcvpkt,0) ) start_timer
Λ

Transport Layer 3-39

 rdt3.0 in action
sende receiver sende receiver
r
send pkt0 r
send pkt0
pkt pkt
0 rcv 0 rcv
ack send
pkt0 ack send
pkt0
rcv ack0 0 ack0 rcv ack0 0 ack0
send pkt1 pkt send pkt1 pkt
1 rcv 1 X
ack send los
pkt1 s
rcv ack1 1 ack1
send pkt0 pkt
0 rcv timeout
ack send
pkt0 resend pkt
0 ack0 pkt1 1 rcv
ack send
pkt1
rcv ack1 1 ack1
send pkt0 pkt
(a) no 0 rcv
loss ack send
pkt0
0 ack0

(b) packet
loss Transport Layer 3-40
 rdt3.0 in action
sende receiver
sende receiver r
send pkt0 pkt
r
send pkt0 0 rcv
pkt
ack send
pkt0
0 rcv
send rcv ack0 0 ack0
ack pkt0 send pkt1 pkt
rcv ack0 0 ack0 1 rcv
send pkt1 pkt
1 rcv send
pkt1
ack ack1 ack1
send
pkt1
X 1
los ack1 timeout
s resend pkt
rcv
timeout
resend pkt rcv pkt1
ack1 1
pkt (detect
pkt1
rcv send pkt0 send
pkt1 1 (detect ack 0
duplicate)
ack pkt1 rcv
ack1
send
duplicate) rcv ack1 1 ack
rcv ack1 1 send pkt0 send
pkt0
send pkt0 pkt ack1 0 pkt ack0
rcv
0 rcv ack 0 (detect
ack send pkt0
pkt0 0 send
duplicate)
0 ack0 ack0
(c) ACK (d) premature timeout/ delayed
loss ACK
Transport Layer 3-41
Performance of rdt3.0
▪ rdt3.0 is correct, but performance stinks
▪ e.g.: 1 Gbps link, 15 ms prop. delay, 8000 bit packet:
L 8000 bits
Dtrans = R = = 8
109 bits/sec
microsecs
▪ U sender: utilization – fraction of time sender busy sending

▪ if RTT=30 msec, 1KB pkt every 30 msec: 33kB/sec thruput

over 1 Gbps link
▪ network protocol limits use of physical resources!
Transport Layer 3-42
 rdt3.0: stop-and-wait operation
sender receiver
first packet bit transmitted, t = 0
last packet bit transmitted, t = L / R

first packet bit arrives

RTT last packet bit arrives, send
ACK

ACK arrives, send next

packet, t = RTT + L / R

Transport Layer 3-43

Pipelined protocols
pipelining: sender allows multiple, “in-flight”, yet-
to-be-acknowledged pkts
• range of sequence numbers must be increased
• buffering at sender and/or receiver

▪ two generic forms of pipelined protocols: go-Back-N,

selective repeat
Transport Layer 3-44
 Pipelining: increased utilization
sender receiver
first packet bit transmitted, t = 0
last bit transmitted, t = L / R

first packet bit arrives

RTT last packet bit arrives, send ACK
last bit of 2nd packet arrives, send ACK
last bit of 3rd packet arrives, send ACK
ACK arrives, send next
packet, t = RTT + L / R
3-packet pipelining increases
utilization by a factor of 3!

Transport Layer 3-45

Pipelined protocols: overview
Go-back-N: Selective Repeat:
▪ sender can have up to ▪ sender can have up to N
N unacked packets in unack’ed packets in
pipeline pipeline
▪ receiver only sends ▪ rcvr sends individual ack
cumulative ack for each packet
• doesn’t ack packet if
there’s a gap
▪ sender has timer for ▪ sender maintains timer
oldest unacked packet for each unacked packet
• when timer expires, • when timer expires,
retransmit all unacked retransmit only that
packets unacked packet

Transport Layer 3-46

Go-Back-N: sender
▪ k-bit seq # in pkt header
▪ “window” of up to N, consecutive unack’ed pkts allowed

▪ ACK(n): ACKs all pkts up to, including seq # n - “cumulative

ACK”
• may receive duplicate ACKs (see receiver)
▪ timer for oldest in-flight pkt
▪ timeout(n): retransmit packet n and all higher seq # pkts in
window
Transport Layer 3-47
GBN: sender extended FSM
rdt_send(data)
if (nextseqnum < base+N) {
sndpkt[nextseqnum] = make_pkt(nextseqnum,data,chksum)
udt_send(sndpkt[nextseqnum])
if (base == nextseqnum)
start_timer
nextseqnum++
}
Λ else
refuse_data(data)
base=1
nextseqnum=1
timeout
start_timer
Wait
udt_send(sndpkt[base])
rdt_rcv(rcvpkt) udt_send(sndpkt[base+1])
&& corrupt(rcvpkt) …
udt_send(sndpkt[nextseqnum-1]
)
rdt_rcv(rcvpkt) &&
notcorrupt(rcvpkt)
base = getacknum(rcvpkt)+1
If (base == nextseqnum)
stop_timer
else
start_timer
Transport Layer 3-48
GBN: receiver extended FSM
defaul
t
udt_send(sndpkt) rdt_rcv(rcvpkt)
&& notcurrupt(rcvpkt)
Λ && hasseqnum(rcvpkt,expectedseqnum)
expectedseqnum=1 Wait extract(rcvpkt,data)
sndpkt = deliver_data(data)
make_pkt(expectedseqnum,ACK,chksum) sndpkt = make_pkt(expectedseqnum,ACK,chksum)
udt_send(sndpkt)
expectedseqnum++

ACK-only: always send ACK for correctly-received

pkt with highest in-order seq #
• may generate duplicate ACKs
• need only remember expectedseqnum
▪ out-of-order pkt:
• discard (don’t buffer): no receiver buffering!
• re-ACK pkt with highest in-order seq #
Transport Layer 3-49
 GBN in action
sender window sende receiver
0 1 2 3 (N=4)
4567 r
send
0 1 2 384 5 6 7 pkt0
send receive pkt0, send ack0
0 1 2 384 5 6 7
pkt1 X los receive pkt1, send ack1
0 1 2 384 5 6 7
8 send s
pkt2 receive pkt3, discard,
01234567 rcv ack0, send
send (re)send ack1
0 1 2 384 5 6 7 pkt4
pkt3 receive pkt4, discard,
8 rcv ack1,(wait)
send (re)send ack1
pkt5
ignore duplicate receive pkt5, discard,
ACK
(re)send ack1
pkt 2
01234567 timeout
send
0 1 2 384 5 6 7 pkt2
0 1 2 384 5 6 7 send rcv pkt2, deliver, send ack2
0 1 2 384 5 6 7 pkt3 rcv pkt3, deliver, send ack3
8 send rcv pkt4, deliver, send ack4
pkt4 rcv pkt5, deliver, send ack5
send
pkt5 Transport Layer 3-50
Selective repeat
▪ receiver individually acknowledges all correctly
received pkts
• buffers pkts, as needed, for eventual in-order
delivery to upper layer
▪ sender only resends pkts for which ACK not
received
• sender timer for each unACKed pkt
▪ sender window
• N consecutive seq #’s
• limits seq #s of sent, unACKed pkts

Transport Layer 3-51

Selective repeat: sender, receiver windows

Transport Layer 3-52

Selective repeat
sende receiver
datarfrom above: pkt n in [rcvbase, rcvbase+N-1]
▪ if next available seq # in ▪ send ACK(n)
window, send pkt ▪ out-of-order: buffer
timeout(n): ▪ in-order: deliver (also
▪ resend pkt n, restart timer deliver buffered, in-order
pkts), advance window to
ACK(n) in [sendbase,sendbase+N]: next not-yet-received pkt
▪ mark pkt n as received
pkt n in [rcvbase-N,rcvbase-1]
▪ if n smallest unACKed pkt,
advance window base to ▪ ACK(n)
next unACKed seq # otherwise:
▪ ignore

Transport Layer 3-53

Selective repeat in action
sender window sende receiver
0 1 2 3 (N=4)
4567 r
send
0 1 2 384 5 6 7 pkt0
send receive pkt0, send ack0
0 1 2 384 5 6 7
pkt1 X los receive pkt1, send ack1
0 1 2 384 5 6 7
8 send s
pkt2 receive pkt3, buffer,
01234567 rcv ack0, send
send send ack3
0 1 2 384 5 6 7 pkt4
pkt3 receive pkt4, buffer,
8 rcv ack1,(wait)
send send ack4
pkt5
record ack3 arrived receive pkt5, buffer,
send ack5
pkt 2
01234567 timeout
send
0 1 2 384 5 6 7 pkt2
record ack4 arrived
0 1 2 384 5 6 7 rcv pkt2; deliver pkt2,
record ack5 arrived
0 1 2 384 5 6 7 pkt3, pkt4, pkt5; send ack2
8
Q: what happens when ack2
arrives?
Transport Layer 3-54
Selective repeat:
sender receiver
window window
(after receipt) (after receipt)
dilemma 012301
2
012301
pkt
0
pkt 012301
2 1
pkt
2
012301
example:
012301
2 2 2
012301
pkt
▪ seq #’s: 0, 1, 2, 3
012301 2
2 3 X
012301
▪ window size=3 2 pkt will accept packet
0 with seq number
(a) no
▪ receiver sees no problem
0

difference in two receiver can’t see sender side.

scenarios! receiver behavior identical in both
cases!
▪ duplicate data something’s (very) wrong!
accepted as new in (b) 012301 pkt
2 0
pkt
012301 012301
2 1
pkt
2
Q: what relationship 012301
2 2X
012301
2
between seq # size
012301
2
X
and window size to timeout
retransmit X
avoid problem in (b)? pkt0
012301 pkt
will accept packet
2 0
with seq number
(b) 0
oops!
Transport Layer 3-55
Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-56

TCP: Overview RFCs: 793,1122,1323, 2018, 2581

▪ point-to-point: ▪ full duplex data:

• one sender, one • bi-directional data flow
receiver in same connection
▪ reliable, in-order byte • MSS: maximum
steam: segment size
• no “message ▪ connection-oriented:
boundaries” • handshaking (exchange
▪ pipelined: of control msgs) inits
sender, receiver state
• TCP congestion and before data exchange
flow control set
window size ▪ flow controlled:
• sender will not
overwhelm receiver
Transport Layer 3-57
 TCP segment structure
32
URG: urgent data bits
source port dest port counting
(generally not by bytes
used) #
sequence number#
ACK: ACK of data
# acknowledgement number (not
valid hea not segments!)
PSH: push data d use
UAP R S F receive
now len d window # bytes
checksu Urg data
(generally not used) rcvr
m pointer
RST, SYN, FIN: willing
options (variable
connection to accept
length)
estab
(setup, teardown
application
commands)
Internet data
checksum (variable
(as in length)
UDP)

Transport Layer 3-58

TCP seq. numbers, ACKs
outgoing segment from
sequence numbers: source port
#
sender
dest port
#
sequence number
• byte stream “number” of acknowledgement number

first byte in segment’s rwn

d
data
checksu urg
m pointer
window
acknowledgements: size
N
• seq # of next byte
expected from other side sender sequence number
• cumulative ACK space

Q: how receiver handles out- sent

ACKe
sent, not- usable not
yet ACKed but not usable
of-order segments d (“in- yet sent
flight”)
• A: TCP spec doesn’t say, - incoming segment to
up to implementor source port
#
sender
dest port
#
sequence number
acknowledgement number
A rwn
checksu d
urg
m pointer
Transport Layer 3-59
TCP seq. numbers, ACKs
Host Host
A B

User
types
‘C’ Seq=42, ACK=79, data =
‘C’ host ACKs
receipt of
‘C’,
Seq=79, ACK=43, data = echoes
host ‘C’
ACKs back ‘C’
receipt
of echoed Seq=43,
‘C’ ACK=80

simple telnet
scenario

Transport Layer 3-60

TCP round trip time, timeout
Q: how to set TCP Q: how to estimate RTT?
timeout value? ▪ SampleRTT: measured
time from segment
▪ longer than RTT transmission until ACK
• but RTT varies receipt
▪ too short: premature • ignore retransmissions
timeout, unnecessary ▪ SampleRTT will vary, want
retransmissions estimated RTT “smoother”
• average several recent
▪ too long: slow measurements, not just
reaction to segment current SampleRTT
loss

Transport Layer 3-61

TCP round trip time, timeout
EstimatedRTT = (1- α)*EstimatedRTT + α*SampleRTT
▪ exponential weighted moving average
▪ influence of past sample decreases exponentially fast
▪ typical value: α = 0.125

RTT: gaia.cs.umass.edu to fantasia.eurecom.fr

(milliseconds)
RTT

sampleRT
T
EstimatedRTT

time Transport Layer 3-62

 TCP round trip time, timeout
▪ timeout interval: EstimatedRTT plus “safety margin”
• large variation in EstimatedRTT -> larger safety margin
▪ estimate SampleRTT deviation from EstimatedRTT:
DevRTT = (1-β)*DevRTT +
β*|SampleRTT-EstimatedRTT|
(typically, β = 0.25)

TimeoutInterval = EstimatedRTT + 4*DevRTT

estimated “safety
RTT margin”

* Check out the online interactive exercises for more

examples: h ttp://gaia.cs.umass.edu/kurose_ross/interactive/ Transport Layer 3-63
Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-64

TCP reliable data transfer
▪ TCP creates rdt service
on top of IP’s unreliable
service
• pipelined segments
• cumulative acks let’s initially consider
• single retransmission simplified TCP sender:
timer • ignore duplicate acks
▪ retransmissions • ignore flow control,
triggered by: congestion control
• timeout events
• duplicate acks

Transport Layer 3-65

TCP sender events:
data rcvd from app: timeout:
▪ create segment with ▪ retransmit segment
seq # that caused timeout
▪ seq # is byte-stream ▪ restart timer
number of first data ack rcvd:
byte in segment ▪ if ack acknowledges
▪ start timer if not previously unacked
already running segments
• think of timer as for • update what is known
oldest unacked to be ACKed
segment • start timer if there are
• expiration interval: still unacked segments
TimeOutInterval

Transport Layer 3-66

TCP sender (simplified)
data received from application above
create segment, seq. #: NextSeqNum
pass segment to IP (i.e., “send”)
NextSeqNum = NextSeqNum +
length(data)
Λ if (timer currently not running)
NextSeqNum = wait
start timer
InitialSeqNum for
SendBase = InitialSeqNum even timeou
t t
retransmit not-yet-acked
segment with smallest seq.
#
ACK received, with ACK field value start timer
y
if (y > SendBase) {
SendBase = y
/* SendBase–1: last cumulatively ACKed byte
*/
if (there are currently not-yet-acked segments)
start timer
else stop timer Transport Layer 3-67
TCP: retransmission scenarios
Host Host Host Host
A B A B

SendBase=9
Seq=92, 8 bytes of 2 Seq=92, 8 bytes of
data data
Seq=100, 20 bytes of
timeo

timeo
ACK=10 data
ut

ut
X 0
ACK=10
0
ACK=12
0
Seq=92, 8 bytes of Seq=92, 8
data SendBase=10 bytes of data
0
SendBase=12
ACK=10 0 ACK=12
0
0
SendBase=12
0
lost ACK premature
scenario timeout
Transport Layer 3-68
TCP: retransmission scenarios
Host Host
A B

Seq=92, 8 bytes of
data
Seq=100, 20 bytes of
data
ACK=10
timeo

X 0
ut

ACK=12
0

Seq=120, 15 bytes of data

cumulative
ACK
Transport Layer 3-69
TCP ACK generation [RFC 1122, RFC 2581]

event at receiver TCP receiver action

arrival of in-order segment with delayed ACK. Wait up to 500ms
expected seq #. All data up to for next segment. If no next segment,
expected seq # already send ACK
ACKed
immediately send single cumulative
arrival of in-order segment with ACK, ACKing both in-order segments
expected seq #. One other
segment has ACK pending
immediately send duplicate ACK,
arrival of out-of-order segment indicating seq. # of next expected
higher-than-expect seq. # . byte
Gap detected

arrival of segment that immediate send ACK, provided that

partially or completely fills gap segment starts at lower end of gap
Transport Layer 3-70
TCP fast retransmit
▪ time-out period often
relatively long: TCP fast
• long delay before retransmit
if sender receives 3
resending lost packet ACKs for same data
▪ detect lost segments (“triple
(“triple duplicate
duplicate ACKs”),
ACKs”),
via duplicate ACKs. resend unacked
• sender often sends segment with smallest
many segments back- seq #
to-back
▪ likely that unacked
• if segment is lost, segment lost, so don’t
there will likely be wait for timeout
many duplicate ACKs.

Transport Layer 3-71

TCP fast retransmit
Host Host
A B

Seq=92, 8 bytes of
data
Seq=100, 20 bytes of
data
X

ACK=10
0
ACK=10
timeo
ut

0
ACK=10
0
ACK=10
0
Seq=100, 20 bytes of
data

fast retransmit after sender

receipt of triple duplicate
Transport Layer 3-72
ACK
Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-73

TCP flow control applicati
on
application may process
remove data from applicatio
TCP socket buffers …. n
TCP socket OS
receiver
… slower than TCP buffers
receiver is delivering
(sender is sending) TCP
code

IP
flow code
receiver controls sender, so
control
sender won’t overflow
receiver’s buffer by transmitting from
too much, too fast sender
receiver protocol
stack
Transport Layer 3-74
TCP flow control
▪ receiver “advertises” free
buffer space by including to application
process
rwnd value in TCP header
of receiver-to-sender
buffered data
segments RcvBuffe
r
• RcvBuffer size set via
socket options (typical default rwn free buffer
is 4096 bytes) d
space
• many operating systems
autoadjust RcvBuffer TCP segment
▪ sender limits amount of payloads
unacked (“in-flight”) data to receiver-side
receiver’s rwnd value buffering
▪ guarantees receive buffer
will not overflow
Transport Layer 3-75
Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-76

Connection Management
before exchanging data, sender/receiver “handshake”:
▪ agree to establish connection (each knowing the other willing
to establish connection)
▪ agree on connection parameters

applicatio applicatio
n n
connection state: ESTAB connection state: ESTAB
connection variables: connection Variables:
seq # client-to-server seq # client-to-server
server-to-client server-to-client
rcvBuffer size rcvBuffer size
at server,client at server,client

network network

Socket clientSocket = Socket connectionSocket =

newSocket("hostname","port welcomeSocket.accept();
number");
Transport Layer 3-77
Agreeing to establish a connection

2-way handshake:
Q: will 2-way handshake
always work in
Let’s
network?
talk ESTA ▪ variable delays
O
ESTA K
B ▪ retransmitted messages (e.g.
B req_conn(x)) due to
message loss
▪ message reordering
choose
req_conn(x
▪ can’t “see” other side
x
) ESTA
acc_conn(x B
ESTA )
B

Transport Layer 3-78

Agreeing to establish a connection
2-way handshake failure scenarios:

choose choose
req_conn(x x req_conn(x
x
) ESTA ) ESTA
retransmit B retransmit acc_conn(x B
acc_conn(x
req_conn(x req_conn(x )
)
) )
ESTA ESTA
B data(x+1 accept
B req_conn(x
retransmit ) data(x+1)
) data(x+1)
connection connection
client x server x server
completes
client completes
terminates forgets x terminates forgets x
req_conn(x
)
ESTA ESTA
data(x+1 accept
half open B B
) data(x+1)
connection!
(no client!) Transport Layer 3-79
 TCP 3-way handshake

client server
state state
LISTE LISTE
N choose init seq num, N
x
SYNSEN send TCP SYN msg SYNbit=1,
T Seq=x choose init seq num,
y
send TCP SYNACK SYN
SYNbit=1, Seq=y msg, acking SYN RCVD
ACKbit=1;
received SYNACK(x) ACKnum=x+1
ESTA indicates server is live;
send ACK for SYNACK;
B this segment may ACKbit=1,
contain
client-to-server data ACKnum=y+1 received ACK(y)
indicates client is
live ESTA
B

Transport Layer 3-80

TCP 3-way handshake: FSM

close
d
Socket connectionSocket =
welcomeSocket.accept();

Λ Socket clientSocket =
SYN(x newSocket("hostname","port
number");
)
SYNACK(seq=y,
ACKnum=x+1) SYN(seq=x
create new socket for liste
)
communication back to client n

SY SY
N N
rcvd sent
SYNACK(seq=y,
ESTAB ACKnum=x+1)
ACK(ACKnum=y+1
ACK(ACKnum=y+1 )
)
Λ

Transport Layer 3-81

TCP: closing a connection
▪ client, server each close their side of connection
• send TCP segment with FIN bit = 1
▪ respond to received FIN with ACK
• on receiving FIN, ACK can be combined with own FIN
▪ simultaneous FIN exchanges can be handled

Transport Layer 3-82

TCP: closing a connection
client server
state state
ESTA ESTA
B clientSocket. B
close() FINbit=1,
FIN_WAIT_ can no longer
1 send but can seq=x
receive data CLOSE_WAI
ACKbit=1; T
can still
FIN_WAIT_ wait for ACKnum=x+1 send
server data
2 close
LAST_AC
FINbit=1, K
TIMED_WAI seq=y can no
longer
T send data
ACKbit=1;
timed wait ACKnum=y+1
for 2*max CLOSED
segment
lifetime

CLOSED

Transport Layer 3-83

Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-84

Principles of congestion control
congestion:
▪ informally: “too many sources sending too much
data too fast for network to handle”
▪ different from flow control!
▪ manifestations:
• lost packets (buffer overflow at routers)
• long delays (queueing in router buffers)
▪ a top-10 problem!

Transport Layer 3-85

 Causes/costs of congestion: scenario 1
original data: λin throughput: λout
▪ two senders, two
receivers Host A

▪ one router, infinite unlimited shared

buffers output link buffers

▪ output link capacity: R

▪ no retransmission
Host B

R/2

dela
λout

y
λin R/2 λin R/2
▪ maximum per-connection ❖ large delays as arrival rate,
throughput: R/2 λin, approaches capacity
Transport Layer 3-86
Causes/costs of congestion: scenario 2
▪ one router, finite buffers
▪ sender retransmission of timed-out packet
• application-layer input = application-layer output: λin =
λout
• transport-layer input includes retransmissions : ‘λin λin

λin : original data

λout
λ' in: original data, plus
retransmitted data

Host A

finite shared output

Host B
link buffers
Transport Layer 3-87
Causes/costs of congestion: scenario 2
R/2
idealization: perfect
knowledge

λout
▪ sender sends only when
router buffers available
λin R/2

λin : original data

cop λout
λ' in: original data, plus
y retransmitted data

A free buffer space!

finite shared output

Host B
link buffers
Transport Layer 3-88
Causes/costs of congestion: scenario 2
Idealization: known
loss packets can be lost,
dropped at router due
to full buffers
▪ sender only resends if
packet known to be lost
λin : original data
cop λout
λ' in: original data, plus
y retransmitted data

A no buffer space!

Host B
Transport Layer 3-89
Causes/costs of congestion: scenario 2
Idealization: known R/
loss packets can be lost, 2
when sending at
dropped at router due R/2, some packets

λout
to full buffers are retransmissions
but asymptotic
▪ sender only resends if goodput is still R/2
packet known to be lost λin R/
(why?)
2

λin : original data

λout
λ' in: original data, plus
retransmitted data

A free buffer space!

Host B
Transport Layer 3-90
Causes/costs of congestion: scenario 2
Realistic: duplicates R/
▪ packets can be lost, dropped 2
at router due to full buffers when sending at
R/2, some packets

λout
▪ sender times out prematurely, are retransmissions

sending two copies, both of including duplicated

that are delivered!
which are delivered λin R/
2

λin
cop
timeout λout
λ' in
y

A free buffer space!

Host B
Transport Layer 3-91
Causes/costs of congestion: scenario 2
Realistic: duplicates R/
▪ packets can be lost, dropped 2
at router due to full buffers when sending at
R/2, some packets

λout
▪ sender times out prematurely, are retransmissions

sending two copies, both of including duplicated

that are delivered!
which are delivered λin R/
2

“costs” of congestion:
▪ more work (retrans) for given “goodput”
▪ unneeded retransmissions: link carries multiple copies of pkt
• decreasing goodput

Transport Layer 3-92

Causes/costs of congestion: scenario 3
▪ four senders Q: what happens as λin and λin’
increase ?
▪ multihop paths
A: as red λin’ increases, all arriving
▪ timeout/retransmit blue pkts at upper queue are
dropped, blue throughput → 0
Host A
λin : original data λout
Host B
λ' in: original data, plus
retransmitted data
finite shared
output link buffers

Host
D Host
C

Transport Layer 3-93

Causes/costs of congestion: scenario 3

C/
2
λout

λin’ C/
2

another “cost” of congestion:

▪ when packet dropped, any “upstream
transmission capacity used for that packet was
wasted!

Transport Layer 3-94

Chapter 3 outline
3.1 transport-layer 3.5 connection-oriented
services transport: TCP
3.2 multiplexing and • segment structure
demultiplexing • reliable data transfer
3.3 connectionless • flow control
transport: UDP • connection management
3.4 principles of reliable 3.6 principles of congestion
data transfer control
3.7 TCP congestion control

Transport Layer 3-95

 TCP congestion control: additive increase
multiplicative decrease
▪ approach: sender increases transmission rate
(window size), probing for usable bandwidth, until
loss occurs
• additive increase: increase cwnd by 1 MSS every
RTT until loss detected
• multiplicative decrease: cut cwnd in half after loss
additively increase window size …
…. until loss occurs (then cut window in
half)
cwnd: TCP sender
congestion window

AIMD saw tooth

behavior:
size

probing
for bandwidth

tim
e Transport Layer 3-96
TCP Congestion Control: details
sender sequence number
space cwn TCP sending rate:
d
▪ roughly: send cwnd
bytes, wait RTT for
last last byte
ACKS, then send
byte
ACKed
sent, not-
yet ACKed
sent more bytes
(“in-
flight”) cwn
▪ sender limits transmission: rat ~
~ d
RT
bytes/
sec
LastByteSent- <
e T
cwn
LastByteAcked d

▪ cwnd is dynamic, function

of perceived network
congestion
Transport Layer 3-97
TCP Slow Start
Host Host
▪ when connection begins, A B
increase rate
exponentially until first one
loss event:

RTT
segment

• initially cwnd = 1 MSS two

• double cwnd every RTT segments

• done by incrementing
cwnd for every ACK four
received segments

▪ summary: initial rate is

slow but ramps up
exponentially fast tim
e

Transport Layer 3-98

 TCP: detecting, reacting to loss
▪ loss indicated by timeout:
• cwnd set to 1 MSS;
• window then grows exponentially (as in slow start)
to threshold, then grows linearly
▪ loss indicated by 3 duplicate ACKs: TCP RENO
• dup ACKs indicate network capable of delivering
some segments
• cwnd is cut in half window then grows linearly
▪ TCP Tahoe always sets cwnd to 1 (timeout or 3
duplicate acks)

Transport Layer 3-99

 TCP: switching from slow start to CA
Q: when should the
exponential
increase switch to
linear?
A: when cwnd gets
to 1/2 of its value
before timeout.

Implementation:
▪ variable ssthresh
▪ on loss event, ssthresh
is set to 1/2 of cwnd just
before loss event

* Check out the online interactive exercises for more

examples: h ttp://gaia.cs.umass.edu/kurose_ross/interactive/ Transport Layer 3-100
Summary: TCP Congestion Control
New
New ACK!
ACK!
duplicate
ACK
dupACKcount+ new
new
cwnd = cwndACK .
+ MSS (MSS/cwnd)
dupACKcount = 0
+ ACK = cwnd+MSS
cwnd transmit new segment(s), as
dupACKcount = 0 allowed
Λ transmit new segment(s), as
cwnd = 1 MSS allowed
ssthresh = 64 KB cwnd >
dupACKcount =
0
slow ssthresh
Λ congestio
timeou n
start t = cwnd/2
ssthresh
cwnd = 1 MSS avoidance duplicate
timeou dupACKcount = 0 ACK
dupACKcount+
t = cwnd/2
ssthresh retransmit missing +
cwnd = 1 MSS segment
dupACKcount = 0
retransmit missing
segment timeou
New
t
ACK!
ssthresh = cwnd/2
cwnd = 1 New
dupACKcount = 0 ACK
retransmit missing cwnd = ssthresh dupACKcount ==
dupACKcount == dupACKcount =
3ssthresh= cwnd/2 segment 0 3
ssthresh= cwnd/2
cwnd = ssthresh + 3 cwnd = ssthresh + 3
retransmit missing retransmit missing
segment fast segment

recovery
duplicate
ACK
cwnd = cwnd + MSS
transmit new segment(s), as
allowed
Transport Layer 3-101
TCP throughput
▪ avg. TCP thruput as function of window size, RTT?
• ignore slow start, assume always data to send
▪ W: window size (measured in bytes) where loss occurs
• avg. window size (# in-flight bytes) is ¾ W
• avg. thruput is 3/4W per RTT
3 W
avg TCP thruput bytes/
4 RT sec
=
T
W

W/
2

Transport Layer 3-102

TCP Futures: TCP over “long, fat pipes”

▪ example: 1500 byte segments, 100ms RTT, want

10 Gbps throughput
▪ requires W = 83,333 in-flight segments
▪ throughput in terms of segment loss probability, L
[Mathis 1997]:
1.2 . MS
TCP throughput
RTT
2 SL
=
➜ to achieve 10 Gbps throughput, need a loss rate of L
= 2·10-10 – a very small loss rate!
▪ new versions of TCP for high-speed

Transport Layer 3-103

TCP Fairness
fairness goal: if K TCP sessions share same
bottleneck link of bandwidth R, each should have
average rate of R/K

TCP connection
1

bottleneck
router
capacity
TCP connection
R
2

Transport Layer 3-104

Why is TCP fair?
two competing sessions:
▪ additive increase gives slope of 1, as throughout increases
▪ multiplicative decrease decreases throughput proportionally

R equal bandwidth share

Connection 2 throughput

loss: decrease window by factor of 2

congestion avoidance: additive increase
loss: decrease window by factor of 2
congestion avoidance: additive increase

Connection 1 throughput R
Transport Layer 3-105
Fairness (more)
Fairness and UDP Fairness, parallel TCP
▪ multimedia apps often connections
do not use TCP ▪ application can open
• do not want rate multiple parallel
throttled by connections between
congestion control
two hosts
▪ instead use UDP:
• send audio/video at
▪ web browsers do this
constant rate, tolerate ▪ e.g., link of rate R with
packet loss 9 existing connections:
• new app asks for 1 TCP, gets
rate R/10
• new app asks for 11 TCPs,
gets R/2

Transport Layer 3-106

Explicit Congestion Notification (ECN)
network-assisted congestion control:
▪ two bits in IP header (ToS field) marked by network
router to indicate congestion
▪ congestion indication carried to receiving host
▪ receiver (seeing congestion indication in IP datagram) )
sets ECE bit on receiver-to-sender ACK segment to
notify sender of congestion
TCP ACK segment
sourc destination
e
application application
ECE=1
transport transport
network network
link link
physical physical

ECN=00 ECN=11

IP datagram
Transport Layer 3-107
Chapter 3: summary
▪ principles behind
transport layer services: next:
• multiplexing, ▪ leaving the
demultiplexing network “edge”
• reliable data transfer (application,
• flow control transport layers)
• congestion control ▪ into the network
▪ instantiation, “core”
implementation in the ▪ two network layer
Internet chapters:
• UDP • data plane
• TCP • control plane
Transport Layer 3-108