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:
▪ 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!)
▪ 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.
Computer Networking: A
For a revision history, see the slide note for this page.
Top-Down
th
Approach
Thanks and enjoy! JFK/KWR 8 edition
All material copyright 1996-2020 Jim Kurose, Keith Ross
J.F Kurose and K.W. Ross, All Rights Reserved Pearson, 2020
Transport Layer: 3-1
Transport layer: overview
Our goal:
▪ understand principles ▪ learn about Internet transport
behind transport layer layer protocols:
services: • UDP: connectionless transport
• multiplexing, • TCP: connection-oriented reliable
demultiplexing transport
• reliable data transfer • TCP congestion control
• flow control
• congestion control

Transport Layer: 3-2

Transport layer: roadmap
▪ Transport-layer services
▪ Multiplexing and demultiplexing
▪ Connectionless transport: UDP
▪ Principles of reliable data transfer
▪ Connection-oriented transport: TCP
▪ Principles of congestion control
▪ TCP congestion control
▪ Evolution of transport-layer
functionality
Transport Layer: 3-3
Transport services and protocols
application
transport

▪ provide logical communication

network
mobile network
data link
physical

between application processes national or global ISP

log
running on different hosts

ica
l en
▪ transport protocols actions in end

d-e
systems:

nd
local or

tra
• sender: breaks application messages regional ISP

nsp
into segments, passes to network layer

ort
home network content
• receiver: reassembles segments into provider
network datacenter
messages, passes to application layer application
transport
network
network

▪ two transport protocols available to data link

physical

Internet applications enterprise

network
• TCP, UDP
Transport Layer: 3-4
Transport vs. network layer services and protocols
household analogy:
12 kids in Ann’s house sending
letters to 12 kids in Bill’s
house:
▪ hosts = houses
▪ processes = kids
▪ app messages = letters in
envelopes
▪ transport protocol = Ann and Bill
who demux to in-house siblings
▪ network-layer protocol = postal
service

Transport Layer: 3-5

Transport vs. network layer services and protocols
household analogy:
▪transport layer: 12 kids in Ann’s house sending
communication between letters to 12 kids in Bill’s
processes house:
• relies on, enhances, network ▪ hosts = houses
layer services ▪ processes = kids
▪ app messages = letters in
envelopes
▪network layer: ▪ transport protocol = Ann and Bill
communication between who demux to in-house siblings
hosts ▪ network-layer protocol = postal
service

Transport Layer: 3-6

Transport Layer Actions

Sender:
application ▪ is passed an application
app. msg
application-layer message
transport ▪ determines segment TThhtransport
app. msg
header fields values
network (IP) ▪ creates segment network (IP)

link ▪ passes segment to IP link

physical physical

Transport Layer: 3-7

Transport Layer Actions

Receiver:
application ▪ receives segment from IP application
▪ checks header values
transport
transport
app. msg ▪ extracts application-layer
message
network (IP)
network (IP) ▪ demultiplexes message up
link to application via socket link

physical physical
Th app. msg

Transport Layer: 3-8

Two principal Internet transport
protocols application

▪TCP: Transmission Control Protocol

transport
network
mobile network
data link
physical
national or global ISP
• reliable, in-order delivery

log
• congestion control

ica
l en
• flow control

d-e
• connection setup

nd
▪UDP: User Datagram Protocol
local or

tra
regional ISP

nsp
• unreliable, unordered delivery

ort
home network content
provider
• no-frills extension of “best-effort” IP network datacenter
application

▪services not available:

network
transport
network
data link

• delay guarantees physical

• bandwidth guarantees enterprise

network

Transport Layer: 3-9

Chapter 3: roadmap
▪ Transport-layer services
▪ Multiplexing and demultiplexing
▪ Connectionless transport: UDP
▪ Principles of reliable data transfer
▪ Connection-oriented transport: TCP
▪ Principles of congestion control
▪ TCP congestion control
▪ Evolution of transport-layer
functionality
Transport Layer: 3-10
 HTTP server
client
application application
HTTP msg
transport

transport network transport

network link network
link physical link
physical physical

Transport Layer: 3-11

 HTTP server
client
application application
HTTP msg
transport
Ht HTTP msg

transport network transport

network link network
link physical link
physical physical

Transport Layer: 3-12

 HTTP server
client
application application
HTTP msg
transport
Ht HTTP msg

transport Hnnetwork
Ht HTTP msg transport
network link network
link physical link
physical physical

Transport Layer: 3-13

 HTTP server
client
application application

transport

transport network transport

network link network
link physical link
physical physical

Hn Ht HTTP msg

Transport Layer: 3-14

 HTTP server
client
application application
HTTP msg
transport
Ht HTTP msg

transport network transport

Hn Hnetwork link network
t HTTP msg

link physical link

physical physical

Transport Layer: 3-15

 HTTP server
client
application application
HTTP msg
transport
Ht HTTP msg

Htransport network transport

t HTTP msg

network link network

link physical link
physical physical

Transport Layer: 3-16

 Q: how did transport layer know to deliver message to Firefox
browser process rather then Netflix process or Skype process?

client
application application
HTTP msg
HTTP msg transport
Ht HTTP msg

transport network transport

network link network
link physical link
physical physical

Transport Layer: 3-17

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

application

application P1 P2 application socket

P3 transport P4
process
transport network transport
network link network
link physical link
physical physical

Transport Layer: 3-18

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

Transport Layer: 3-19

Connectionless demultiplexing
Recall: when receiving host receives
UDP segment:
▪ when creating socket, must • checks destination port # in
specify host-local port #: segment
DatagramSocket mySocket1 • directs UDP segment to
= new socket with that port #
DatagramSocket(12534);
▪ when creating datagram to
send into UDP socket, must
specify IP/UDP datagrams with same dest.
port #, but different source IP
• destination IP address addresses and/or source port
• destination port # numbers will be directed to same
socket at receiving host
Transport Layer: 3-20
Connectionless demultiplexing: an example
mySocket =
socket(AF_INET,SOCK_DGRAM)
mySocket.bind(myaddr,6428);
mySocket = mySocket =
socket(AF_INET,SOCK_STREAM) socket(AF_INET,SOCK_STREAM)
mySocket.bind(myaddr,9157); mySocket.bind(myaddr,5775);
application
application application
P1
P3 P4
transport
transport transport
network
network link network
link physical link
physical physical

B D
source port: 6428 source port: ?
dest port: 9157 dest port: ?

A C
source port: 9157 source port: ?
dest port: 6428 dest port: ?
Connection-oriented demultiplexing
▪ TCP socket identified by ▪ server may support many
4-tuple: simultaneous TCP sockets:
• source IP address • each socket identified by its
• source port number own 4-tuple
• dest IP address • each socket associated with
• dest port number a different connecting client
▪ demux: receiver uses all
four values (4-tuple) to
direct segment to
appropriate socket
Transport Layer: 3-22
Connection-oriented demultiplexing: example
application
application P4 P5 P6 application
P1 P2 P3
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: A,9157 source IP,port: C,5775 address C
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 sockets
Transport Layer: 3-23
Summary
▪ Multiplexing, demultiplexing: based on segment, datagram
header field values
▪ UDP: demultiplexing using destination port number (only)
▪ TCP: demultiplexing using 4-tuple: source and destination IP
addresses, and port numbers
▪ Multiplexing/demultiplexing happen at all layers

Transport Layer: 3-24

Chapter 3: roadmap
▪ Transport-layer services
▪ Multiplexing and demultiplexing
▪ Connectionless transport: UDP
▪ Principles of reliable data transfer
▪ Connection-oriented transport: TCP
▪ Principles of congestion control
▪ TCP congestion control
▪ Evolution of transport-layer
functionality
Transport Layer: 3-25
UDP: User Datagram Protocol
Why is there a UDP?
▪ “no frills,” “bare bones”
Internet transport protocol ▪ no connection
establishment (which can
▪ “best effort” service, UDP add RTT delay)
segments may be: ▪ simple: no connection state
• lost at sender, receiver
• delivered out-of-order to app ▪ small header size
▪ connectionless: ▪ no congestion control
▪ UDP can blast away as fast as
• no handshaking between UDP desired!
sender, receiver ▪ can function in the face of
• each UDP segment handled congestion
independently of others
Transport Layer: 3-26
UDP: User Datagram Protocol
▪ UDP use:
▪ streaming multimedia apps (loss tolerant, rate sensitive)
▪ DNS
▪ SNMP
▪ HTTP/3
▪ if reliable transfer needed over UDP (e.g., HTTP/3):
▪ add needed reliability at application layer
▪ add congestion control at application layer

Transport Layer: 3-27

UDP: User Datagram Protocol [RFC 768]

Transport Layer: 3-28

UDP: Transport Layer Actions

SNMP client SNMP server

application application

transport transport
(UDP) (UDP)

network (IP) network (IP)

link link

physical physical

Transport Layer: 3-29

UDP: Transport Layer Actions

SNMP client SNMP server

UDP sender actions:
application ▪ is passed an application
SNMP msg
application-layer message
transport
transport ▪ determines UDP segment UDP
UDPhh SNMP msg
(UDP) (UDP)
header fields values
network (IP) ▪ creates UDP segment network (IP)

link ▪ passes segment to IP link

physical physical

Transport Layer: 3-30

UDP: Transport Layer Actions

SNMP client SNMP server

UDP receiver actions:
application ▪ receives segment from IP application
▪ checks UDP checksum transport
transport header value
SNMP msg (UDP)
(UDP) ▪ extracts application-layer
network message network (IP)
h SNMP(IP)
msg
UDP
▪ demultiplexes message up link
link to application via socket
physical physical

Transport Layer: 3-31

UDP segment header
32 bits
source port # dest port #
length checksum

application length, in bytes of

data UDP segment,
(payload) including header

data to/from
UDP segment format application layer

Transport Layer: 3-32

UDP checksum
Goal: detect errors (i.e., flipped bits) in transmitted segment
1st number 2nd number sum

Transmitted: 5 6 11

Received: 4 6 11

receiver-computed sender-computed
checksum
= checksum (as received)

Transport Layer: 3-33

Internet checksum
Goal: detect errors (i.e., flipped bits) in transmitted segment
sender: receiver:
▪ treat contents of UDP ▪ compute checksum of received
segment (including UDP header segment
fields and IP addresses) as
sequence of 16-bit integers ▪ check if computed checksum equals
▪ checksum: addition (one’s checksum field value:
complement sum) of segment • not equal - error detected
content • equal - no error detected. But maybe
▪ checksum value put into errors nonetheless? More later ….
UDP checksum field
Transport Layer: 3-34
Internet checksum: an example
example: add two 16-bit integers
1 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0
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 0 1 1 1 0 1 1 1 0 1 1 1 1 0 0
checksum 0 1 0 0 0 1 0 0 0 1 0 0 0 0 1 1

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: http://gaia.cs.umass.edu/kurose_ross/interactive/
Transport Layer: 3-35
Internet checksum: weak protection!
example: add two 16-bit integers
0 1
1 1 1 0 0 1 1 0 0 1 1 0 0 1 1 0 1 0
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 Even though
numbers have
sum 1 0 1 1 1 0 1 1 1 0 1 1 1 1 0 0 changed (bit
flips), no change
checksum 0 1 0 0 0 1 0 0 0 1 0 0 0 0 1 1 in checksum!

Transport Layer: 3-36

Summary: UDP
▪ “no frills” protocol:
• segments may be lost, delivered out of order
• best effort service: “send and hope for the best”
▪ UDP has its plusses:
• no setup/handshaking needed (no RTT incurred)
• can function when network service is compromised
• helps with reliability (checksum)
▪ build additional functionality on top of UDP in application layer
(e.g., HTTP/3)
Chapter 3: roadmap
▪ Transport-layer services
▪ Multiplexing and demultiplexing
▪ Connectionless transport: UDP
▪ Principles of reliable data transfer
▪ Connection-oriented transport: TCP
▪ Principles of congestion control
▪ TCP congestion control
▪ Evolution of transport-layer
functionality
Transport Layer: 3-38
 Principles of reliable data transfer

sending receiving
process process
applicatio data data
ntranspor
t reliable
channel
reliable service abstraction

Transport Layer: 3-39

 Principles of reliable data transfer

sending receiving sending receiving

process process process process
applicatio data data application data data
ntranspor transport
t reliable
channel
sender-side of receiver-side
reliable service abstraction reliable data of reliable data
transfer protocol transfer protocol

transport
network
unreliable channel

reliable service implementation

Transport Layer: 3-40

Principles of reliable data transfer

sending receiving
process process
application data data
transport

sender-side of receiver-side
Complexity of reliable data reliable data
transfer protocol
of reliable data
transfer protocol
transfer protocol will depend
(strongly) on characteristics of transport
network
unreliable channel (lose, unreliable channel
corrupt, reorder data?)
reliable service implementation

Transport Layer: 3-41

Principles of reliable data transfer

sending receiving
process process
application data data
transport

sender-side of receiver-side
reliable data of reliable data
Sender, receiver do not know transfer protocol transfer protocol
the “state” of each other, e.g.,
was a message received? transport

▪ unless communicated via a

network
unreliable channel

message
reliable service implementation

Transport Layer: 3-42

 Reliable data transfer protocol (rdt): interfaces
rdt_send(): called from above, deliver_data(): called by rdt
(e.g., by app.). Passed data to to deliver data to upper layer
deliver to receiver upper layer
sending receiving
process process
rdt_send() data data
deliver_data()

sender-side data receiver-side

implementation of implementation of
rdt reliable data packet rdt reliable data
transfer protocol transfer protocol

udt_send() Header data Header data rdt_rcv()

unreliable channel
udt_send(): called by rdt rdt_rcv(): called when packet
to transfer packet over arrives on receiver side of
Bi-directional communication over
unreliable channel to receiver unreliable channel channel
Transport Layer: 3-43
Reliable data transfer: getting started
We will:
▪ incrementally develop sender, receiver sides of reliable data transfer
protocol (rdt)
▪ consider only unidirectional data transfer
• but control info will flow in 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 uniquely state state
determined by next 1 event
event
2
actions

Transport Layer: 3-44

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)

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

Transport Layer: 3-45

rdt2.0: channel with bit errors
▪ underlying channel may flip bits in packet
• checksum (e.g., Internet checksum) to detect bit errors
▪ the question: how to recover from errors?

How do humans recover from “errors” during conversation?

Transport Layer: 3-46

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

stop and wait

sender sends one packet, then waits for receiver response
Transport Layer: 3-47
rdt2.0: FSM specifications
rdt_send(data)
snkpkt = make_pkt(data, checksum)
udt_send(sndpkt)
rdt_rcv(rcvpkt) &&
Wait for Wait for isNAK(rcvpkt)
sender call from ACK or udt_send(sndpkt) rdt_rcv(rcvpkt) && corrupt(rcvpkt)
above NAK
udt_send(NAK)

rdt_rcv(rcvpkt) && isACK(rcvpkt)

Wait for
Λ call from receiver
below

rdt_rcv(rcvpkt) && notcorrupt(rcvpkt)

extract(rcvpkt,data)
deliver_data(data)
udt_send(ACK)

Transport Layer: 3-48

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

rdt_rcv(rcvpkt) && isACK(rcvpkt)

Wait for
Λ call from receiver
below

Note: “state” of receiver (did the receiver get my

rdt_rcv(rcvpkt) && notcorrupt(rcvpkt)
message correctly?) isn’t known to sender unless
extract(rcvpkt,data)
somehow communicated from receiver to sender deliver_data(data)
▪ that’s why we need a protocol! udt_send(ACK)

Transport Layer: 3-49

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

rdt_rcv(rcvpkt) && isACK(rcvpkt)

Wait for
Λ call from receiver
below

rdt_rcv(rcvpkt) && notcorrupt(rcvpkt)

extract(rcvpkt,data)
deliver_data(data)
udt_send(ACK)

Transport Layer: 3-50

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

rdt_rcv(rcvpkt) && isACK(rcvpkt)

Wait for
Λ call from receiver
below

rdt_rcv(rcvpkt) && notcorrupt(rcvpkt)

extract(rcvpkt,data)
deliver_data(data)
udt_send(ACK)

Transport Layer: 3-51

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

stop and wait

sender sends one packet, then
waits for receiver response
Transport Layer: 3-52
rdt2.1: sender, handling garbled ACK/NAKs
rdt_send(data)
sndpkt = make_pkt(0, data, checksum)
udt_send(sndpkt) rdt_rcv(rcvpkt) &&
(corrupt(rcvpkt) ||
Wait for Wait for isNAK(rcvpkt) )
call 0 from ACK or
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: 3-53

rdt2.1: receiver, handling 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 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: 3-54

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

Transport Layer: 3-55

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

As we will see, TCP uses this approach to be NAK-free

Transport Layer: 3-56

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) || Λ
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: 3-57
rdt3.0: channels with errors and loss
New channel assumption: underlying channel can also lose
packets (data, ACKs)
• checksum, sequence #s, ACKs, retransmissions will be of help …
but not quite enough

Q: How do humans handle lost

sender-to-receiver words in
conversation?
Transport Layer: 3-58
rdt3.0: channels with errors and loss
Approach: sender waits “reasonable” amount of time for ACK
▪ retransmits if no ACK received in this time
▪ if pkt (or ACK) just delayed (not lost):
• retransmission will be duplicate, but seq #s already handles this!
• receiver must specify seq # of packet being ACKed
▪ use countdown timer to interrupt after “reasonable” amount
of time
timeout

Transport Layer: 3-59

rdt3.0 sender
rdt_send(data)
sndpkt = make_pkt(0, data, checksum)
udt_send(sndpkt)
start_timer

Wait for Wait

call 0 from for
above ACK0
rdt_rcv(rcvpkt)
&& notcorrupt(rcvpkt) rdt_rcv(rcvpkt)
&& isACK(rcvpkt,1) && notcorrupt(rcvpkt)
stop_timer && isACK(rcvpkt,0)
stop_timer
Wait Wait for
for call 1 from
ACK1 above

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

Transport Layer: 3-60

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 Λ
Λ Wait for Wait
for timeout
call 0 from
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) Λ
rdt_rcv(rcvpkt) &&
( corrupt(rcvpkt) || sndpkt = make_pkt(1, data, checksum)
isACK(rcvpkt,0) ) udt_send(sndpkt)
start_timer
Λ

Transport Layer: 3-61

rdt3.0 in action
sender receiver sender receiver
send pkt0 pkt0 send pkt0 pkt0
rcv pkt0 rcv pkt0
ack0 send ack0 ack0 send ack0
rcv ack0 rcv ack0
send pkt1 pkt1 send pkt1 pkt1
rcv pkt1 X
loss
ack1 send ack1
rcv ack1
send pkt0 pkt0
rcv pkt0 timeout
ack0 send ack0 resend pkt1 pkt1
rcv pkt1
ack1 send ack1
rcv ack1
send pkt0 pkt0
(a) no loss rcv pkt0
ack0 send ack0

(b) packet loss

Transport Layer: 3-62
rdt3.0 in action
sender receiver
sender receiver send pkt0
pkt0
rcv pkt0
send pkt0 pkt0 send ack0
ack0
rcv pkt0 rcv ack0
ack0 send ack0 send pkt1 pkt1
rcv ack0 rcv pkt1
send pkt1 pkt1 send ack1
rcv pkt1 ack1
ack1 send ack1
X timeout
loss resend pkt1
pkt1 rcv pkt1
timeout
resend pkt1 pkt1
rcv pkt1 rcv ack1 (detect duplicate)
send pkt0 pkt0 send ack1
(detect duplicate)
ack1 send ack1 ack1 rcv pkt0
rcv ack1 rcv ack1 send ack0
send pkt0 pkt0 (ignore) ack0
rcv pkt0
ack0 send ack0 pkt1

(c) ACK loss (d) premature timeout/ delayed ACK

Transport Layer: 3-63
Performance of rdt3.0 (stop-and-wait)
▪ U sender: utilization – fraction of time sender busy sending

▪ example: 1 Gbps link, 15 ms prop. delay, 8000 bit packet

• time to transmit packet into channel:
L 8000 bits
Dtrans = R = 9 = 8 microsecs
10 bits/sec

Transport Layer: 3-64

rdt3.0: stop-and-wait operation
sender receiver
first packet bit transmitted, t = 0

first packet bit arrives

RTT last packet bit arrives, send ACK

ACK arrives, send next

packet, t = RTT + L / R

Transport Layer: 3-65

rdt3.0: stop-and-wait operation
sender receiver

L/R L/R
Usender=
RTT + L / R
.008 RTT
=
30.008
= 0.00027

▪ rdt 3.0 protocol performance stinks!

▪ Protocol limits performance of underlying infrastructure (channel)

Transport Layer: 3-66

rdt3.0: pipelined protocols operation
pipelining: sender allows multiple, “in-flight”, yet-to-be-acknowledged
packets
• range of sequence numbers must be increased
• buffering at sender and/or receiver

Transport Layer: 3-67

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: 3-68

Go-Back-N: sender
▪ sender: “window” of up to N, consecutive transmitted but unACKed pkts
• k-bit seq # in pkt header

▪ cumulative ACK: ACK(n): ACKs all packets up to, including seq # n

• on receiving ACK(n): move window forward to begin at n+1
▪ timer for oldest in-flight packet
▪ timeout(n): retransmit packet n and all higher seq # packets in window
Transport Layer: 3-69
Go-Back-N: receiver
▪ ACK-only: always send ACK for correctly-received packet so far, with
highest in-order seq #
• may generate duplicate ACKs
• need only remember rcv_base
▪ on receipt of out-of-order packet:
• can discard (don’t buffer) or buffer: an implementation decision
• re-ACK pkt with highest in-order seq #

Receiver view of sequence number space:

received and ACKed

… … Out-of-order: received but not ACKed

rcv_base
Not received
Transport Layer: 3-70
Go-Back-N in action
sender window (N=4) sender receiver
012345678 send pkt0
012345678 send pkt1
send pkt2 receive pkt0, send ack0
012345678
send pkt3 Xloss receive pkt1, send ack1
012345678
(wait)
receive pkt3, discard,
012345678 rcv ack0, send pkt4 (re)send ack1
012345678 rcv ack1, send pkt5 receive pkt4, discard,
(re)send ack1
ignore duplicate ACK receive pkt5, discard,
(re)send ack1
pkt 2 timeout
012345678 send pkt2
012345678 send pkt3
012345678 send pkt4 rcv pkt2, deliver, send ack2
012345678 send pkt5 rcv pkt3, deliver, send ack3
rcv pkt4, deliver, send ack4
rcv pkt5, deliver, send ack5

Transport Layer: 3-71

Selective repeat
▪receiver individually acknowledges all correctly received packets
• buffers packets, as needed, for eventual in-order delivery to upper
layer
▪sender times-out/retransmits individually for unACKed packets
• sender maintains timer for each unACKed pkt
▪sender window
• N consecutive seq #s
• limits seq #s of sent, unACKed packets

Transport Layer: 3-72

Selective repeat: sender, receiver windows

Transport Layer: 3-73

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

Transport Layer: 3-74

Selective Repeat in action
sender window (N=4) sender receiver
012345678 send pkt0
012345678 send pkt1
012345678 send pkt2 receive pkt0, send ack0
012345678 send pkt3 Xloss receive pkt1, send ack1
(wait)
receive pkt3, buffer,
012345678 rcv ack0, send pkt4 send ack3
012345678 rcv ack1, send pkt5
receive pkt4, buffer,
record ack3 arrived send ack4
receive pkt5, buffer,
pkt 2 timeout send ack5
012345678 send pkt2
012345678 (but not 3,4,5)
012345678 rcv pkt2; deliver pkt2,
012345678 pkt3, pkt4, pkt5; send ack2

Q: what happens when ack2 arrives?

Transport Layer: 3-75

 sender window receiver window

Selective repeat: (after receipt)

0123012 pkt0
(after receipt)

a dilemma! 0123012
0123012
pkt1
pkt2
0123012
0123012
0123012
example: 0123012 pkt3
X
▪ seq #s: 0, 1, 2, 3 (base 4 counting)
0123012
pkt0 will accept packet

▪ window size=3
with seq number 0
(a) no problem

0123012 pkt0
0123012 pkt1 0123012
0123012 pkt2 X 0123012
X 0123012
X
timeout
retransmit pkt0
0123012 pkt0
will accept packet
with seq number 0
(b) oops!
Transport Layer: 3-76
 sender window receiver window

Selective repeat: (after receipt)

0123012 pkt0
(after receipt)

a dilemma! 0123012
0123012
pkt1
pkt2
0123012
0123012
0123012
example: 0123012 pkt3
X
▪ seq #s: 0, 1, 2, 3 (base 4 counting) ▪ receiver can’t
0123012
pkt0 will accept packet
see sender side
▪ window size=3
with seq number 0
▪ receiver
(a) no problem
behavior
identical in both
cases!
▪0something’s
123012 pkt0
Q: what relationship is needed 0(very)
1 2 3 0 1wrong!
2 pkt1 0123012

between sequence # size and 0123012 pkt2 X

X
0123012
0123012
window size to avoid problem X
timeout
in scenario (b)? retransmit pkt0
0123012 pkt0
will accept packet
with seq number 0
(b) oops!
Transport Layer: 3-77
Chapter 3: roadmap
▪ Transport-layer services
▪ Multiplexing and demultiplexing
▪ Connectionless transport: UDP
▪ Principles of reliable data transfer
▪ Connection-oriented transport: TCP
• segment structure
• reliable data transfer
• flow control
• connection management
▪ Principles of congestion control
▪ TCP congestion control
Transport Layer: 3-78
TCP: overview RFCs: 793,1122, 2018, 5681, 7323
▪ point-to-point: ▪ cumulative ACKs
• one sender, one receiver ▪ pipelining:
▪ reliable, in-order byte • TCP congestion and flow control
steam: set window size
• no “message boundaries" ▪ connection-oriented:
▪ full duplex data: • handshaking (exchange of control
• bi-directional data flow in messages) initializes sender,
same connection receiver state before data
• MSS: maximum segment size exchange
▪ flow controlled:
• sender will not overwhelm receiver
Transport Layer: 3-79
 TCP segment structure
32 bits

source port # dest port # segment seq #: counting

ACK: seq # of next expected sequence number bytes of data into bytestream
byte; A bit: this is an ACK (not segments!)
acknowledgement number
head not
length (of TCP header) len used C EUAP R SF receive window flow control: # bytes
Internet checksum checksum Urg data pointer receiver willing to accept
options (variable length)
C, E: congestion notification
TCP options
application data sent by
RST, SYN, FIN: connection data application into
management (variable length) TCP socket

Transport Layer: 3-80

TCP sequence numbers, ACKs
outgoing segment from sender
Sequence numbers: source port # dest port #
sequence number
• byte stream “number” of first acknowledgement number
rwnd
byte in segment’s data checksum urg pointer

window size
Acknowledgements: N

• seq # of next byte expected

from other side sender sequence number space

• cumulative ACK sent sent, usable not

ACKed not-yet but not usable
ACKed yet sent
Q: how receiver handles (“in-flight”)

out-of-order segments outgoing segment from receiver

source port # dest port #
• A: TCP spec doesn’t say, - up sequence number

to implementor acknowledgement number

A rwnd
checksum urg pointer
Transport Layer: 3-81
TCP sequence numbers, ACKs
Host A Host B

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

simple telnet scenario

Transport Layer: 3-82
TCP round trip time, timeout
Q: how to set TCP timeout Q: how to estimate RTT?
value? ▪ SampleRTT:measured time
▪ longer than RTT, but RTT varies! from segment transmission until
ACK receipt
▪ too short: premature timeout,
• ignore retransmissions
unnecessary retransmissions
▪ SampleRTT will vary, want
▪ too long: slow reaction to estimated RTT “smoother”
segment loss • average several recent
measurements, not just current
SampleRTT

Transport Layer: 3-83

TCP round trip time, timeout
EstimatedRTT = (1- α)*EstimatedRTT + α*SampleRTT
▪ exponential weighted moving average (EWMA)
▪ influence of past sample decreases exponentially fast
▪ typical value: α = 0.125
RTT: gaia.cs.umass.edu to fantasia.eurecom.fr

(milliseconds)
RTT

sampleRTT
EstimatedRTT

time (seconds)
Transport Layer: 3-84
TCP round trip time, timeout
▪ timeout interval: EstimatedRTT plus “safety margin”
• large variation in EstimatedRTT: want a larger safety margin
TimeoutInterval = EstimatedRTT + 4*DevRTT

estimated RTT “safety margin”

▪ DevRTT: EWMA of SampleRTT deviation from EstimatedRTT:

DevRTT = (1-β)*DevRTT + β*|SampleRTT-EstimatedRTT|
(typically, β = 0.25)

* Check out the online interactive exercises for more examples: http://gaia.cs.umass.edu/kurose_ross/interactive/
Transport Layer: 3-85
TCP Sender (simplified)
event: data received from event: timeout
application ▪ retransmit segment that
caused timeout
▪ create segment with seq #
▪ restart timer
▪ seq # is byte-stream number
of first data byte in segment
event: ACK received
▪ start timer if not already
running ▪ if ACK acknowledges
previously unACKed segments
• think of timer as for oldest
unACKed segment • update what is known to be
ACKed
• expiration interval:
TimeOutInterval • start timer if there are still
unACKed segments
Transport Layer: 3-86
TCP Receiver: ACK generation [RFC 5681]
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: 3-87

TCP: retransmission scenarios
Host A Host B Host A Host B

SendBase=92
Seq=92, 8 bytes of data Seq=92, 8 bytes of data

Seq=100, 20 bytes of data

timeo

timeo
ACK=100
ut

ut
X
ACK=100
ACK=120

Seq=92, 8 bytes of data Seq=92, 8

SendBase=100 bytes of data send cumulative
SendBase=120 ACK for 120
ACK=100
ACK=120

SendBase=120

lost ACK scenario premature timeout

Transport Layer: 3-88

TCP: retransmission scenarios
Host A Host B

Seq=92, 8 bytes of data

Seq=100, 20 bytes of data

ACK=100
X
ACK=120

Seq=120, 15 bytes of data

cumulative ACK covers

for earlier lost ACK

Transport Layer: 3-89

TCP fast retransmit
Host A Host B
TCP fast retransmit
if sender receives 3 additional
Seq=92
ACKs for same data (“triple Seq=1
, 8 byte
s of da
ta
00, 20
duplicate ACKs”), resend unACKed bytes
of data
segment with smallest seq # X
▪ likely that unACKed segment lost,
=100
so don’t wait for timeout ACK

timeout
C K =100
A
=100
ACK
C K =100
Receipt of three duplicate ACKs A

indicates 3 segments received Seq=100, 20 bytes of data

after a missing segment – lost

segment is likely. So retransmit!

Transport Layer: 3-90

Chapter 3: roadmap
▪ Transport-layer services
▪ Multiplexing and demultiplexing
▪ Connectionless transport: UDP
▪ Principles of reliable data transfer
▪ Connection-oriented transport: TCP
• segment structure
• reliable data transfer
• flow control
• connection management
▪ Principles of congestion control
▪ TCP congestion control
Transport Layer: 3-91
TCP flow control
applicati
on
Q: What happens if network Application removing process
layer delivers data faster than data from TCP socket
buffers
application layer removes TCP socket
data from socket buffers? receiver buffers

TCP
code
Network layer
delivering IP datagram
payload into TCP
IP
socket buffers code

from sender

receiver protocol stack

Transport Layer: 3-92

TCP flow control
applicati
on
Q: What happens if network Application removing process
layer delivers data faster than data from TCP socket
buffers
application layer removes TCP socket
data from socket buffers? receiver buffers

TCP
code
Network layer
delivering IP datagram
payload into TCP
IP
socket buffers code

from sender

receiver protocol stack

Transport Layer: 3-93

TCP flow control
applicati
on
Q: What happens if network Application removing process
layer delivers data faster than data from TCP socket
buffers
application layer removes TCP socket
data from socket buffers? receiver buffers

TCP
code

receive window
flow control: # bytes
receiver willing to accept IP
code

from sender

receiver protocol stack

Transport Layer: 3-94

TCP flow control
applicati
on
Q: What happens if network Application removing process
layer delivers data faster than data from TCP socket
buffers
application layer removes TCP socket
data from socket buffers? receiver buffers

TCP
flow control code

receiver controls sender, so

sender won’t overflow IP
code
receiver’s buffer by
transmitting too much, too fast
from sender

receiver protocol stack

Transport Layer: 3-95

TCP flow control
▪ TCP receiver “advertises” free buffer
space in rwnd field in TCP header to application process
• RcvBuffer size set via socket
options (typical default is 4096 bytes) RcvBuffer buffered data
• many operating systems autoadjust
RcvBuffer
rwnd free buffer space

▪ sender limits amount of unACKed

(“in-flight”) data to received rwnd TCP segment payloads

▪ guarantees receive buffer will not TCP receiver-side buffering

overflow

Transport Layer: 3-96

TCP flow control
flow control: # bytes receiver willing to accept

▪ TCP receiver “advertises” free buffer

space in rwnd field in TCP header
• RcvBuffer size set via socket
receive window
options (typical default is 4096 bytes)
• many operating systems autoadjust
RcvBuffer
▪ sender limits amount of unACKed
(“in-flight”) data to received rwnd
▪ guarantees receive buffer will not
overflow
TCP segment format

Transport Layer: 3-97

TCP connection management
before exchanging data, sender/receiver “handshake”:
▪ agree to establish connection (each knowing the other willing to establish connection)
▪ agree on connection parameters (e.g., starting seq #s)

application application

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 number"); welcomeSocket.accept();
Transport Layer: 3-98
Agreeing to establish a connection
2-way handshake:

Q: will 2-way handshake always

Let’s talk work in network?
ESTAB
ESTAB
OK ▪ variable delays
▪ retransmitted messages (e.g.
req_conn(x)) due to message loss
▪ message reordering
choose x
req_conn(x) ▪ can’t “see” other side
ESTAB
acc_conn(x)
ESTAB

Transport Layer: 3-99

 2-way handshake scenarios
choose x
req_conn(x)
ESTAB
acc_conn(x)

ESTAB
data(x+1) accept
ACK(x+1) data(x+1)

connection
x completes

No problem!

Transport Layer: 3-100

2-way handshake scenarios

choose x
req_conn(x)
ESTAB
retransmit acc_conn(x)
req_conn(x)

ESTAB
req_conn(x)

connection
client x completes server
terminates forgets x

ESTAB
acc_conn(x)
Problem: half open
connection! (no client)
Transport Layer: 3-101
2-way handshake scenarios
choose x
req_conn(x)
ESTAB
retransmit acc_conn(x)
req_conn(x)

ESTAB
data(x+1) accept
data(x+1)
retransmit
data(x+1)
connection
x completes server
client
terminates forgets x
req_conn(x)
ESTAB
data(x+1) accept
data(x+1)
Problem: dup data
accepted!
 TCP 3-way handshake
Server state
serverSocket = socket(AF_INET,SOCK_STREAM)
Client state serverSocket.bind((‘’,serverPort))
serverSocket.listen(1)
clientSocket = socket(AF_INET, SOCK_STREAM) connectionSocket, addr = serverSocket.accept()
LISTEN
clientSocket.connect((serverName,serverPort)) LISTEN
choose init seq num, x
send TCP SYN msg
SYNSENT SYNbit=1, Seq=x
choose init seq num, y
send TCP SYNACK
msg, acking SYN SYN RCVD
SYNbit=1, Seq=y
ACKbit=1; ACKnum=x+1
received SYNACK(x)
ESTAB indicates server is live;
send ACK for SYNACK;
this segment may contain ACKbit=1, ACKnum=y+1
client-to-server data
received ACK(y)
indicates client is live
ESTAB

Transport Layer: 3-103

A human 3-way handshake protocol

1. On belay?

2. Belay on.
3. Climbing.

Transport Layer: 3-104

Closing a TCP 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: 3-105

Chapter 3: roadmap
▪ Transport-layer services
▪ Multiplexing and demultiplexing
▪ Connectionless transport: UDP
▪ Principles of reliable data transfer
▪ Connection-oriented transport: TCP
▪ Principles of congestion control
▪ TCP congestion control
▪ Evolution of transport-layer
functionality
Transport Layer: 3-106
Principles of congestion control
Congestion:
▪ informally: “too many sources sending too much data too fast for
network to handle”
▪ manifestations:
• long delays (queueing in router buffers)
• packet loss (buffer overflow at routers)
▪ different from flow control! congestion control:
▪ a top-10 problem! too many senders,
sending too fast

flow control: one sender

too fast for one receiver
Transport Layer: 3-107
Causes/costs of congestion: scenario 1
original data: λ throughput: λout
in
Simplest scenario: Host A
▪ one router, infinite buffers
▪ input, output link capacity: R infinite shared
output link buffers

▪ two flows
R R
▪ no retransmissions needed
Host B

R/2
Q: What happens as
λout

delay
arrival rate λin throughput:

approaches R/2?
λin R/2 λin R/2
maximum per-connection large delays as arrival rate
throughput: R/2 λin approaches capacity
Transport Layer: 3-108
Causes/costs of congestion: scenario 2
▪ one router, finite buffers
▪ sender retransmits lost, timed-out packet
• application-layer input = application-layer output: λin = λout
• transport-layer input includes retransmissions : λ’in λin

Host A λin : original data

λout
λ'in: original data, plus
retransmitted data

R R

Host B finite shared output

link buffers
Transport Layer: 3-109
Causes/costs of congestion: scenario 2
Idealization: perfect knowledge R/2

▪ sender sends only when router buffers available

throughput: λout
Host A λin : original data λin
copy λout R/2
λ'in: original data, plus
retransmitted data

free buffer space!

R R

Host B finite shared output

link buffers
Transport Layer: 3-110
Causes/costs of congestion: scenario 2
Idealization: some perfect knowledge
▪ packets can be lost (dropped at router) due
to full buffers
▪ sender knows when packet has been
dropped: only resends if packet known to be
lost
Host A λin : original data
copy λ'in: original data, plus
retransmitted data

no buffer space!

R R

Host B finite shared output

link buffers
Transport Layer: 3-111
Causes/costs of congestion: scenario 2
Idealization: some perfect knowledge R/2
“wasted” capacity due
▪ packets can be lost (dropped at router) due to retransmissions

throughput: λout
to full buffers
when sending at
▪ sender knows when packet has been R/2, some packets
are needed
dropped: only resends if packet known to be retransmissions
lost
Host A λin : original data λin R/2
λ'in: original data, plus
retransmitted data

free buffer space!

R R

Host B finite shared output

link buffers
Transport Layer: 3-112
Causes/costs of congestion: scenario 2
Realistic scenario: un-needed duplicates R/2
▪ packets can be lost, dropped at router due to “wasted” capacity due

throughput: λout
full buffers – requiring retransmissions to un-needed
retransmissions
▪ but sender times can time out prematurely,
sending two copies, both of which are delivered when sending at
R/2, some packets
are retransmissions,
including needed
λin : original data and un-needed
Host A λin duplicates, that are
copy
timeout R/2
λ'in: original data, plus delivered!
retransmitted data

free buffer space!

R R

Host B finite shared output

link buffers
Transport Layer: 3-113
Causes/costs of congestion: scenario 2
Realistic scenario: un-needed duplicates R/2
▪ packets can be lost, dropped at router due to “wasted” capacity due

throughput: λout
full buffers – requiring retransmissions to un-needed
retransmissions
▪ but sender times can time out prematurely,
sending two copies, both of which are delivered when sending at
R/2, some packets
are retransmissions,
including needed
and un-needed
λin R/2 duplicates, that are
delivered!
“costs” of congestion:
▪ more work (retransmission) for given receiver throughput
▪ unneeded retransmissions: link carries multiple copies of a packet
• decreasing maximum achievable throughput

Transport Layer: 3-114

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

Host D
λout
Host C

Transport Layer: 3-115

Causes/costs of congestion: scenario 3
R/2
λout

λin’ R/2

another “cost” of congestion:

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

Transport Layer: 3-116

Causes/costs of congestion: insights
▪ throughput can never exceed capacity

▪ delay increases as capacity approached

▪ loss/retransmission decreases effective

throughput

▪ un-needed duplicates further decreases

effective throughput
▪ upstream transmission capacity / buffering
wasted for packets lost downstream
Transport Layer: 3-117
Approaches towards congestion control
End-end congestion control:
▪ no explicit feedback from
network
▪ congestion inferred from ACKs
data data
observed loss, delay ACKs

▪ approach taken by TCP

Transport Layer: 3-118

Approaches towards congestion control
Network-assisted congestion
control: explicit congestion info

▪ routers provide direct feedback

to sending/receiving hosts with data data
ACKs
flows passing through congested ACKs

router
▪ may indicate congestion level or
explicitly set sending rate
▪ TCP ECN, ATM, DECbit protocols
Transport Layer: 3-119
Chapter 3: roadmap
▪ Transport-layer services
▪ Multiplexing and demultiplexing
▪ Connectionless transport: UDP
▪ Principles of reliable data transfer
▪ Connection-oriented transport: TCP
▪ Principles of congestion control
▪ TCP congestion control
▪ Evolution of transport-layer
functionality
Transport Layer: 3-120
TCP congestion control: AIMD
▪ approach: senders can increase sending rate until packet loss
(congestion) occurs, then decrease sending rate on loss event
Additive Increase Multiplicative Decrease
increase sending rate by 1 cut sending rate in half at each
maximum segment size every loss event
RTT until loss detected
TCP sender Sending rate

AIMD sawtooth
behavior: probing
for bandwidth

time Transport Layer: 3-121

TCP AIMD: more
Multiplicative decrease detail: sending rate is
▪ Cut in half on loss detected by triple duplicate ACK (TCP Reno)
▪ Cut to 1 MSS (maximum segment size) when loss detected by
timeout (TCP Tahoe)

Why AIMD?
▪ AIMD – a distributed, asynchronous algorithm – has been
shown to:
• optimize congested flow rates network wide!
• have desirable stability properties

Transport Layer: 3-122

TCP congestion control: details
sender sequence number space
cwnd TCP sending behavior:
▪ roughly: send cwnd bytes,
wait RTT for ACKS, then
send more bytes
last byte
available but cwnd
ACKed sent, but TCP rate ~
~ bytes/sec
not-yet ACKed not used RTT
(“in-flight”) last byte sent

▪ TCP sender limits transmission: LastByteSent- LastByteAcked < cwnd

▪ cwnd is dynamically adjusted in response to observed

network congestion (implementing TCP congestion control)
Transport Layer: 3-123
TCP slow start
Host A Host B
▪ when connection begins,
increase rate exponentially
one segm
until first loss event: ent

RTT
• initially cwnd = 1 MSS two segm
ents
• double cwnd every RTT
• done by incrementing cwnd
four segm
for every ACK received ents

▪ summary: initial rate is

slow, but ramps up
time
exponentially fast
Transport Layer: 3-124
TCP: from slow start to congestion
avoidance
Q: when should the exponential
increase switch to linear?
X
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: http://gaia.cs.umass.edu/kurose_ross/interactive/
Transport Layer: 3-125
Summary: TCP congestion control
New
New ACK!
ACK! new ACK
duplicate ACK
dupACKcount++ new ACK .
cwnd = cwnd + MSS (MSS/cwnd)
dupACKcount = 0
cwnd = cwnd+MSS transmit new segment(s), as allowed
dupACKcount = 0
Λ transmit new segment(s), as allowed
cwnd = 1 MSS
ssthresh = 64 KB cwnd > ssthresh
dupACKcount = 0 slow Λ congestion
start timeout avoidance
ssthresh = cwnd/2
cwnd = 1 MSS duplicate ACK
timeout dupACKcount = 0 dupACKcount++
ssthresh = cwnd/2 retransmit missing segment
cwnd = 1 MSS
dupACKcount = 0
retransmit missing segment
timeout New
ACK!
ssthresh = cwnd/2
cwnd = 1 New ACK
dupACKcount = 0
dupACKcount == 3 cwnd = ssthresh dupACKcount == 3
retransmit missing segment 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: 3-126

TCP CUBIC
▪ Is there a better way than AIMD to “probe” for usable bandwidth?
▪ Insight/intuition:
• Wmax: sending rate at which congestion loss was detected
• congestion state of bottleneck link probably (?) hasn’t changed much
• after cutting rate/window in half on loss, initially ramp to to Wmax faster, but then
approach Wmax more slowly

Wmax classic TCP

TCP CUBIC - higher

Wmax/2 throughput in this
example

Transport Layer: 3-127

TCP CUBIC
▪ K: point in time when TCP window size will reach Wmax
• K itself is tuneable
▪ increase W as a function of the cube of the distance between current
time and K
• larger increases when further away from K
• smaller increases (cautious) when nearer K
▪ TCP CUBIC default Wmax
in Linux, most
TCP Reno
popular TCP for TCP CUBIC
popular Web TCP
sending
servers rate

time
t0 t1 t2 t3 t4
Transport Layer: 3-128
TCP and the congested “bottleneck link”
▪ TCP (classic, CUBIC) increase TCP’s sending rate until packet loss occurs
at some router’s output: the bottleneck link

sourc destination
e
application application
TCP TCP
network network
link link
physical physical
packet queue almost
never empty, sometimes
overflows packet (loss)

bottleneck link (almost always busy)

Transport Layer: 3-129
TCP and the congested “bottleneck link”
▪ TCP (classic, CUBIC) increase TCP’s sending rate until packet loss occurs
at some router’s output: the bottleneck link
▪ understanding congestion: useful to focus on congested bottleneck link

insight: increasing TCP sending rate will

sourc not increase end-end throughout destination
e
application
with congested bottleneck
application
TCP TCP
network network
link link
physical physical

insight: increasing TCP

sending rate will
increase measured RTT
Goal: “keep the end-end pipe just full, but not fuller”
RTT
Transport Layer: 3-130
Delay-based TCP congestion control
Keeping sender-to-receiver pipe “just full enough, but no fuller”: keep
bottleneck link busy transmitting, but avoid high delays/buffering
# bytes sent in
measured last RTT interval
RTTmeasured throughput =
RTTmeasured
Delay-based approach:
▪ RTTmin - minimum observed RTT (uncongested path)
▪ uncongested throughput with congestion window cwnd is cwnd/RTTmin
if measured throughput “very close” to uncongested throughput
increase cwnd linearly /* since path not congested */
else if measured throughput “far below” uncongested throughout
decrease cwnd linearly /* since path is congested */
Transport Layer: 3-131
Delay-based TCP congestion control
▪ congestion control without inducing/forcing loss
▪ maximizing throughout (“keeping the just pipe full… ”) while keeping
delay low (“…but not fuller”)
▪ a number of deployed TCPs take a delay-based approach
▪ BBR deployed on Google’s (internal) backbone network

Transport Layer: 3-132

Explicit congestion notification (ECN)
TCP deployments often implement network-assisted congestion control:
▪ two bits in IP header (ToS field) marked by network router to indicate congestion
• policy to determine marking chosen by network operator
▪ congestion indication carried to destination
▪ destination sets ECE bit on ACK segment to notify sender of congestion
▪ involves both IP (IP header ECN bit marking) and TCP (TCP header C,E bit marking)
sourc TCP ACK segment
destination
e
application
ECE=1 application
TCP TCP
network network
link link
physical physical

ECN=10 ECN=11

IP datagram
Transport Layer: 3-133
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 connection 2 router
capacity R

Transport Layer: 3-134

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

R equal bandwidth share

Is TCP fair?
Connection 2 throughput

A: Yes, under idealized

loss: decrease window by factor of 2 assumptions:
congestion avoidance: additive increase ▪ same RTT
▪ fixed number of sessions
loss: decrease window by factor of 2
congestion avoidance: additive increase
only in congestion
avoidance

Connection 1 throughput R
Transport Layer: 3-135
Fairness: must all network apps be “fair”?
Fairness and UDP Fairness, parallel TCP
▪ multimedia apps often do not connections
use TCP ▪ application can open multiple
• do not want rate throttled by
congestion control parallel connections between two
hosts
▪ instead use UDP:
• send audio/video at constant rate, ▪ web browsers do this , e.g., link of
tolerate packet loss rate R with 9 existing connections:
▪ there is no “Internet police” • new app asks for 1 TCP, gets rate R/10
policing use of congestion • new app asks for 11 TCPs, gets R/2
control

Transport Layer: 3-136

Transport layer: roadmap
▪ Transport-layer services
▪ Multiplexing and demultiplexing
▪ Connectionless transport: UDP
▪ Principles of reliable data transfer
▪ Connection-oriented transport: TCP
▪ Principles of congestion control
▪ TCP congestion control
▪ Evolution of transport-layer
functionality
Transport Layer: 3-137
Evolving transport-layer functionality
▪ TCP, UDP: principal transport protocols for 40 years
▪ different “flavors” of TCP developed, for specific scenarios:
Scenario Challenges
Long, fat pipes (large data Many packets “in flight”; loss shuts down
transfers) pipeline
Wireless networks Loss due to noisy wireless links, mobility;
TCP treat this as congestion loss
Long-delay links Extremely long RTTs
Data center networks Latency sensitive
Background traffic flows Low priority, “background” TCP flows

▪ moving transport–layer functions to application layer, on top of UDP

• HTTP/3: QUIC
Transport Layer: 3-138
QUIC: Quick UDP Internet Connections
▪ application-layer protocol, on top of UDP
• increase performance of HTTP
• deployed on many Google servers, apps (Chrome, mobile YouTube app)

HTTP/2 HTTP/2 (slimmed)

Application HTTP/3
TLS QUIC

Transport TCP UDP

Network IP IP

HTTP/2 over TCP HTTP/2 over QUIC over UDP

Transport Layer: 3-139

QUIC: Quick UDP Internet Connections
adopts approaches we’ve studied in this chapter for
connection establishment, error control, congestion control
• error and congestion control: “Readers familiar with TCP’s loss
detection and congestion control will find algorithms here that parallel
well-known TCP ones.” [from QUIC specification]
• connection establishment: reliability, congestion control,
authentication, encryption, state established in one RTT

▪ multiple application-level “streams” multiplexed over single QUIC

connection
• separate reliable data transfer, security
• common congestion control
Transport Layer: 3-140
QUIC: Connection establishment

TCP handshake
(transport layer) QUIC handshake

data
TLS handshake
(security)
data

TCP (reliability, congestion control QUIC: reliability, congestion control,

state) + TLS (authentication, crypto authentication, crypto state
state)
▪ 1 handshake
▪ 2 serial handshakes

Transport Layer: 3-141

 QUIC: streams: parallelism, no HOL blocking
HTTP HTTP
GET GET HTTP
GET
HTTP HTTP
application

GET GET
HTTP
GET QUIC QUIC QUIC QUIC QUIC QUIC
encrypt encrypt encrypt encrypt encrypt encrypt
QUIC QUIC QUIC
TLS encryption TLS encryption RDT RDT RDT error! QUIC
QUIC QUIC
RDT RDT RDT

QUIC Cong. Cont. QUIC Cong. Cont.

TCP RDT TCP
error! RDT
transport

TCP Cong. Contr. TCP Cong. Contr. UDP UDP

(a) HTTP 1.1 (b) HTTP/2 with QUIC: no HOL blocking

Transport Layer: 3-142
Chapter 3: summary
▪ principles behind transport Up next:
layer services: ▪ leaving the network
• multiplexing, demultiplexing “edge” (application,
• reliable data transfer transport layers)
• flow control ▪ into the network “core”
• congestion control
▪ two network-layer
▪ instantiation, implementation chapters:
in the Internet • data plane
• UDP • control plane
• TCP

Transport Layer: 3-143

Additional Chapter 3 slides

Transport Layer: 3-144

Go-Back-N: 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: 3-145
Go-Back-N: receiver extended FSM
any other event
udt_send(sndpkt) rdt_rcv(rcvpkt)
&& notcorrupt(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 packet with highest

in-order seq #
• may generate duplicate ACKs
• need only remember expectedseqnum
▪ out-of-order packet:
• discard (don’t buffer): no receiver buffering!
• re-ACK pkt with highest in-order seq #
Transport Layer: 3-146
TCP sender (simplified)
data received from application above
create segment, seq. #: NextSeqNum
pass segment to IP (i.e., “send”)
NextSeqNum = NextSeqNum + length(data)
if (timer currently not running)
Λ start timer
NextSeqNum = InitialSeqNum wait
SendBase = InitialSeqNum for
event timeout
retransmit not-yet-acked segment
with smallest seq. #
start timer
ACK received, with ACK field value 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: 3-147
TCP 3-way handshake FSM
closed
Socket connectionSocket =
welcomeSocket.accept();
Λ Socket clientSocket =
newSocket("hostname","port number");
SYN(x)
SYNACK(seq=y,ACKnum=x+1) SYN(seq=x)
create new socket for communication
back to client
listen

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

Transport Layer: 3-148

Closing a TCP connection
client state server state
ESTAB ESTAB
clientSocket.close(
) FINbit=1, seq=x
FIN_WAIT_1 can no longer
send but can
receive data CLOSE_WAIT
ACKbit=1; ACKnum=x+1
can still
FIN_WAIT_2 wait for server send data
close

LAST_ACK
FINbit=1, seq=y
TIMED_WAIT can no longer
send data
ACKbit=1; ACKnum=y+1
timed wait
for 2*max CLOSED
segment lifetime

CLOSED

Transport Layer: 3-149

TCP throughput
▪ avg. TCP thruput as function of window size, RTT?
• ignore slow start, assume there is 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/sec
4 RTT
W

W/2
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]:
1.22 . MSS
TCP throughput =
RTT L
➜ to achieve 10 Gbps throughput, need a loss rate of L = 2·10-10 – a
very small loss rate!
▪ versions of TCP for long, high-speed scenarios

Transport Layer: 3-151