Chapter 6

Errors, Error Detection,

and Error Control
 Introductio
n
 All transmitted signals will contain some rate
of error (>0%)

 Popular error control methods include:

 Parity bits (add a 1 or 0 to the end of each seven
bits)
 Longitudinal redundancy checking (LRC)
 Polynomial checking
 What’s an
“error”?
 Human errors:
 Incorrect IP address assignment, or subnet mask,
etc., etc.

 Network errors:
 Lost data
 Corrupted data (received, but garbled)
Line Noise and Distortion
Errors

Source Likely cause

Line outage Storm, accident

White noise Movement of electrons

Impulse noise Random spikes of power

Cross-talk Guardbands, wires too close

Echo Reflective feedback

Jitter Timing irregularities

Delay distortion Propagation speed

Attenuation Wires too long

 Error
Prevention
 To prevent errors from happening, several
techniques may be applied:
 Proper shielding of cables to reduce interference
 Telephone line conditioning or equalization
 Replacing older media and equipment with new,
possibly digital components
 Proper use of digital repeaters and analog
amplifiers
 Observe the stated capacities of the media
 Error Detection
Methods
 The only way to do error detection and
correction is to send extra data with each
message
 Two common error detection methods:
 Parity checking
 Simple parity
 Longitudinal parity
 Cyclic redundancy checksum (CRC)
 Simple
Parity
 Add an additional bit to each byte in the
message:
 Even parity causes the sum of all bits (including the
parity bit) to be even
 Odd parity causes the sum of all bits to be odd

1 Even parity
0 1 0 1 0 1 0
0 Odd parity
 Exampl
e

Letter 7-bit ASCII Parity bit

D 1 0 1
0 1
0 1 0 0 1

A 1 0 0 0 0 0 1 1

T 1 0 1 0 1 0 0 0

A 1 0 0 0 0 0 1 1
 Longitudinal
Parity
 Add block check character (BCC) to the end of
the message:
 Perform odd parity checking on the block of bits for
each character in the message

0 1 0 1 0 1 0 0
1 0 0 1 1 1 0 1
BCC 0 0 1 1 0 1 1 1
 Exampl
e

Letter 7-bit ASCII Parity bit

D 101 10
0010 1
A 101
0001
01 1
T 1010100 0
A 1000001 1
BCC 1101111 1
 Parity
Checks
 Both simple parity and longitudinal parity do
not catch all errors
 Simple parity only catches odd numbers of bit
errors
 Longitudinal parity is better at catching errors
 But requires too many check bits added to a block of
data
 We need a better error detection method
 What about cyclic redundancy checksum?
 Arithmetic
Checksum
 Used in TCP and IP on the Internet
 Characters to be transmitted are converted to
numeric form and summed
 Sum is placed in some form at the end of the
transmission
 Receiver performs same conversion and
summing and compares new sum with sent
sum
 TCP and IP processes a little more complex
but idea is the same
 Polynomial
Checking
 Adds a character (or series of characters) to
the end of the message based on a
mathematical algorithm:
 Checksum
 Sum the message values and divide by 255. The
remainder is the checksum
D 68
A 65 282
=1 remainder 27
T 84 255
A 65
Checksum 0 0 1 1 0 1 1 1
 Cyclical redundancy
check
 CRC error detection method treats packet of data to be
transmitted as a large polynomial
 Transmitter
 Using polynomial arithmetic, divides polynomial by a
given generating polynomial
 Quotient is discarded
 Remainder is “attached” to the end of message
 Message (with the remainder) is transmitted to the
receiver
 Receiver divides the message polynomial plus the
remainder (checksum) by same generating polynomial
 If a remainder of zero results Ú no error during
transmission
 If a remainder not equal to zero results Ú error during
transmission
 Example:
CRC

7 6 5 4 3 2 1 0
0 0 1 1 0 1 1 1

Message polynomial
x5+x4+x2+x1+x0  x5+x4+x2+x+1

Generating polynomial
ATM CRC x8 + x2 + x + 1
CRC-16 x16 + x15 + x2 + 1
 Error
Control
 Once an error is detected, the receiver can:

1. Toss the frame/packet

 Some newer systems such as frame relay perform
this type of error control
2. Return an error message to the transmitter
 Stop-and-wait error control
 Sliding window error control

3. Fix the error with no further help from the

transmitter
 Toss
frame/packet
 Seems like a strange way to control errors but
some lower-layer protocols such as frame
relay perform this type of error control
 For example, if frame relay detects an error, it
simply tosses the frame
 No message is returned
 Frame relay assumes a higher protocol (such
as TCP/IP) will detect the tossed frame and
ask for retransmission
 Return A
Message
 Once an error is detected, an error
message is returned to the transmitter
 Two basic forms:
 Stop-and-wait error control
 Sliding window error control
 Stop-and-wait Error
Control
 A transmitter sends
a frame then stops
and waits for an
acknowledgment
 If a positive
acknowledgment
(ACK) is received,
the next frame is
sent
 If a negative
acknowledgment
(NAK) is received,
the same frame is
transmitted again
 Sliding Window Error
Control
 These techniques assume that multiple
frames are in transmission at one time
 A sliding window protocol allows the
transmitter to send a number of data packets
at one time before receiving any
acknowledgments
 Depends on window size
 When a receiver does acknowledge receipt,
the returned ACK contains the number of the
frame expected next
 Older sliding window protocols numbered
each frame or packet that was transmitted
 More modern sliding window protocols
number each byte within a frame
 Notice that an ACK is not always sent after
each frame is received
 It is more efficient to wait for a few received frames
before returning an ACK
 How long should you wait until you return an
ACK?
 TCP/IP
 Rule 1: If a receiver just
received data and wants to
send its own data, piggyback
an ACK along with that data
 Rule 2: If a receiver has no data
to return and has just ACKed
the last packet, receiver waits
500 ms for another packet
 If while waiting, another packet
arrives, send the ACK
immediately
 Rule 3: If a receiver has no data
to return and has just ACKed
the last packet, receiver waits
500 ms
 No packet, send ACK
 Packet
Lost
 If a frame is lost, the
following frame will
be “out of
sequence”
 The receiver will hold
the out of sequence
bytes in a buffer and
request the sender to
retransmit the
missing frame
 ACK
Lost
 If an ACK is lost, the
sender will wait for
the ACK to arrive
and eventually time
out
 When the time-out
occurs, the sender
will resend the last
frame
 Correct the
Error
 For a receiver to correct the error with no
further help from the transmitter requires a
large amount of redundant information to
accompany the original data
 This redundant information allows the receiver to
determine the error and make corrections
 This type of error control is often called
forward error correction and involves codes
called Hamming codes
 Hamming codes add additional check bits to a
character
 These check bits perform parity checks on various bits
 For example, what if bit b9 flips?
 The c8 check bit checks bits b12, b11, b10, b9 and c8
(01000)
 This would cause a parity error
 The c4 check bit checks bits b12, b7, b6, b5 and c4
(00101)
 This would not cause a parity error (even number of 1s)
 The c2 check bit checks bits b11, b10, b7, b6, b3 and
c2 (100111)
 This would not cause a parity error
 The c1 check bit checks b11, b9, b7, b5, b3 and c1
(100011)
 This would cause a parity error