IERG2310 Tutorial 11

Xiangyu Zhong

. . . . . .
Xiangyu Zhong (CUHK) IERG2310 Tutorial 11 1 / 12
Outline

Error Control Coding

I Basic Concepts

I Linear Binary Block Code

I Exercises

. . . . . .
Xiangyu Zhong (CUHK) IERG2310 Tutorial 11 2 / 12
Basic Concepts
Error Control Coding
I The coding procedure done to control the occurrences of errors,
including error detection, and/or error correction.

I Basic idea: add redundancy (i.e., extra bits). Or, put another way,
disallow some codewords; if you receive such an invalid codeword,
correct it to the closest valid codeword.

Binary Block Codes

I An (n, k) binary block code contains successive k message bits and
(n − k) check/partiy bits:

I Check bits are computed from the message bits according to the
specific encoding rule.
. . . . . .
Xiangyu Zhong (CUHK) IERG2310 Tutorial 11 3 / 12
Basic concepts

Let C is the set of all valid codewords for an (n, k) binary block code,
and c1 , c2 ∈ C are 2 valid codewords.

Concept Definition
Code Rate k/n
Hamming Distance dH (c1 , c2 ) := |c1 − c2 |
d∗H := min dH (c1 , c2 )
Minimum Distance c1 ̸=c2 ,
c1 ,c2 ∈C

Hamming Weight wH (c) := dH (c, 0)

∗ :=
Minimum Weight wH min wH (c)
c̸=0,
c∈C

. . . . . .
Xiangyu Zhong (CUHK) IERG2310 Tutorial 11 4 / 12
Basic concepts

Table: Error Detection and Correction Capabilities

Description Expression
A code can detect up to d errors
d∗H ≥ d + 1
per codeword if:
A code can correct up to c errors
d∗H ≥ 2c + 1
per codeword if:
A code can detect up to d errors
and correct up to c < d errors per d∗H ≥ d + c + 1
codeword if:

. . . . . .
Xiangyu Zhong (CUHK) IERG2310 Tutorial 11 5 / 12
Linear Binary Block Code

Linearity: Any linear combination of two valid codewords is also an valid

codeword.

The all-zero sequence 0 is always a codeword.

The minimum distance equals the minimum weight, i.e., d∗H = wH

∗
.

Example:
Hamming codes — can detect up to two-bit errors or correct single-bit
errors.
Repetition codes (p.3) and single parity check bit codes (p.2).

. . . . . .
Xiangyu Zhong (CUHK) IERG2310 Tutorial 11 6 / 12
Exercise

Exercise 1:
Consider a binary linear block code of the form:
[d1 d2 d3 p1 p2 p3 ]
where d1 , d2 , d3 are message bits and p1 , p2 , p3 are check bits given by
p1 = d1 + d2 mod 2
p2 = d2 + d3 mod 2
p3 = d1 + d3 mod 2
list out all the codewords.

. . . . . .
Xiangyu Zhong (CUHK) IERG2310 Tutorial 11 7 / 12
Exercise
Exercise 1:
Consider a binary linear block code of the form:
[d1 d2 d3 p1 p2 p3 ]
where d1 , d2 , d3 are message bits and p1 , p2 , p3 are check bits given by
p1 = d1 + d2 mod 2
p2 = d2 + d3 mod 2
p3 = d1 + d3 mod 2
list out all the codewords.

Solution:
[0 0 0 0 0 0], [0 0 1 0 1 1]
[0 1 0 1 1 0], [0 1 1 1 0 1]
[1 0 0 1 0 1], [1 0 1 1 1 0]
[1 1 0 0 1 1], [1 1 1 0 0 0]
. . . . . .
Xiangyu Zhong (CUHK) IERG2310 Tutorial 11 8 / 12
Exercise

Exercise 2:
Consider an (8, 4) binary linear block code with minimum distance of 4.
I How many valid codewords are there?
I What is the code rate?
I What is the minimum weight of the code?
I If the code is used for error detection only, how many errors can it
detect?
I If the code is used for error correction, how many errors can it correct?

. . . . . .
Xiangyu Zhong (CUHK) IERG2310 Tutorial 11 9 / 12
Exercise
Exercise 2:
Consider an (8, 4) binary linear block code with minimum distance of 4.
I How many valid codewords are there?
I What is the code rate?
I What is the minimum weight of the code?
I If the code is used for error detection only, how many errors can it
detect?
I If the code is used for error correction, how many errors can it correct?

Solution:
I 24 = 16.
I 4/8 = 0.5.
∗ ∗
I wH = 4, since wH = d∗H for linear block codes.
∗
I d = 4, by dH ≥ d + 1.
I c = 1, by d∗H ≥ 2c + 1.

. . . . . .
Xiangyu Zhong (CUHK) IERG2310 Tutorial 11 10 / 12
Exercise

Exercise 3:
Consider a binary linear block code of the form:
[d1 d2 p1 p2 p3 ]
where d1 , d2 are message bits and p1 , p2 , p3 are check bits given by
p1 = d1 + d2 mod 2
p2 = d1 mod 2
p3 = d2 mod 2
Get the error detecting and correcting capabilities.

. . . . . .
Xiangyu Zhong (CUHK) IERG2310 Tutorial 11 11 / 12
Exercise
Exercise 3:
Consider a binary linear block code of the form:
[d1 d2 p1 p2 p3 ]
where d1 , d2 are message bits and p1 , p2 , p3 are check bits given by
p1 = d1 + d2 mod 2
p2 = d1 mod 2
p3 = d2 mod 2
Get the error detecting and correcting capabilities.
Solution:
First, we can list all the valid codewords as follows:
[0 0 0 0 0], [0 1 1 0 1]
[1 0 1 1 0], [1 1 0 1 1]
∗
As observed, the minimum weight wH = 3. Since this is a linear block code,
∗ ∗
we have dH = wH = 3.
Thus, the error detection capability is d = 2 and the code correction
capability is c = 1.
. . . . . .
Xiangyu Zhong (CUHK) IERG2310 Tutorial 11 12 / 12
 Assignment 5

Xiangyu Zhong (CUHK) ENGG2310 Tutorial 4 13 / 18

Q16

The low-pass signal x(t) with bandwidth W is sampled with a sampling interval of
Ts , and the signal
∞
X
xp (t) = x(nTs )p(t − nTs )
n=−∞

is generated, where p(t) is an arbitrary shaped pulse (not necessarily time-limited

to the interval [0, Ts ]).
(a) Find the Fourier transform of xp (t) in terms of the Fourier transform of p(t),
namely P (f ).
(b) Find the conditions for perfect reconstruction of x(t) from xp (t), and specify
the required reconstruction filter.

Xiangyu Zhong (CUHK) ENGG2310 Tutorial 4 14 / 18

Q17

Consider a quantizer. The output of the quantizer xQ is given by:



 −0.75, x < −0.5

 −0.25, −0.5 ≤ x < 0
xQ =


 0.25, 0 ≤ x < 0.5
0.75, 0.5 ≤ x


The input samples are distributed according to the probability density function:
(
|x|, 0 ≤ x ≤ 1
fX (x) =
0, else

(a) Determine the signal-to-quantization noise ratio (SQNR).

(b) Write a program to determine the optimal quantizer.

Xiangyu Zhong (CUHK) ENGG2310 Tutorial 4 15 / 18

Q17

clear; clc;

% PDF and range

pdf = @(x) abs(x) .* (x >= -1 & x <= 1); % PDF of X (symmetric |x| o
bins = 4; % Number of quantization levels
thresholds = linspace(-1, 1, bins+1); % Initial decision boundaries
delta = 1e-6; diff = 1; % Convergence criteria

while diff > delta

% Step 1: compute centroids (quantization levels)
for i = 1:bins
a = thresholds(i); b = thresholds(i+1);
num(i) = integral(@(x) x .* pdf(x), a, b); % \int x * f_X(x)
den(i) = integral(@(x) pdf(x), a, b); % \int f_X(x) dx
end
q = num ./ den; % Updated quantization levels

Xiangyu Zhong (CUHK) ENGG2310 Tutorial 4 16 / 18

Q17

% Step 2: update thresholds (except for first and last)

new_thresholds = thresholds;
for i = 2:bins
new_thresholds(i) = (q(i-1) + q(i)) / 2;
end

% Convergence check
diff = max(abs(new_thresholds - thresholds));
thresholds = new_thresholds; % Update thresholds
end

% Display results
disp(’Optimal Quantization Levels:’);
disp(q);
disp(’Optimal Decision Boundaries:’);
disp(thresholds);

Xiangyu Zhong (CUHK) ENGG2310 Tutorial 4 17 / 18

Q18

Simulate the sigma-delta modulation with MATLAB according to the following

steps:
Generate the signal x(t) = sin(πt) + 0.5 cos 3πt


1 2
R
2 Generate y(t) = x(t)dt
3 Apply delta modulation to y(t) at a very high sampling rate.
4 Apply 1-bit D/A, i.e., construct a pulse train using the outputs of the delta
modulator.
5 Smooth the resulting signal.

Xiangyu Zhong (CUHK) ENGG2310 Tutorial 4 18 / 18