Scribd
Upload
4 views

Tutorial Correlation With Answers

The document contains a tutorial on correlation and error evaluation, including calculations for auto-correlation and cross-correlation of codewords, as well as the probability of bit errors in transmission. It discusses the Index of Discrimination for various codewords and the impact of bit errors on the validity of the correlation process. Additionally, it provides formulas and results for calculating the Bit Error Rate (BER) needed to maintain certain error probabilities in codewords of varying lengths.

Uploaded by

ibrahim
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
4 views

Tutorial Correlation With Answers

The document contains a tutorial on correlation and error evaluation, including calculations for auto-correlation and cross-correlation of codewords, as well as the probability of bit errors in transmission. It discusses the Index of Discrimination for various codewords and the impact of bit errors on the validity of the correlation process. Additionally, it provides formulas and results for calculating the Bit Error Rate (BER) needed to maintain certain error probabilities in codewords of varying lengths.

Uploaded by

ibrahim
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Tutorial correlation

1) Find the value of R(2) (auto-correlation) for the codeword: 01011100

2) Find the value of the Index of Discrimination (D) for the auto-correlation given the following
codewords:

i) 11110000
ii) 10101010
iii) 1101

Which of these codewords would be any good for auto-correlation? Why?

3) a) If the BER of a transmission link is BER=1e-3, (i) what is the probability that any single bit in a
codeword is in error?
(ii) What is the probability that there are 0 errors in a codeword of length 32 bits?
(iii) Find (approximately) the BER allowable for the system it we want to ensure that our 32 bit
codewords has 4 errors with probability 1e-6.
Check the accuracy of this approximation by using the full Binomial distribution.

4) Find the value of R(2) for the cross-correlation of the 2 codewords:

a) a[n]=0000 , b[n]=1111, b) a[n]=01010101 , b[n]=11110000

5) For codeword 1101, how many bits in error are required to invalidate the autocorrelation
process? (This is open ended, as there is no certain answer …) Assuming that we want to have the:

probability(correlation process is corrupted) = P(k in error in the codeword of length N bits) = 1e-6

Calculate the BER we need on the communication channel.

6) For some codeword D = 8. How many errors would likely be needed to invalidate the
codeword? What BER is needed to make this number of errors occur with a probability no larger
than 0.001? Assume N=16.
Answers to tutorial questions on correlation and error evaluation
1) R(2) = 01011100
00010111
------------
10110100 So R(2) = 4

2)
11110000 has R(0)=8, R(1)=6, R(2)=4, R(3)=2, R(4)=0, R(5)=2, R(6)=4, R(7)=6 D=2

10101010 has R(0)=8, R(1)=0, R(2)=8, R(3)=0, R(4)=8, R(5)=0, R(6)=8, R(7)=0 D=0

1101 has R(0)=4, R(1)=2, R(2)=2, R(3)=2 D=2

The first is perhaps OK, although not the best for N=8. The second is useless, and the last is as good as can
be achieved with N=4.

3) a) i) p=BER = Pr(any single bit is in error) = 0.001

ii) P(0 errors in 32) = 32C0 . (1-0.001)32 . (0.001)0 = 0.96849

iii) BER ≈ exp( [ln(0.000001) - ln(32C4) ] / 4) = 0.00229


32
P(4 errors in 32, BER=0.0029) = C4 . (1-0.00229)28 . (0.00229)4 = 9.27e-7

4) a) R(2) = 0000
1111 (still looks the same after shift of 2 places!)
------
0000 so R(2) = 0

b) 01010101
00111100
-------------
10010110 so R(2) = 4

5) The question of how many bits are needed to be in error is a complex one, but I would accept 2:
e.g. for shift of 1:

1101
1110 and 2 changes to the lower (shifted) version would suffice to make them appear the same. So:

BER  exp{ (ln(1e-6) - ln(4C2) ) / 2 } = 4e-4

This can then be checked by using the Binomial distribution:

Pr(r=2) = 4C2 . (1 - 4e-4)2 . (4e-4)2 = 9.99e-7 which is OK!


6) For some codeword D = 8. How many errors would likely be needed to invalidate the codeword?
What BER is needed to make this number of errors occur with a probability no larger than 0.001?
Assume N=16.

Difficult to be certain, but we recall that with N=4, D=2, and the number of errors that would
invalidate the autocorrelation process was (we decided) 2. So in this case (N=16, D=8) would
suggest maybe 8 errors would invalidate the autocorrelation process:

ln(Pk ) − ln(NCk) ln(0.001) − ln(16C8)


𝐵𝐸𝑅 ≈ exp ( ) = exp ( )
𝑘 8
ln(0.001) − ln(16C8) −6.9 − 9.46
𝐵𝐸𝑅 ≈ exp ( ) = exp ( ) = 0.129
8 8

You might also like