Tuesday, July 2, 2024 6:35 PM

error control

topic : error control

if data received is not same as the data sent then this means error has occured.

types of error :
(i) single-bit error

(ii) burst error (most likely to occur)

example :
b= 1kbps = 103bits/second
noise duration = 1/10sec
number of corrupted bits (or) affected bits = 103bits/second x 1/10sec = 100bits

formula :
number of corrupted bits = bandwidth * noise duration
(number of corrupted bits or affected bits depends on the data rate and duration of noise)

concept : redundancy

Chapter - 2 (Error control) Page 1

 data : 10101010
- one bit : 8c1 = 8
- two bit : 8c2 = 28

error detection error correction

data + data(redundancy) sent : 10101010
- aryan@9927 recv : 10001000
- aryam@9927 or.d : 10101010

receiver can only detect receiver can detect + how many and
which bits are changed

concept : logic for error detection

(a) datawords : k
- combinations : 2k
(b) redundant bits : r
(c) codeword(n) = k+r
- combinations : 2k
(d) invalid codeword : 2k-2n

Chapter - 2 (Error control) Page 2

 dataword codeword (valid)
00 000
01 011
10 101
11 110
(k) n = (k+r)
n>k

(i) error detection using block code

dataword codeword (valid)

00 000
01 011
10 101
11 110

(i) if the recieved codeword is same as the one of the valid codeword, the word is accepted.

no error
000 000 valid codeword (accepted by receiver)
sender receiver

(ii) the received codeword is not valid, it is discarded.

one bit error

000 100 invalid codeword (rejected by receiver and
sender receiver receiver able to detect the error)

dataword codeword (valid)

00 000
01 011
codeword not available
10 101
11 110

(iii) the codeword is corrupted during transmission but the recieved word still matches a valid
codeword, the error remains undetected.

two bit error

000 011 valid codeword (accepted by receiver and
Chapter - 2 (Error control) Page 3
 two bit error
000 011 valid codeword (accepted by receiver and
sender receiver receiver is not able to detect the error)

dataword codeword (valid)

00 000
01 011 codeword available
10 101
11 110

error because
sent : 000
recieved : 011

topic : hamming distance

concept : minimum hamming distance

Chapter - 2 (Error control) Page 4

example : d(a,b)=3 010
(i) d(a,c)=1 101
(a)010 d(a,d)=2 111
(b)101 d(b,c)=2 101
(c)110 d(b,d)=1 110
(d)001 d(c,d)=3 011
minimum hamming distance : 1

concept : minimum hamming distance for error detection

(i) minimum hamming distance = 1

one bit error

010 110 valid codeword (accepted by receiver and
sender receiver receiver is not able to detect the error)

codeword (valid) d(a,b)=3

000 d(a,c)=1
101 d(a,d)=2
110 codeword available d(b,c)=2
001 d(b,d)=1
d(c,d)=3

valid one bit error valid

(receiver is not able to detect the error)
codeword codeword

sender reciever

(ii) minimum hamming distance = 2

(a) one bit error : can detect one bit error

one bit error

000 100 invalid codeword (able to detect the error)
sender receiver

Chapter - 2 (Error control) Page 5

 one bit error
000 100 invalid codeword (able to detect the error)
sender receiver

codeword (valid) d(a,b)=2

000 d(a,c)=2
011 codeword not d(a,d)=2
101 available d(b,c)=2
110 d(b,d)=2
d(c,d)=2

(b) two bit error : cannot detect two bit error

two bit error

000 011 valid codeword (accepted by receiver and
sender receiver receiver is not able to detect the error)

codeword (valid) d(a,b)=2

000 d(a,c)=2
011 codeword available d(a,d)=2
101 d(b,c)=2
110 d(b,d)=2
d(c,d)=2

valid one bit error invalid one bit error valid

codeword (receiver is able to codeword (receiver is able to codeword
detect the error) detect the error)

sender reciever
sender reciever

two bit error

(receiver is not able to detect the error)

- minimum hamming distance : 1

1bit change karne par dusra valid codeword ban jayega
(not able to detect error)

- minimum hamming distance : 2

1bit change karne par invalid codeword ban jayega
(able to detect error)

- minimum hamming distance : 2

Chapter - 2 (Error control) Page 6
 - minimum hamming distance : 2
2bit change karne par dusra valid codeword ban jayega
(not able to detect error)

- minimum hamming distance : 3

1bit change karne par invalid codeword ban jayega
(able to detect error)

- minimum hamming distance : 3

2bit change karne par invalid codeword ban jayega
(able to detect error)

- minimum hamming distance : 3

3bit change karne par dusra valid codeword ban jayega
(not able to detect error)

formula :
(i) if minimum hamming distance = d
then we can detect upto (d-1) bit error.
(ii) to detect "d" bit error, minimum hamming distance required = d+1

topic : simple parity

(i) simple parity concept one extra bit (parity bit) is added to each data word.
(ii) simple parity check can detect all single bit error.
(iii) simple parity check can not detect an even number of errors, but it can detect an odd
number of errors.

even parity odd parity

- number of 1's must be - number of 1's must be
even in each codeword odd in each codeword
including the parity including the parity
bit. bit

message : 00|01|10|11 (dataword)

by using even parity : number of 1's must be even.

Chapter - 2 (Error control) Page 7
 by using even parity : number of 1's must be even.

dataword codeword (valid)

00 000 parity bit
01 011
10 101 even 1's
11 110

k=2 bit
r=1
codeword (n) = k+r = 2+1=3

(a) one bit error : can detect one bit error

one bit error

000 100 invalid codeword (able to detect the error)
sender receiver

codeword (valid)
000
011 codeword not
101 available
110

(b) two bit error : cannot detect two bit error

two bit error

000 110 valid codeword (accepted by receiver and
sender receiver receiver is not able to detect the error)

codeword (valid)
000
011
101
110 codeword available

(c) three bit error : can detect odd bit error

Chapter - 2 (Error control) Page 8

(c) three bit error : can detect odd bit error

two bit error

000 111 invalid codeword (able to detect the error)
sender receiver

codeword (valid)
000
011 codeword not available
101
110

topic : 2d parity check code

(i) can detect and correct all single bit error

(ii) can detect two and three bit error that can occur anywhere in matrix
(iii) only some pattern with four or more error can be detected
(iv) the information bits are organized in a matrix consisting of row and columns, for each row and
each column one parity check bits is calculated.

using even parity = number of 1's even

1st row 2nd row 3rd row 4th row 5th row

number of rows : 5
number of columns : 6

0 row parity
1
1
1
0
0 1 0 0 0 0 1

column parity

number of rows : 6
number of columns : 7

transmitted 0100001
data :
6th row

reciever will also divide the data in row and column

Chapter - 2 (Error control) Page 9

 transmitted 0100001
data :
6th row

reciever will also divide the data in row and column

one bit error

it is 1 instead of 0 means
there is error in bits

it is 1 instead
original :
of 0 means
there is error
in bits

- it can detect and correct all single bit error.

- one bit error will effect a parity bit.
(11 : 1 row + 1 column)

maximum affected parity

original :
- it can detect two bit but cannot correct two bit error.
- two bit error will effect maximum 4 parity.
(1011: 2 row + 2 column)

minimum affected parity

Chapter - 2 (Error control) Page 10

original :
- two bit error will effect minimum 2 parity.
(00 : 2 column)

- kisi bhi 2 bit mai change karne par wahi same distrubance
hoga. saari bits change karne par bhi islie not able to find
kaunse bits ki wajah se error hoga.

maximum affected parity

original :
- it can detect three bit error but cannot correct three bit error.

- three bit error will effect maximum 6 parity.

(011100 : 3 row + 3 column)

Chapter - 2 (Error control) Page 11

 minimum affected parity

original :
- three bit error will effect minimum 2 parity.
(01 : 1 row + 1 column)

it cannot detect 4 bit error it can detect 4 bit error

original :
it can and sometimes cannot detect 4 bit error

disadvantages of 2d parity :
- if we have a error in the parity then this scheme does not work fine.

Chapter - 2 (Error control) Page 12

 0 error in parity (it should be 1)

1
error in parity (it should be 0)

0 changed to 1 but actual mai yeh change nahi hui.

topic : linear block codes

valid codeword
(xor)
valid codeword
valid codeword

Chapter - 2 (Error control) Page 13

 a : 000
(xor)
b : 011
r: 011
a : 000
(xor)
c : 101
r: 101

minimum number of
1's in a non-zero valid
code word :
- valid code words:
011 : 2
101 : 2
110 : 2

(ii) assume this is a linear block code : minimum number of 1's in a non-zero valid code word :
000 - valid code words:
001 000 : 0
110 001 : 1
111 110 : 2
010 111 : 3
011 010 : 1
011 : 2

minimum hamming distance = 1

topic : cyclic code

linear block + 1 extra property

right shift left shift

example : valid codeword

000
011
101
110

right shift left shift

Chapter - 2 (Error control) Page 14
110

right shift left shift

- 011 - 101 - 011 - 110
- 101 - 110 - 101 - 011
- 110 - 011 - 110 - 101
the length of the codeword : 3

- 3 shift operation karne baad wahi same codeword wapas aa jayega

011 - (i) 110 - (ii) 101 - (iii) 011

concept : crc(cyclic redundancy check)

(i) length of the dataword = n

(ii) length of the divisor= k
(iii) append (k-1) zero's to the original message
(iv) perform modulo 2 divison (xor)
(v) remainder of the divison = crc
(vi) total bits in code word = n+k-1 (bits)
(vii) code word = dataword with appended (k-1) zeros + crc

point to remember :
crc must be k-1 bits

example :

data = 1001001
divisor or crc generator = 1101

(i) length of the dataword = n = 1001001 = 7

(ii) length of the divisor= k = 1101 = 4
(iii) append (k-1) zero's to the original message
- (k-1) = (4-1) = 3
1001001 = 1001001000
(iv) perform modulo 2 divison
1101|1001001000
1101
0100001000
1101
010101000
1101
01111000
1101
0010000
1101

Chapter - 2 (Error control) Page 15

 1101
0010000
1101
01010
1101
0111

(v) remainder of the divison = crc

crc = k-1 bits
crc = 4-1 = 3 bits
crc = 111
(vi) code word = n+k-1 bit = 7+4-1 = 10bit
(vii) code word = dataword with appended (k-1) zeros + crc
- 1001001000
+111
1001001111

(a) if reciever recieves uncorrupted data:

perform modulo 2 divison

1101|1001001111
1101
0100001111
1101
010101111
1101
01111111
1101
0010111
1101
01101
1101
0000 syndrome = 0 (no error)
dataword accepted.

(b) if reciever recieves corrupted data:

1101|1011001111
1101
0110001111
1101
000101111
1101
011011
1101
00001 syndrome ≠ 0 (error)
dataword rejected.

Chapter - 2 (Error control) Page 16

 00001
dataword rejected.

concept : polynomial notation in crc

(i) dataword = d(x)
(ii) codeword = c(x)
(iii) generator = g(x)
(iv) syndrome = s(x)
(v) error = e(x)

how to apply crc step by step :

(i) determine the degree 'r' of g(x) (highest power)
(ii) determine xrd(x)
(iii) append (k-1) zero's to the original message
(iii) determine the remainder by dividing x rd(x) by g(x)
(iv) codeword = xrd(x)+remainder (crc)

example :
dataword d(x) = 1001001
divisor (or) generator g(x) = 1101

- dataword d(x) = 1001001

1 0 0 1 0 0 1
6 5 4 3
x x x x x2 x 1 x 0
d(x) = x6+x3+x0(1)
d(x) = x6+x3+1

- divisor (or) generator g(x) = 1101

1 1 0 1
3 2
x x x1 x0
g(x) = x3+x2+x0(1)
g(x) = x3+x2+1

(i) determine the degree 'r' of g(x) (highest power)

g(x) = x3+x2+1
1101
r:3
(ii) determine xrd(x)
x3(x6+x3+1)
x9+x6+x3
1001001000
total : 10bits
number of 1's : 3
number of 0's : 7

Chapter - 2 (Error control) Page 17

 1001001 000
d(x) (k-1)

(iii) derermine the remainder by dividing x rd(x) by g(x)

x3+x2+1 | x9+x6+x3 1101|1001001000

x9+x8+x6 1101
x8+x3 0100001000
x8+x7+x5 1101
x7+x5+x3 010101000
x7+x6+x4 1101
x6+x5+x4+x3 01111000
x6+x5+x3 1101
x4 0010000
x4+x3+x 1101
x3+x 01010
x3+x2+1 1101
x2+x+1 0111
x2+x+1
crc/remainder : 111

(iv) codeword = xrd(x)+remainder (crc)

= x9+x6+x3+x2+x+1
= 1001001 111
d(x) (k-1)

concept : cycle code analysis

sent code word

error

Chapter - 2 (Error control) Page 18

 sent code word

error

c(x) - sent codeword = recieved codeword

=0 - rem = 0 (100%) - crc
g(x)

sent code word c(x)

1 bit error

recieved code word

c(x) + e(x) x3 x 2 x1 x 0

e(x) = x3
1 0 0 0 e(x) = 1000
error = 1000

sent code word : 10110101 point to remember :

error : 1000 (+) recieved codeword have error, is error correct?
recieved word : 10111101 (i) bit changed from 0 to 1
- add (codeword + error) to check if error is correct.
(ii) bit changed from 1 to 0
- subtract (codeword - error) to check if error is correct.

(i) either e(x) = 0(no error) crc scheme is working fine

e(x)
=0
g(x)
e(x)
(ii) e(x)≠0, but we are getting = 0 it means e(x) is divisible by g(x)
g(x)
ex . (i)
dataword = 1101001
divisor = 1001

Chapter - 2 (Error control) Page 19

divisor = 1001

sender : 1001 1101001000

1001
0100001000
1001
000101000
1001
001100
1001
0101 (crc = k-1 = 4-1 = 3) = 101

codeword = xrd(x)+remainder (crc)

1101001000 + 101
1101001000
101
1101001101

sent codeword : 1 1 0 1 0 0 1 1 0 1

recieved
codeword : 1 1 0 1 0 0 0 1 0 0
x3 x0

c(x): 1 1 0 1 0 0 1 1 0 1 1 0 0 1 e(x) = x3 + x0
e(x) = 1001

1101001101
1001 (-)
1101000100

reciever : 1001 1101000100

1001
0100000100
1001
000100100
1001
000000 (crc = k-1 = 4-1 = 3) = 000

syndrome = 0 (no error)

dataword accepted, but crc scheme
failed to detect error.

Chapter - 2 (Error control) Page 20

 failed to detect error.

original sent codeword : 1101001101

recieved codeword : 1101000100

e(x)
=0
g(x)

1001
=0
1001

point to remember :
(i) crc is not perfect scheme if e(x) is divisible by g(x) then that error cannot be detected
(ii) probability of such error is very less, hence error detection probability of crc is very high.

syndrome : s(x)
(i) if s(x)≠0, then code word is rejected and crc scheme is working fine.
(ii) if s(x)=0 and e(x)=0 then code word is accepted and crc scheme is working fine.
(iii) if s(x)=0 and e(x)≠0 [e(x) is divisible by g(x)] then codeword accepted and scheme failed
to detect the error.

concept : choosing divisor in crc

sent code word = 1 0 1 0 1 0 1 0

recv code word = 1 0 1 1 1 0 1 0
x4
error : x4+x3.0+x2.0+x1.0+x0.0 = x4= 10000

sent code word = 1 0 1 0 1 0 1 0

recv code word = 1 0 1 0 0 0 1 0
x3
error : x3.1+x2.0+x1.0+x0.0 = x3= 1000

point to remember :
last bit ki error ko chodkar saari error even number de rahi hai (in single bit)
if e(x) = 10 - [2nd bit error] ; 2
if e(x) = 100 - [3rd bit error] ; 4
if e(x) = 1000 - [4th bit error] ; 8
if e(x) = 10000 - [5th bit error] ; 16
if e(x) = 100000 - [6th bit error] ; 32
g(x) = 10 ; 2
Chapter - 2 (Error control) Page 21
g(x) = 10 ; 2
g(x) = 100 ; 4
g(x) = 1000 ; 8
g(x) = 10000 ; 16

not able to detect error when :

e(x) = 10 even
g(x) = 10 = =0
even

problem : except for last bit error, single bit error gives even number so divisor must not be
even. if divisor is also even then remainder will be zero, so we can not caught single bit
error.

solution : to detect all single bit error the last bit of the divisor must be 1, so that divisor
becomes an odd number and hence all single bit error detected.

points to remember :
(i) if the generator has more than one term (x+1) and coefficient of x 0 is 1, all single bit error
can be detected.
(ii) if generator cannot divide x t+1 (t between 0 and n-1) then all isolated double error can be
detected.
(iii) a generator that contains a factor of x+1 can detect all odd numbered errors.

example :
(i)generator = 1
data : 1011
it is invalid generator, r : 0 (no redundant bits)

(ii)generator = x : x1.1+x0.0 = 10
data : 1011

sender : 10 10110
10
00110
10
010
10
00 crc (or) remainder

one bit error

10110 10010
sender receiver

reciever : 10 10010
10
Chapter - 2 (Error control) Page 22
 10
00010
10
00 syndrome = 0 (dataword accepted)
crc failed to detect error.

e(x) = x2 = 100(4)
g(x) = 10 (2)

e(x) = 100
0
g(x) = 10 =

point to remember :
g(x) should not contain 'x'

(iii)generator = x+1: x1.1+x0.1 = 11

data : 1011

sender : 11 10110
11
01110
11
0010
11
01 crc (or) remainder

one bit error

10111 10101
sender receiver

reciever : 11 10101
11
01110
11
0010
11
01 syndrome ≠ 0 (dataword rejected)
crc passed to detect error.

e(x) = x1 = 10(2)
g(x) = 11 (3)

e(x) = 10
0
g(x) = 11 ≠

Chapter - 2 (Error control) Page 23

topic : hamming code
(i) hamming code can detect + correct 1 bit error.
(ii) hamming code can detect upto 2 bit error.
(iii) hamming code is used for error correction.

- message = m bits
- n(codeword)=m(dataword)+r(redundant/check/parity bits)

- according to the hamming code, number of redundant bits are :

m+r+1 ≤ 2r (this equation should be satisfied)
where, r = lower limitation
yeh equation satisfy honi chaiye 'r' ki minimum value ke lie toh mujhe redundant bit mil jayegi

example :
- dataword/message : 1010111, m = 7
m+r+1 ≤ 2r

r:1 = 7+1+1 ≤ 21 = 9 ≤ 2 (no)

r:2 = 7+2+1 ≤ 22 = 10 ≤ 4 (no)
r:3 = 7+3+1 ≤ 23 = 11 ≤ 8 (no)
r:4 = 7+4+1 ≤ 24 = 12 ≤ 16 (yes) : minimum value of 'r'.
r:5 = 7+5+1 ≤ 25 = 13 ≤ 32 (yes)
r:6 = 7+6+1 ≤ 26 = 14 ≤ 64 (yes)

minimum value of 'r' = 4 (4 extra redundant bits)

Chapter - 2 (Error control) Page 24

 n =m+r bits
n = 7+4=11 bits

positioning of redundant bits :

2i = 20, 21, 22, 23, 24...
1, 2, 4, 8, 16 positions

even parity : number of 1's even

11bits : 1 2 3 4 5 6 7 8 9 10 11
positions : 1, 2, 4, 8

p1 p2 3 p4 5 6 7 p8 9 10 11

1010111 : 1 0 1 0 1 1 1

- calculating p1 : - calculating p2 :
choose 1 bit and skip 1 bit start from 2, choose 2 bit and skip 2 bit

p1 3 5 7 9 11 p2 3 6 7 10 11
1 0 0 1 1 1 1 0 1 1
p1 3 5 7 9 11 p2 3 5 7 10 11
1 1 0 0 1 1 0 1 1 0 1 1

- calculating p4 : - calculating p8 :
start from 4, choose 4 bit and skip 4 bit start from 8, choose 8 bit and skip 8 bit

p4 5 6 7 p8 9 10 11
0 1 0 1 1 1
p4 5 6 7 p8 9 10 11
1 0 1 0 1 1 1 1

- 1011

p1 p2 3 p4 5 6 7 p8 9 10 11
1 0 1 1 0 1 0 1 1 1 1

codeword : 10110101111

1 2 3 4 5 6 7 8 9 10 11
transmitted data : 1 0 1 1 0 1 0 1 1 1 1

Chapter - 2 (Error control) Page 25

(i) if reciever recieved uncorrupted data
1 2 3 4 5 6 7 8 9 10 11
recieved data : 1 0 1 1 0 1 0 1 1 1 1

p1 : p2 :
1 3 5 7 9 11 2 3 6 7 10 11
1 1 0 0 1 1 0 1 1 0 1 1
number of 1's : even (o) number of 1's : even (o)

p4 : p8 :
4 5 6 7 8 9 10 11
1 0 1 0 1 1 1 1
number of 1's : even (o) number of 1's : even (o)

p8 p 4 p 2 p 1
0 0 0 0 sabki value 0, no error.

(ii) if reciever recieved corrupted data

1 2 3 4 5 6 7 8 9 10 11
recieved data : 1 0 1 1 0 1 0 1 0 1 1
p1 : p2 :
1 3 5 7 9 11 2 3 6 7 10 11
1 1 0 0 0 1 0 1 1 0 1 1
number of 1's : odd(1) number of 1's : even (o)

p4 : p8 :
4 5 6 7 8 9 10 11
1 0 1 0 1 0 1 1
number of 1's : even (o) number of 1's : odd(1)

p8 p4 p2 p1
1 0 0 1 1, error.
decimal value : 1001 = 9
error in 9th bit, 9th bit got corrupted.

(iii) if reciever recieved corrupted data (2bit error)

1 2 3 4 5 6 7 8 9 10 11
recieved data : 1 0 1 1 1 1 0 1 0 1 1

p1 : p2 :
1 3 5 7 9 11 2 3 6 7 10 11
1 1 1 0 0 1
Chapter - 2 (Error control) Page 26
p1 : p2 :
1 3 5 7 9 11 2 3 6 7 10 11
1 1 1 0 0 1 0 1 1 0 1 1
number of 1's : even (0) number of 1's : even (o)

p4 : p8 :
4 5 6 7 8 9 10 11
1 1 1 0 1 0 1 1
number of 1's : odd(1) number of 1's : even(0)

p8 p4 p2 p1
1 1 0 0 1, error.
decimal value : 1100 = 12
error in 12th bit, 12th bit does not exist.

point to remember :
if parity is not given then odd parity is preferable over even parity.

Chapter - 2 (Error control) Page 27

Chapter - 2 (Error control) Page 28
Chapter - 2 (Error control) Page 29
Chapter - 2 (Error control) Page 30
topic : checksum
to detect error
it can be 4bit, 8bit, 16bit, 32bit

dataword : 01111011110000000110
lets assume checksum = 4bit.
dataword : 0111 1011 1100 0000 0110
decimal : 7 11 12 0 6
checksum : 7+11+12+0+6=36

technique (i) : send the data along with checksum, reciever will divide checksum into 4 bits and
calculate the sum, if the sum of checksum(sent) is equal to checksum(reciever) then no error.
transmitted data : 01111011110000000110 36
checksum

technique (ii) : send the data along with checksum, reciever will recieve the checkum in negative, if
the difference between checksum is 0 then no error.

transmitted data : 01111011110000000110 36

checksum

recieved data : 01111011110000000110 - 36

checksum
36-36=0

but there is a problem with both techniques, the maximum bits we can store in 4 bits is 15 (1111) but
the checksum is 36.

better approach : internet checksum

Chapter - 2 (Error control) Page 31

(i) checksum = 36 : 100100(bits needed to write 36), still 6bits are needed.
so we will add the extra bits (wrap extra bits)

100100
+10
0110
1001 one's complement

decimal value : 9 (checksum)

transmitted data : 01111011110000000110 1001
checksum
(i) if reciever recieves uncorrupted data

0111 1011 1100 0000 0110 1001

decimal : 7 11 12 0 6 9
checksum : 7+11+12+0+6+9=45 : 101101

101101
+10
1111
0000 one's complement
all zero : no error

Chapter - 2 (Error control) Page 32