2 of 16

Chapter 3 outline

3.1 Transport-layer services

3.2 Multiplexing and � demultiplexing

3.3 Connectionless � transport: UDP

3.4 Principles of reliable data� transfer

3.5 Connection-oriented � transport: TCP

segment structure
reliable data transfer
flow control
connection management

3.6 Principles of congestion� control

3.7 TCP congestion control

3 of 16

TCP reliable data transfer

TCP creates reliable service on top of IP’s unreliable service
Pipelined segments �(for performance)
Cumulative ACKs
Retransmissions

TCP uses single retransmission timer
even if multiple Tx but unACKed segments

Retransmissions are triggered by:

timeout events
duplicate ACK �(for performance reason)

Initially consider simplified TCP sender:

use only timeouts
ignore duplicate ACKs
also ignore flow control, congestion control

4 of 16

TCP sender events

data rcvd from app:

Create segment with �seq#

seq# is byte-stream number of first data byte in segment

start timer if not already running

think of timer as for oldest unACKed segment
expiration interval…

TimeOutInterval

timeout:

retransmit segment that caused timeout
restart timer

Ack rcvd:

if ACKs previously unACKed segments

update what is known to be ACKed
start timer if there are outstanding segments

5 of 16

TCP: retransmission scenarios

Host A

Seq=92, 8 bytes data

ACK=100

loss

timeout

reTx due to lost ACK

Host B

Seq=92, 8 bytes data

ACK=100

time

SendBase

= 100

Host A

Seq=100, 20 bytes data

ACK=100

time

Seq 100 not reTx

Host B

Seq=92, 8 bytes data

ACK=120

Seq=92, 8 bytes data

Seq=92 timeout

ACK=120

Seq=92 timeout

SendBase

= 120

SendBase

= 120

Sendbase

= 100

6 of 16

TCP: retransmission scenarios

Host A

Seq=92, 8 bytes data

ACK=100

loss

timeout

cum ACK avoids reTx

Host B

Seq=100, 20 bytes data

ACK=120

time

SendBase

= 120

7 of 16

Double Timeout Interval

8 of 16

TCP ACK generation [RFC 1122, RFC 2581]

Event at Receiver

Arrival of in-order segment with

expected seq #; All data up to

expected seq # already ACKed

Arrival of in-order segment with

expected seq #; One other in �order segment has ACK pending

Arrival of out-of-order segment

Higher than expected seq#;

Gap detected

Arrival of segment that

partially or completely fills gap

TCP Receiver action

Delayed ACK; Wait up to 500ms

for next segment; If no next segment,

send ACK

Immediately send single Cum ACK,

ACKing both in-order segments

Immediately send Duplicate ACK,

indicating seq# of expected byte

Imm send ACK, provided segment

starts at lower end of gap

Ack the “largest in-order byte” seq #

9 of 16

TCP fast retransmit

time-out period often relatively long:

long delay before resending lost packet

detect lost segments via duplicate ACKs

sender often sends many segments back-to-back
if a segment is lost, there will likely be many duplicate ACKs

if sender receives 3 ACKs for same data

(“triple duplicate ACKs”)

resend unACKed segment with smallest seq #

likely that unACKed segment got lost
so don’t wait for timeout

TCP fast retransmit

(Implicit NAK)

10 of 16

TCP fast retransmit

Fast retransmit �after sender receipt of �triple duplicate ACK

Host B

Host A

Seq=92, 8 bytes of data

ACK=100

timeout

ACK=100

Seq=100, 20 bytes of data

11 of 16

Principles in practice

TCP uses Seq Nums to allow receiver to identify lost or duplicate segments
TCP provides reliable data transfer by using positive ACKs and timeouts
TCP uses ACKs even when data not received correctly, and retransmits when segments or ACKs lost/corrupted
Certain TCP versions also have Implicit NAK with fast retransmit;

receipt of three duplicate ACKs for a segment serves as �Implicit NAK for the following segment
triggers reTx of that segment before timeout

12 of 16

Principles in practice

As in case of reliable data transfer protocol , TCP sender cannot itself tell for certain if a segment or its ACK, is lost, corrupted, or overly delayed

At sender, TCP response will be the same; �reTx the segment in question

TCP also uses pipelining, allowing sender to have multiple transmitted but yet-to-be ACKed segments outstanding at any given time
Pipelining can greatly improve session thruput, when ratio of segment size to round trip delay is small

13 of 16

Chapter 3 outline

3.1 Transport-layer services

3.2 Multiplexing and � demultiplexing

3.3 Connectionless � transport: UDP

3.4 Principles of reliable data� transfer

3.5 Connection-oriented � transport: TCP

segment structure
reliable data transfer
flow control
connection management

3.6 Principles of congestion� control

3.7 TCP congestion control

14 of 16

TCP Flow control

receive side of TCP connection has a receive buffer

speed-matching required: match the send rate to the receiving app’s drain rate

app process may be slow at reading from buffer

sender won’t overflow

receiver’s buffer �by transmitting too much,

too fast

flow control

15 of 16

TCP Receive Window size

(assume TCP receiver discards out-of-order segments)

Spare room in buffer

= RcvWindow

= RcvBuffer - [LastByteRcvd - LastByteRead]

16 of 16

TCP Flow control

Rcv Buffer size typical default is 4096 bytes
many operating systems auto adjust Rcv Buffer

Sender limits amount of unACKed data to receiver rwnd value

buffered data

free buffer space

rwnd

RcvBuffer

TCP segment payloads

to application process

Receiver-side buffering

Receiver “advertises” free buffer space by including rwnd value in TCP header of receiver-to-sender segments

LastByteSent-LastByteAcked ≤ rwnd

guarantees Rcv Buffer will �not overflow