Chapter 3

Transport Layer

A note on the use of these ppt slides:

We’re making these slides freely available to all (faculty, students, readers). Computer
They’re in PowerPoint form so you can add, modify, and delete slides
(including this one) and slide content to suit your needs. They obviously
Networking: A Top
represent a lot of work on our part. In return for use, we only ask the Down Approach
following:
 If you use these slides (e.g., in a class) in substantially unaltered form, 5th edition.
that you mention their source (after all, we’d like people to use our book!) Jim Kurose, Keith
 If you post any slides in substantially unaltered form on a www site, that
you note that they are adapted from (or perhaps identical to) our slides, and Ross
note our copyright of this material. Addison-Wesley, April
Thanks and enjoy! JFK/KWR 2009.
All material copyright 1996-2009
J.F Kurose and K.W. Ross, All Rights Reserved
Transport Layer 3-1
Chapter 3: Transport Layer
Our goals:
 understand  learn about transport
principles behind layer protocols in the
transport layer Internet:
services:  UDP: connectionless
 multiplexing/ transport
demultiplexing  TCP: connection-oriented
 reliable data transport
transfer  TCP congestion control
 flow control
 congestion control

Transport Layer 3-2

Chapter 3 outline
 3.1 Transport-layer  3.5 Connection-
services oriented transport:
 3.2 Multiplexing and TCP
demultiplexing  segment structure
 3.3 Connectionless
 reliable data transfer
 flow control
transport: UDP
 connection
 3.4 Principles of
management
reliable data transfer  3.6 Principles of
congestion control
 3.7 TCP congestion
control
Transport Layer 3-3
Transport services and
protocols applicatio
n
 provide logical transport
network
communication between app data link
physical

lo
processes running on

gi
ca
different hosts

enl
 transport protocols run in

d-
en
end systems

d
tr
 send side: breaks app

a ns
po
messages into segments,

tr
passes to network layer applicatio
 rcv side: reassembles n
transport
segments into messages, network
data link
passes to app layer physical

 more than one transport

protocol available to apps
 Internet: TCP and UDP

Transport Layer 3-4

Transport vs. network layer
 network layer: Household analogy:
logical 12 kids sending letters
communication to 12 kids
between hosts  processes = kids
 transport layer:  app messages =
logical letters in envelopes
communication  hosts = houses
between processes  transport protocol =
 relies on, enhances, Ann and Bill
network layer services  network-layer protocol
= postal service

Transport Layer 3-5

Internet transport-layer
protocols
 reliable, in-order applicatio
n
transport
delivery (TCP) network
data link
physical network
 congestion control

lo
data link
network

gi
physical
data link

ca
 flow control physical

l en
 connection setup

d-
en
 unreliable, unordered network

d
data link

tr
a
physicalnetwork
delivery: UDP

ns
data link

po
physical

r
 no-frills extension of

t
network
data link
applicatio
“best-effort” IP physical network
data link
n
transport
 services not available: physical network
data link
physical
 delay guarantees
 bandwidth guarantees

Transport Layer 3-6

Chapter 3 outline
 3.1 Transport-layer  3.5 Connection-
services oriented transport:
 3.2 Multiplexing and TCP
demultiplexing  segment structure
 3.3 Connectionless
 reliable data transfer
 flow control
transport: UDP
 connection
 3.4 Principles of
management
reliable data transfer  3.6 Principles of
congestion control
 3.7 TCP congestion
control
Transport Layer 3-7
 Multiplexing/demultiplexing
Demultiplexing at rcv host: Multiplexing at send host:
gathering data from multiple
delivering received segments
sockets, enveloping data with
to correct socket
header (later used for
demultiplexing)
= socket = process

P3 P1
P1 P2 P4 application
application application

transport transport transport

network network network

link link link

physical physical physical

host 2 host 3
host 1
Transport Layer 3-8
How demultiplexing works
 host receives IP datagrams
 each datagram has source IP
address, destination IP address 32 bits
 each datagram carries 1
source port # dest port #
transport-layer segment
 each segment has source,
destination port number other header fields
 host uses IP addresses & port
numbers to direct segment to
appropriate socket
application
data
(message)

TCP/UDP segment format

Transport Layer 3-9

Connectionless
demultiplexing
 When host receives
 Create sockets with port
UDP segment:
numbers:
DatagramSocket mySocket1 = new
 checks destination port
DatagramSocket(12534); number in segment
DatagramSocket mySocket2 = new  directs UDP segment to
DatagramSocket(12535); socket with that port
 UDP socket identified by number
 IP datagrams with
two-tuple:
(dest IP address, dest port number) different source IP
addresses and/or
source port numbers
directed to same
socket
Transport Layer 3-10
 Connectionless demux (cont)
DatagramSocket serverSocket = new DatagramSocket(6428);

P2 P1
P1
P3

SP: 6428 SP: 6428

DP: 9157 DP: 5775

SP: 9157 SP: 5775

client DP: 6428 DP: 6428 Client
server
IP: A IP: C IP:B

SP provides “return address”

Transport Layer 3-11

 Connection-oriented demux
 TCP socket identified  Server host may
by 4-tuple: support many
 source IP address simultaneous TCP
 source port number sockets:
 dest IP address  each socket identified
 dest port number by its own 4-tuple
 receiving host uses all  Web servers have
four values to direct different sockets for
segment to each connecting client
appropriate socket  non-persistent HTTP will
have different socket for
each request

Transport Layer 3-12

Connection-oriented demux
(cont)

P1 P4 P5 P6 P2 P1P3

SP: 5775
DP: 80
S-IP: B
D-IP:C

SP: 9157 SP: 9157

client DP: 80 DP: 80 Client
server
IP: A S-IP: A S-IP: B IP:B
IP: C
D-IP:C D-IP:C

Transport Layer 3-13

Connection-oriented demux:
Threaded Web Server

P1 P4 P2 P1P3

SP: 5775
DP: 80
S-IP: B
D-IP:C

SP: 9157 SP: 9157

client DP: 80 DP: 80 Client
server
IP: A S-IP: A S-IP: B IP:B
IP: C
D-IP:C D-IP:C

Transport Layer 3-14

Chapter 3 outline
 3.1 Transport-layer  3.5 Connection-
services oriented transport:
 3.2 Multiplexing and TCP
demultiplexing  segment structure
 3.3 Connectionless
 reliable data transfer
 flow control
transport: UDP
 connection
 3.4 Principles of
management
reliable data transfer  3.6 Principles of
congestion control
 3.7 TCP congestion
control
Transport Layer 3-15
UDP: User Datagram Protocol [RFC
768]
 “no frills,” “bare bones”
Internet transport protocol Why is there a UDP?
 “best effort” service, UDP
 no connection
segments may be:
establishment (which can
 lost
add delay)
 delivered out of order
 simple: no connection
to app state at sender, receiver
 connectionless:
 small segment header
 no handshaking
 no congestion control:
between UDP sender,
UDP can blast away as
receiver
fast as desired
 each UDP segment
handled independently
of others

Transport Layer 3-16

UDP: more
 often used for streaming
multimedia apps 32 bits
 loss tolerant
Length, in source port # dest port #
 rate sensitive bytes of UDP length checksum
segment,
 other UDP uses
including
 DNS header
 SNMP
 reliable transfer over Application
UDP: add reliability at data
application layer (message)
 application-specific
error recovery!
UDP segment format

Transport Layer 3-17

UDP checksum
Goal: detect “errors” (e.g., flipped bits) in
transmitted segment

Sender: Receiver:
 treat segment contents  compute checksum of
as sequence of 16-bit received segment
integers  check if computed checksum
 checksum: addition (1’s equals checksum field value:
 NO - error detected
complement sum) of
segment contents  YES - no error detected.

 sender puts checksum But maybe errors

nonetheless? More later
value into UDP
….
checksum field

Transport Layer 3-18

 Internet Checksum Example
 Note
 When adding numbers, a carryout from the
most significant bit needs to be added to
the result
 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

sum 1 1 0 1 1 1 0 1 1 1 0 1 1 1 1 0 0
checksum 1 0 1 0 0 0 1 0 0 0 1 0 0 0 0 1 1
Transport Layer 3-19
Chapter 3 outline
 3.1 Transport-layer  3.5 Connection-
services oriented transport:
 3.2 Multiplexing and TCP
demultiplexing  segment structure
 3.3 Connectionless
 reliable data transfer
 flow control
transport: UDP
 connection
 3.4 Principles of
management
reliable data transfer  3.6 Principles of
congestion control
 3.7 TCP congestion
control
Transport Layer 3-20
Principles of Reliable data
transfer
 important in app., 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 app., 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 app., 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
(e.g., by app.). Passed data to by rdt to deliver data to
deliver to receiver upper layer upper

send receive
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 transition
state: when in this
“state” next state state state
uniquely determined 1 event
2
by next event actions

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 read 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)

sender receiver

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
 new mechanisms in rdt2.0 (beyond rdt1.0):
 error detection
 receiver feedback: control msgs (ACK,NAK) rcvr-
>sender

Transport Layer 3-27

rdt2.0: FSM specification
rdt_send(data)
snkpkt = make_pkt(data, checksum) receiver
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

call from
below
sender
rdt_rcv(rcvpkt) &&
notcorrupt(rcvpkt)
extract(rcvpkt,data)
deliver_data(data)
udt_send(ACK)

Transport Layer 3-28

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
 call from
below

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

Transport Layer 3-29

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
 call from
below

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

Transport Layer 3-30

rdt2.0 has a fatal flaw!
What happens if Handling duplicates:
ACK/NAK corrupted?  sender retransmits current
 sender doesn’t know pkt if ACK/NAK garbled
 sender adds sequence
what happened at
receiver! number to each pkt
 can’t just retransmit:  receiver discards (doesn’t

possible duplicate deliver up) duplicate pkt

stop and wait

Sender sends one packet,
then waits for receiver
response

Transport Layer 3-31

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)
&& notcorrupt(rcvpkt) rdt_rcv(rcvpkt)
&& isACK(rcvpkt) && notcorrupt(rcvpkt)
&& isACK(rcvpkt)


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 3-32

 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) && rdt_rcv(rcvpkt) &&
(corrupt(rcvpkt) (corrupt(rcvpkt)
sndpkt = make_pkt(NAK, chksum) sndpkt = make_pkt(NAK, chksum)
udt_send(sndpkt) udt_send(sndpkt)
Wait for Wait for
rdt_rcv(rcvpkt) && 0 from 1 from rdt_rcv(rcvpkt) &&
not corrupt(rcvpkt) && below below not corrupt(rcvpkt) &&
has_seq1(rcvpkt) 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 3-33

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

Transport Layer 3-34

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 3-35

 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
fragment rdt_rcv(rcvpkt)
&& notcorrupt(rcvpkt)
rdt_rcv(rcvpkt) && && isACK(rcvpkt,0)
(corrupt(rcvpkt) || 
has_seq1(rcvpkt)) Wait for receiver FSM
0 from
udt_send(sndpkt) 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 3-36
rdt3.0: channels with errors and loss

New assumption: Approach: sender waits

underlying channel “reasonable” amount
can also lose packets of time for ACK
(data or ACKs)  retransmits if no ACK
 checksum, seq. #, received in this time
ACKs, retransmissions  if pkt (or ACK) just delayed
will be of help, but not (not lost):
enough  retransmission will be
duplicate, but use of seq.
#’s already handles this
 receiver must specify
seq # of pkt being
ACKed
 requires countdown timer
Transport Layer 3-37
rdt3.0 sender
rdt_send(data)
rdt_rcv(rcvpkt) &&
sndpkt = make_pkt(0, data, checksum) ( corrupt(rcvpkt) ||
udt_send(sndpkt) isACK(rcvpkt,1) )
rdt_rcv(rcvpkt) start_timer 
 Wait for Wait
for timeout
call 0from
ACK0 udt_send(sndpkt)
above
start_timer
rdt_rcv(rcvpkt)
&& notcorrupt(rcvpkt) rdt_rcv(rcvpkt)
&& isACK(rcvpkt,1) && notcorrupt(rcvpkt)
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) 
rdt_rcv(rcvpkt) &&
sndpkt = make_pkt(1, data, checksum)
( corrupt(rcvpkt) || udt_send(sndpkt)
isACK(rcvpkt,0) ) start_timer


Transport Layer 3-38

rdt3.0 in action

Transport Layer 3-39

rdt3.0 in action

Transport Layer 3-40

Performance of rdt3.0
 rdt3.0 works, but performance stinks
 ex: 1 Gbps link, 15 ms prop. delay, 8000 bit packet:

L 8000bits
d trans   9 8 microseconds
R 10 bps
 U sender : utilization – fraction of time sender busy sending

L/ R .008
U = = = 0.00027
sender 30.008
RTT +L / R microsec
 1KB pkt every 30 msec -> 33kB/sec thruput over 1 Gbps link onds
 network protocol limits use of physical resources!

Transport Layer 3-41

 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

L/ R .008
U = = = 0.00027
sender 30.008
RTT +L / R microsec
onds

Transport Layer 3-42

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 3-43
 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

Increase utilization
by a factor of 3!
3*L/ R .024
U = = = 0.0008
sender 30.008
RTT +L / R microsecon
ds
Transport Layer 3-44
Pipelining Protocols
Go-back-N: overview Selective Repeat:
 sender: up to N overview
unACKed pkts in  sender: up to N
pipeline unACKed packets in
 receiver: only sends pipeline
cumulative ACKs  receiver: ACKs individual
 doesn’t ACK pkt if pkts
there’s a gap  sender: maintains timer
 sender: has timer for for each unACKed pkt
oldest unACKed pkt  if timer expires:
 if timer expires: retransmit only unACKed
retransmit all packet
unACKed packets

Transport Layer 3-45

Go-Back-N
Sender:
 k-bit seq # in pkt header
 “window” of up to N, consecutive unACKed pkts allowed

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

 may receive duplicate ACKs (see receiver)
 timer for each in-flight pkt
 timeout(n): retransmit pkt n and all higher seq # pkts in window

Transport Layer 3-46

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++
}
 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 3-47
GBN: receiver extended FSM
default
udt_send(sndpkt) rdt_rcv(rcvpkt)
&& notcurrupt(rcvpkt)
 && 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 3-48

GBN in
action

Transport Layer 3-49

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
 again limits seq #s of sent, unACKed pkts

Transport Layer 3-50

Selective repeat: sender, receiver
windows

Transport Layer 3-51

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

Transport Layer 3-52

Selective repeat in action

Transport Layer 3-53

Selective repeat:
dilemma
Example:
 seq #’s: 0, 1, 2, 3
 window size=3

 receiver sees no
difference in two
scenarios!
 incorrectly passes
duplicate data as new
in (a)

Q: what relationship
between seq # size
and window size?
Transport Layer 3-54
Chapter 3 outline
 3.1 Transport-layer  3.5 Connection-
services oriented transport:
 3.2 Multiplexing and TCP
demultiplexing  segment structure
 3.3 Connectionless
 reliable data transfer
 flow control
transport: UDP
 connection
 3.4 Principles of
management
reliable data transfer  3.6 Principles of
congestion control
 3.7 TCP congestion
control
Transport Layer 3-55
 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) init’s
sender, receiver state
 TCP congestion and flow
before data exchange
control set window size
 flow controlled:
 send
application & receive buffers
w rites data
application
reads data
 sender will not
sock et sock et
door
TCP TCP
door overwhelm receiver
send buffer receive buffer
segm ent

Transport Layer 3-56

 TCP segment structure
32 bits
URG: urgent data counting
(generally not used) source port # dest port #
by bytes
sequence number of data
ACK: ACK #
valid acknowledgement (not segments!)
U A Pnumber
head not
PSH: push data now len used
R S F Receive window
(generally not used) # bytes
checksum Urg data pointer
rcvr willing
RST, SYN, FIN: to accept
Options (variable length)
connection estab
(setup, teardown
commands)
application
Internet data
checksum (variable length)
(as in UDP)

Transport Layer 3-57

 TCP seq. #’s and ACKs
Seq. #’s:
Host A Host B
 byte stream
“number” of first User Seq=4
2, A C
byte in segment’s types K=79,
da t a =
‘C’ ‘C’
data host ACKs
ACKs: receipt of
’
C ‘C’, echoes
 seq # of next byte , d a ta = ‘
3
9 , A CK=4 back ‘C’
expected from e q =7
S
other side
 cumulative ACK host ACKs
receipt Seq=4
Q: how receiver handles of echoed 3, ACK
=80
out-of-order segments ‘C’
 A: TCP spec doesn’t
say, - up to
implementer time
simple telnet scenario

Transport Layer 3-58

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

Transport Layer 3-59

TCP Round Trip Time and
Timeout
EstimatedRTT = (1- )*EstimatedRTT + *SampleRTT

 Exponential weighted moving average

 influence of past sample decreases exponentially fast
 typical value:  = 0.125

Transport Layer 3-60

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

350

300

250
RTT (milliseconds)

200

150

100
1 8 15 22 29 36 43 50 57 64 71 78 85 92 99 106
time (seconnds)

SampleRTT Estimated RTT

Transport Layer 3-61

 TCP Round Trip Time and
Timeout
Setting the timeout
 EstimtedRTT plus “safety margin”
 large variation in EstimatedRTT -> larger safety margin
 first estimate of how much SampleRTT deviates from EstimatedRTT:

DevRTT = (1-)*DevRTT +
*|SampleRTT-EstimatedRTT|

(typically,  = 0.25)

Then set timeout interval:

TimeoutInterval = EstimatedRTT + 4*DevRTT

Transport Layer 3-62

Chapter 3 outline
 3.1 Transport-layer  3.5 Connection-
services oriented transport:
 3.2 Multiplexing and TCP
demultiplexing  segment structure
 3.3 Connectionless
 reliable data transfer
 flow control
transport: UDP
 connection
 3.4 Principles of
management
reliable data transfer  3.6 Principles of
congestion control
 3.7 TCP congestion
control
Transport Layer 3-63
TCP reliable data transfer
 TCP creates rdt  retransmissions are
service on top of IP’s triggered by:
unreliable service  timeout events
 pipelined segments  duplicate ACKs
 cumulative ACKs  initially consider
 TCP uses single simplified TCP
retransmission timer sender:
 ignore duplicate ACKs
 ignore flow control,
congestion control

Transport Layer 3-64

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 acknowledges
 start timer if not
previously unACKed
already running (think segments
of timer as for oldest  update what is known
unACKed segment) to be ACKed
 expiration interval:  start timer if there are
TimeOutInterval outstanding segments

Transport Layer 3-65

NextSeqNum = InitialSeqNum
SendBase = InitialSeqNum

loop (forever) {
TCP
switch(event)
sender
event: data received from application above
create TCP segment with sequence number NextSeqNum
(simplified
if (timer currently not running) )
start timer
pass segment to IP
Comment:
NextSeqNum = NextSeqNum + length(data)
• SendBase-1: last
event: timer timeout cumulatively
retransmit not-yet-acknowledged segment with ACKed byte
smallest sequence number Example:
start timer • SendBase-1 = 71;
y= 73, so the rcvr
event: ACK received, with ACK field value of y wants 73+ ;
if (y > SendBase) {
y > SendBase, so
SendBase = y
if (there are currently not-yet-acknowledged segments)
that new data is
start timer ACKed
}

} /* end of loop forever */

Transport Layer 3-66
 TCP: retransmission scenarios
Host A Host B Host A Host B

Seq=9 Seq=9
2, 8 b 2, 8 b
y t es d y t es d
at a

Seq=92 timeout
at a Seq=
100,
20 b y
t es d
timeout

ata
=100
ACK 0
10
X CK
A AC
=
K =120
loss
Seq=9 Seq=9
2, 8 b 2, 8 b
y t es d Sendbase y t es d
at a
at a

Seq=92 timeout
= 100
SendBase
= 120 =1 20
K
CK =100 AC
A

SendBase
= 100 SendBase
= 120 premature timeout
time time
lost ACK scenario
Transport Layer 3-67
 TCP retransmission scenarios
(more)
Host A Host B

Seq=9
2, 8 b
y t es d
at a

=100
timeout

Seq=1 A CK
00 , 2 0
b y t es
dat a
X
loss

SendBase CK =120
A
= 120

time
Cumulative ACK scenario

Transport Layer 3-68

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 ACKed send ACK

Arrival of in-order segment with Immediately send single cumulative

expected seq #. One other ACK, ACKing both in-order segments
segment has ACK pending

Arrival of out-of-order segment Immediately send duplicate ACK,

higher-than-expect seq. # . indicating seq. # of next expected 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 3-69

Fast Retransmit
 time-out period  If sender receives 3
often relatively long: ACKs for same data, it
 long delay before assumes that
resending lost packet segment after ACKed
 detect lost segments data was lost:
via duplicate ACKs.  fast retransmit: resend
 sender often sends segment before timer
many segments back- expires
to-back
 if segment is lost,
there will likely be
many duplicate ACKs
for that segment

Transport Layer 3-70

 Host A Host B

seq # x1
seq # x2
seq # x3
ACK x1
seq # x4 X
seq # x5
ACK x1
ACK x1
ACK x1
triple
duplicate
ACKs re s e n
d s eq
X2
timeout

time

Transport Layer 3-71

 Fast retransmit algorithm:

event: ACK received, with ACK field value of y

if (y > SendBase) {
SendBase = y
if (there are currently not-yet-acknowledged segments)
start timer
}
else {
increment count of dup ACKs received for y
if (count of dup ACKs received for y = 3) {
resend segment with sequence number y
}

a duplicate ACK for fast retransmit

already ACKed segment

Transport Layer 3-72

Chapter 3 outline
 3.1 Transport-layer  3.5 Connection-
services oriented transport:
 3.2 Multiplexing and TCP
demultiplexing  segment structure
 3.3 Connectionless
 reliable data transfer
 flow control
transport: UDP
 connection
 3.4 Principles of
management
reliable data transfer  3.6 Principles of
congestion control
 3.7 TCP congestion
control
Transport Layer 3-73
 TCP Flow Control
flow control
sender won’t
 receive side of TCP
overflow
connection has a receiver’s buffer by
receive buffer: transmitting too
much,
too fast
(currently)  speed-matching
IP TCP data application
unused buffer
datagrams space
(in buffer) process service: matching
send rate to
receiving
application’s drain
 app process may be
rate
slow at reading from
buffer
Transport Layer 3-74
 TCP Flow control: how it
works
(currently)  receiver: advertises
IP TCP data application
unused buffer
datagrams space
(in buffer) process unused buffer space
by including rwnd
rwnd value in segment
RcvBuffer
header
(suppose TCP receiver  sender: limits # of
discards out-of-order unACKed bytes to
segments) rwnd
 unused buffer space:  guarantees receiver’s
buffer doesn’t overflow
= rwnd
= RcvBuffer-[LastByteRcvd -
LastByteRead]

Transport Layer 3-75

Chapter 3 outline
 3.1 Transport-layer  3.5 Connection-
services oriented transport:
 3.2 Multiplexing and TCP
demultiplexing  segment structure
 3.3 Connectionless
 reliable data transfer
 flow control
transport: UDP
 connection
 3.4 Principles of
management
reliable data transfer  3.6 Principles of
congestion control
 3.7 TCP congestion
control
Transport Layer 3-76
TCP Connection Management
Recall: TCP sender, receiver Three way handshake:
establish “connection”
before exchanging data Step 1: client host sends TCP
segments SYN segment to server
 initialize TCP variables:  specifies initial seq #
 seq. #s  no data
 buffers, flow control info
Step 2: server host receives
(e.g. RcvWindow) SYN, replies with SYNACK
 client: connection initiator segment
Socket clientSocket = new
Socket("hostname","port
 server allocates buffers
number");  specifies server initial

 server: contacted by client seq. #

Socket connectionSocket = Step 3: client receives SYNACK,
welcomeSocket.accept(); replies with ACK segment,
which may contain data

Transport Layer 3-77

TCP Connection Management (cont.)

Closing a connection: client server

client closes socket: close

FIN
clientSocket.close();

Step 1: client end system

ACK
sends TCP FIN control close
segment to server FIN

Step 2: server receives FIN,

timed wait
ACK
replies with ACK. Closes
connection, sends FIN.

closed

Transport Layer 3-78

TCP Connection Management (cont.)

Step 3: client receives FIN, client server

replies with ACK.
closing
FIN
 Enters “timed wait” - will
respond with ACK to
received FINs
ACK
closing
Step 4: server, receives ACK. FIN
Connection closed.

Note: with small

timed wait
ACK
modification, can handle
simultaneous FINs. closed

closed

Transport Layer 3-79

TCP Connection Management
(cont)

TCP server
lifecycle

TCP client
lifecycle

Transport Layer 3-80

Chapter 3 outline
 3.1 Transport-layer  3.5 Connection-
services oriented transport:
 3.2 Multiplexing and TCP
demultiplexing  segment structure
 3.3 Connectionless
 reliable data transfer
 flow control
transport: UDP
 connection
 3.4 Principles of
management
reliable data transfer  3.6 Principles of
congestion control
 3.7 TCP congestion
control
Transport Layer 3-81
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 3-82

 Causes/costs of congestion: scenario
1
Host A out
in : original data
 two senders, two
receivers
Host B unlimited shared
 one router, output link buffers

infinite buffers
 no
retransmission

 large delays
when congested
 maximum
achievable
throughput

Transport Layer 3-83

 Causes/costs of congestion: scenario
2
 one router, finite buffers
 sender retransmission of lost packet

Host A in : original data out

'in : original data, plus

retransmitted data

Host B finite shared output

link buffers

Transport Layer 3-84

 Causes/costs of congestion: scenario
2
 always: = 
 (goodput)
in out
 > out
 “perfect” retransmission only when loss:
in

 retransmission of delayed (not lost) packet makes
in
larger (than perfect case) forsame
out
R/2 R/2 R/2

R/3
out

out

out
R/4

R/2 R/2 R/2

in in in

a. b. c.
“costs” of congestion:
 more work (retrans) for given “goodput”
 unneeded retransmissions: link carries multiple copies of pkt

Transport Layer 3-85

Causes/costs of congestion: scenario
3
 four senders
Q: what happens as
 multihop paths in
and increase ?
 timeout/retransmit in
Host A out
in : original data
'in : original data, plus
retransmitted data
finite shared output
link buffers

Host B

Transport Layer 3-86

Causes/costs of congestion: scenario
3
H 
o
s o
t
u
A
t

H
o
s
t
B

another “cost” of congestion:

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

Transport Layer 3-87

Approaches towards congestion
control
two broad approaches towards congestion control:

end-end congestion network-assisted

control: congestion control:
 no explicit feedback from  routers provide feedback
network to end systems
 congestion inferred from  single bit indicating
end-system observed congestion (SNA,
loss, delay DECbit, TCP/IP ECN,
 approach taken by TCP ATM)
 explicit rate sender
should send at

Transport Layer 3-88

Case study: ATM ABR congestion
control
ABR: available bit RM (resource
rate: management) cells:
 “elastic service”  sent by sender, interspersed
 if sender’s path with data cells
 bits in RM cell set by switches
“underloaded”:
 sender should use (“network-assisted”)
 NI bit: no increase in rate
available bandwidth
 if sender’s path
(mild congestion)
 CI bit: congestion
congested:
indication
 sender throttled to
 RM cells returned to sender
minimum
by receiver, with bits intact
guaranteed rate

Transport Layer 3-89

Case study: ATM ABR congestion
control

 two-byte ER (explicit rate) field in RM cell

 congested switch may lower ER value in cell
 sender’ send rate thus maximum supportable rate on
path
 EFCI bit in data cells: set to 1 in congested switch
 if data cell preceding RM cell has EFCI set, sender sets CI
bit in returned RM cell
Transport Layer 3-90
Chapter 3 outline
 3.1 Transport-layer  3.5 Connection-
services oriented transport:
 3.2 Multiplexing and TCP
demultiplexing  segment structure
 3.3 Connectionless
 reliable data transfer
 flow control
transport: UDP
 connection
 3.4 Principles of
management
reliable data transfer  3.6 Principles of
congestion control
 3.7 TCP congestion
control
Transport Layer 3-91
TCP congestion control:
 goal: TCP sender should transmit as fast as possible, but without congesting network
 Q: how to find rate just below congestion level

 decentralized: each TCP sender sets its own rate, based on implicit feedback:
 ACK: segment received (a good thing!), network not congested, so increase sending
rate
 lost segment: assume loss due to congested network, so decrease sending rate

Transport Layer 3-92

TCP congestion control: bandwidth
probing
 “probing for bandwidth”: increase transmission rate on receipt
of ACK, until eventually loss occurs, then decrease transmission
rate
 continue to increase on ACK, decrease on loss (since available bandwidth
is changing, depending on other connections in network)

ACKs being received,

Xloss, so decrease rate
so increase rate
X
X
sending rate

X
TCP’s
X “sawtooth”
behavior

time

 Q: how fast to increase/decrease?

 details to follow
Transport Layer 3-93
TCP Congestion Control: details
 sender limits rate by limiting number of
unACKed bytes “in pipeline”:
LastByteSent-LastByteAcked  cwnd
 cwnd: differs from rwnd (how, why?)
 sender limited by min(cwnd,rwnd)
 roughly, cwnd
bytes

cwnd
rate = bytes/sec
RTT
 cwnd is dynamic, function of perceived
RTT
network congestion
ACK(s)

Transport Layer 3-94

TCP Congestion Control: more
details
segment loss event: ACK received: increase
reducing cwnd cwnd
 timeout: no response  slowstart phase:
from receiver  increase exponentially
 cut cwnd to 1 fast (despite name) at
 3 duplicate ACKs: at connection start, or
following timeout
least some segments
 congestion avoidance:
getting through (recall
 increase linearly
fast retransmit)
 cut cwnd in half, less
aggressively than on
timeout

Transport Layer 3-95

 TCP Slow Start
 when connection begins, cwnd
= 1 MSS Host A Host B
 example: MSS = 500 bytes
& RTT = 200 msec one s e gm
ent

RTT
 initial rate = 20 kbps
 available bandwidth may be
two segm
en ts
>> MSS/RTT
 desirable to quickly ramp up
to respectable rate
four segm
 increase rate exponentially ents
until first loss event or when
threshold reached
 double cwnd every RTT
 done by incrementing cwnd
time
by 1 for every ACK received

Transport Layer 3-96

Transitioning into/out of
slowstart
ssthresh: cwnd threshold maintained by TCP
 on loss event: set ssthresh to cwnd/2
 remember (half of) TCP rate when congestion last occurred
 when cwnd >= ssthresh: transition from slowstart to
congestion avoidance phase

duplicate ACK
dupACKcount++ new ACK
cwnd = cwnd+MSS
dupACKcount = 0
 transmit new segment(s),as allowed
cwnd = 1 MSS
ssthresh = 64 KB cwnd > ssthresh
dupACKcount = 0
slow  congestion
start timeout avoidance
ssthresh = cwnd/2
cwnd = 1 MSS
dupACKcount = 0
timeout
retransmit missing segment
ssthresh = cwnd/2
cwnd = 1 MSS
dupACKcount = 0
retransmit missing segment

Transport Layer 3-97

TCP: congestion avoidance
 when cwnd > ssthresh AIMD
grow cwnd linearly
 ACKs: increase cwnd
 increase cwnd by 1
by 1 MSS per RTT:
MSS per RTT
additive increase
 approach possible
 loss: cut cwnd in half
congestion slower
than in slowstart
(non-timeout-
 implementation: cwnd
detected loss ):
multiplicative
= cwnd + MSS/cwnd
decrease
for each ACK received AIMD: Additive Increase
Multiplicative Decrease

Transport Layer 3-98

TCP congestion control FSM: overview

slow cwnd > ssthresh

congestion
start avoidance
loss:
timeout
loss:
timeout

loss: new ACK loss:

timeout 3dupACK
fast
loss: recovery
3dupACK

Transport Layer 3-99

TCP congestion control FSM: details

duplicate ACK
dupACKcount++ new ACK
new ACK
.
cwnd = cwnd + MSS (MSS/cwnd)
dupACKcount = 0
cwnd = cwnd+MSS transmit new segment(s),as allowed
dupACKcount = 0
 transmit new segment(s),as allowed
cwnd = 1 MSS
ssthresh = 64 KB cwnd > ssthresh
dupACKcount = 0
slow  congestion
start timeout avoidance
ssthresh = cwnd/2
cwnd = 1 MSS duplicate ACK
dupACKcount = 0
timeout dupACKcount++
retransmit missing segment
ssthresh = cwnd/2
cwnd = 1 MSS
dupACKcount = 0
retransmit missing segment
timeout
ssthresh = cwnd/2
cwnd = 1 New ACK
dupACKcount = 0
retransmit missing segment cwnd = ssthresh dupACKcount == 3
dupACKcount == 3 dupACKcount = 0
ssthresh= cwnd/2 ssthresh= cwnd/2
cwnd = ssthresh + 3 cwnd = ssthresh + 3
retransmit missing segment retransmit missing segment
fast
recovery
duplicate ACK
cwnd = cwnd + MSS
transmit new segment(s), as allowed

Transport Layer 3-100

Popular “flavors” of TCP

TCP Reno
cwnd window size (in

ssthresh

ssthresh
segments)

TCP Tahoe

Transmission round

Transport Layer 3-101

Summary: TCP Congestion Control
 when cwnd < ssthresh, sender in slow-start
phase, window grows exponentially.
 when cwnd >= ssthresh, sender is in
congestion-avoidance phase, window grows
linearly.
 when triple duplicate ACK occurs, ssthresh
set to cwnd/2, cwnd set to ~ ssthresh
 when timeout occurs, ssthresh set to cwnd/2,
cwnd set to 1 MSS.

Transport Layer 3-102

TCP throughput
 Q: what’s average throughout of TCP as
function of window size, RTT?
 ignoring slow start
 let W be window size when loss occurs.
 when window is W, throughput is
W/RTT
 just after loss, window drops to W/2,
throughput to W/2RTT.
 average throughout: .75 W/RTT

Transport Layer 3-103

TCP Futures: TCP over “long, fat
pipes”
 example: 1500 byte segments, 100ms RTT,
want 10 Gbps throughput
 requires window size W = 83,333 in-flight
segments
 throughput in terms of loss rate:

1.22 MSS
RTT L
 ➜ L = 2·10-10 Wow
 new versions of TCP for high-speed

Transport Layer 3-104

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
TCP
router
connection 2
capacity R

Transport Layer 3-105

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 3-106

Fairness (more)
Fairness and UDP Fairness and parallel TCP
 multimedia apps connections
 nothing prevents app
often do not use TCP
 do not want rate from opening parallel
throttled by connections between 2
congestion control hosts.
 instead use UDP:  web browsers do this
 pump audio/video at  example: link of rate R
constant rate, tolerate supporting 9
packet loss
connections;
 new app asks for 1 TCP,
gets rate R/10
 new app asks for 11 TCPs,
gets R/2 !

Transport Layer 3-107

Chapter 3: Summary
 principles behind transport
layer services:
 multiplexing,
demultiplexing
 reliable data transfer
 flow control
 congestion control
Next:
 instantiation and  leaving the network
implementation in the Internet “edge” (application,
 UDP
transport layers)
 TCP
 into the network
“core”

Transport Layer 3-108