Types of Errors

• A network system
y that cannot gguarantee that
the data received are completely identical to
Chapter 5 the data transmitted is essentiallyy useless.
• Reliable systems must have a mechanism for

Error Detection detecting and correcting the errors.

errors
• There are two types of errors:
– Single‐bit
Si l bi error: one bit
bi is
i changed
h d ffrom 0 to 1 or
from 1 to 0
– Burst
B t error: multiple
lti l bit
bits are changed
h d
1 2

Figure 5-1 Figure 5-2

Types of Errors
Single Bit Error
Single-Bit

ASCII LF ASCII STX

3 4
 Figure 5-3

Single‐Bit
Single Bit Error Burst Error of Length
g Five

• Single‐bit
Single bit errors are least likely type of error in
serial transmission – the duration of the noise
is normally much longer than that of a bit bit.
• However single‐bit error can happen in
parallel transmission
transmission, such as data transfer
inside a computer, between CPU and memory,
or between a computer and a printer
printer.
– E.g. one noisy line out of eight in a parallel printer
cable can corrupt one bit in each byte
byte.
5 6

Burst Error Error Detection

• Burst errors is most likely to happen in a serial • To detect an error, the redundancy technique is
transmission. used, where a short group of bits are appended to
• Burst errors does not mean errors occur in the end of each data unit to be sent.
consecutive bits. • The extra redundant bits carry information for error
detection for the data unit.
• The number of bits affected depends on the data
• If the received data stream pass the checking
rate and duration of noise.
function, the redundant bits are discarded.
• E.g.
E g if date rate is 1 Mbps and noise duration is
• Three types of redundancy checks are used: vertical
1/100 s, the number of bits affected is redundancy check (VRC), longitudinal redundancy
1 Mbps × 1/100 s = 10
10,000
000 bits check
h k (LRC)
(LRC), and
d cyclic
li redundancy
d d check
h k (CRC).
(CRC)
7 8
Figure 5-4 Figure 5-5
Redundancy
Detection Methods

9 10

Figure 5-6

Even Parity VRC Concept

Vertical Redundancy Check (VRC)
• Most common and least expensive mechanism
for error detection, often called parity check
• A parity bit is appended to every data unit so that
the total number of 1s in the unit, including the
parity bit, is even (or odd).
• Can detect all single‐bit errors and burst errors
that change odd (or even) number of bits.
• Cannot detect errors that change even (or odd)
number of bits
bits.
11 12
 Figure 5-7

LRC
Longitudinal Redundancy Check (LRC)
• A block of data is organized in a table of rows and
columns. The parity bit for each column is then
calculated. All parity bits are grouped and appended
to the original data to be sent (Fig. 5‐7).
• LRC increases the likelihood of detecting burst errors
– an LRC of n bits can detect a burst error of n bits.
• Cannot detect errors where there are even number
off d
damageddbbits that
h are located
l d in the
h same column
l
position of different rows.

13 14

An LRC Example Cyclic Redundancy Check (CRC)

• Suppose the following block is sent: • The most powerful redundancy checking technique
<=== 10101001 00111001 11011101 11100111 10101010
• A sequence of redundant bits, called CRC, is
(LRC)
appended
pp to the end of a data unit so that the
• However it is hit by a burst noise of length eight resulting data unit is exactly divisible by a
<=== 10100011 10001001 11011101 11100111 10101010 predetermined binary number.
(LRC)
• At the destination, the data unit is divided by the
• The receiver checks the LRC and finds some bits do same binary number. If remainder is 0, the data is
not follow even
even‐parity
parity rule
rule, so the whole block is assumed d to b
be intact andd accepted.d Iff remainder
d is
discarded. not 0, the data is considered damaged and rejected.
<<=== 10100011 10001001 11011101 11100111 10101010
(LRC)
15 16
 Figure 5-8

CRC Generation CRC Generator and Checker

• CRC bits are derived by dividing the data unit by
the predetermined divisor – the remainder is the
CRC.
• CRC is always one bit less than the divisor.
• CRC is found as follows:
1) First a string of n 0s is appended to the data unit
(assume divisor has n+1 bits)
2) Second the new data unit is divided by the divisor
using binary division. The remainder is the CRC.
3)) Third
h d the
h n CRC bits
b replacel the
h appended
d d 0s.
17 18

Figure 5-9 Figure 5-10

Binary Di
Bi Division
ii
in a Binary Division
CRC GGenerator in a
CRC Checker

19 20
 Binary Division in a CRC Generator Binary Division in a CRC Checker

• Suppose we want to transmit the If no bits are lost or

information string: 1111101.
1111101 corrupted,
t d dividing
di idi theth
• The receiver and sender decide received information
to use the (arbitrary) polynomial string by the agreed upon
pattern, 1101. pattern will give a
• The information string is shifted remainder of zero.
left by one position less than the
number of positions in the We see this is so in the
divisor. calculation at the right.
• The remainder is found through R l applications
Real li ti use
modulo 2 division (at right) and longer binary numbers to
added to the information string: cover larger information
1111101000 + 111 = 1111101111. strings.
21 22

Figure 5-11

Polynomials A Polynomial of Degree Seven

• The CRC divisor is most often represented as an

algebraic polynomial, rather than a binary number,
for purpose of mathematical proof.
• The degree of a polynomial is its highest power.
• Anyy polynomial
p y selected must have the followingg
properties:
– It should not be divisible by x
– It should be divisible by x+1
• Several common polynomials are used (see Fig. 5‐13)

23 24
Figure 5-12 Figure 5-13

A Polynomial
y Representing
p g a Divisor Standard Polynomials

CRC‐32 is used in both Ethernet and Token Ring network.

25 26

Performance Error Correction

• CRC can detect all burst errors affecting an odd • Error correction can be handled in two ways:
y
number of bits. – When an error is detected, the sender is notified and the
entire data unit is retransmitted
• CRC can detect all burst errors of length
g less than or
– The receiver uses an error‐correcting code to correct the
equal to the degree of the polynomial
errors automatically
• It can detect with a veryy high
g probability
p y burst errors • Error correction by retransmission is practical and
of length greater than the degree of the polynomial. common method in networking and communication.
• E.g. CRC‐12 will detect all burst errors affecting an • Error correcting codes are more complicated than
odd number of bits, will detect all burst errors with error detecting codes. Hamming code is a popular
length ≤ 12, and will detect 99.97% of the burst single‐bit
g error correction method.
errors with l th ≥ 12.
ith a length 12
27 28
 Summaryy
• Errors can be categorized as a single bit or burst.
• Three
h types off redundancy
d d checks
h k are used d in
i LANs:
VRC, LRC and CRC.
• In
I VRC,
VRC a parityit bit is
i added
dd d tto th
the d
data
t unit.
it
• In LRC, data unit is organized into a table, and a
parity bit is calculated from each column.
column
• CRC, the most powerful of the three, is based on
binary division.
division
• In CRC, the generator, using a specific divisor, creates
redundant bits that are appended to the datadata. The
checker uses the same divisor to verify the data.
29