You are on page 1of 8

Experiment 2

American International University- Bangladesh


Department of Faculty of Engineering
EEE 4209: Telecommunication Engineering Laboratory

Title: Estimation of Bit Error Probability for Various Modulation Schemes in Wireless
Communications.

Introduction:

The choice of appropriate modulation scheme for transmission depends on the communication system
constraints, i.e. communication system is mainly constrained by signal power and channel bandwidth.
It is thus, indeed, an important aspect which modulation scheme is to choice so that both constraints
can be overcome.

Eb/N0 can be an important parameter for performance measurement, which can eventually provide
with the bit error probability of different modulation schemes. Where bit energy is denoted by Eb (watt-
s), and noise power spectral density by N0 (Watt/Hz).

In this experiment, we are primarily concerned with finding an appropriate modulation scheme at the
transmitter side using bit error probability measurement. We consider modulation schemes such as
M-PAM, M-PSK, M-QAM and M-point orthogonal signal sets. We then draw a conclusion, based
upon the results obtained from these considered schemes, on using a modulation scheme that can
better be fitted in power-limited region and bandwidth-limited region.

Theory and Methodology:

Consider a pair-wise error between two signal points as shown in Figure 2.1. The performance of an
optimal receiver can be found as follows [1] by the pair-wise error probability.

(1)

where the term Q (.) defines Q-function. d represents distance between sm the points and sm [1].

Figure 2.1: Pairwise error between two signal points


If dmin is the minimum distance between the points and Kmin,m is the number of the nearest neighbor from Sm at
tan away of dmin, using union bound estimate (nearest neighbors to Sm will dominate in the union Pr {𝜀𝑚 }, the
union bound estimate of can be found as follows.

(2)

Where Em denote the event in which a decision error is made given hypothesis H=m, i.e. a=Sm for more on
union bound estimate, please refer to [1].

If Kmin is the average value of Kmin,m Equation (2) for symbol error probability Ps can be given as

(3)

Hence, the bit error probability can be given as

(4)

Where M denote the number of points.

Error Performance for PAM, QAM, and Orthogonal Signal Sets

Show in figure 2.2 is a typical gray encoded PAM signal sets (M=16). Consider Es and Eb
respectively denote expected per PAM symbol energy and per PAM bit energy. Note that for binary
PAM Es=Eb. Consequently, Ps=Pb. Hence the two signal points are √Eb and -√Eb. The distance
between them is 2√Eb From equation (3) we can find the following binary PAM (M=2).

(5)

Figure 2.2: Gray encoding for 16-PAM signal sets


Following this approach for M-PAM, it can be found that

(6)

Similarly M’ X M’ for QAM signals sets (see fig 2.3)

(7)

Where M’ = √M

And for orthogonal signal sets (See fig 2.4)

(8)

For PSK, B-PSK is the same as 2-PAM and Q-PSK is the same as 4-QAM and hence so do the per bit
error probability expressions However, for 8- PSK,

(9)

Figure 2.3: Gray encoding for 4 X 4 - QAM signal sets.

Figure 2.4: Typical 2-point orthogonal signal sets [2].


Figure 2.5: Typical M-PSK signal sets.

7.3 Simulation algorithm


Step 1: set the value for M (degree of modulation),
Step 2: set range of values (in dB) for Eb/N0,
Step 3: define expressions of bit error probability for M-PAM, M-QAM and M-point orthogonal
signal sets,
Step 4: run simulation until Eb/N0 reaches the maximum value,
Step 5: plot bit error probability versus Eb/N0.

7.4 Performance Evaluation

Plot experimental graphs and evaluate performance with relevant and appropriate methods. The
followings (Figures 2.6 and 2.7) are example line graphs for bit error probability versus Eb/N0
performance evaluation. So far as from these Figures, it can be found that irrespective of the degree of
modulation, PAM is susceptible more to bit errors than PSK, QAM and orthogonal signal sets. In
addition, increase in degree of modulation results in more errors for PAM, PSK and QAM; however,
the opposite is the case for orthogonal signal sets. Overall, PAM and QAM are bandwidth-efficient
and orthogonal signal sets are power-efficient.

Figure 2.6: Bit error probability for 4-PAM, 2 X 2 QAM and 4-point orthogonal signal sets.
Figure 2.7: Bit error probability for 16-PAM, 4 X 4 QAM and 16-point orthogonal signal sets.

Apparatus:

(1) A PC/Laptop with MATLAB installed.

Report Questions:

Question 01
Plot bit error probability versus Eb / N0 for M-PAM, M-PSK, M-QAM and M-point orthogonal signal sets. Draw a
conclusion on which modulation scheme should you consider for high data rates, based on the results using comparative
analysis method.

Question 02
Repeat the first part of Question 01 for the value of M=64. Mention notable changes from when the value of M=4.

Question 03
For mobile wireless communication systems, which modulation scheme/s should you consider? Justify your
consideration.

Question 04
Find the variables that influence most the bit error probability. You can use equations (6-9) for your analysis.

Question 05
Write a summary on what you have learned, observed, things went wrong (if in your experiment, you have not had the
predicted result), and how you can overcome the similar problem in future.
Reference(s):

[1] P. Saengudomlert, “Digital Communications”, Lecture notes, AT77.13, Asian Institute of Technology
(AIT), Thailand, Jan 2011.

[2] R. G. Gallagar, “Principle of Digital Communication”, course materials for 6.450 Principles of Digital
Communications I, Massachusetts Institute of Technology (MIT), USA, Fall 2006.

Appendix:

Following MATLAB Functions would be helpful for this experiment.

• randint (1, x, [p q]): Generate one dimensional matrix of uniformly distributed random integers
between p and q with a maximum of x number.

• for i=1:0.5:10 end: this command continues a loop for a maximum of 10 with an increment of 0.5
starting from 1.

• semilogx(x) is the same as the command plot(x), but a logarithmic (with base 10) scale is used for
the x-axis. • disp(‘x’): display x on the command window.

• Legend(x1, x2, ...): puts a legend on the current plot using the specified strings as labels (x1, x2).
legend works on line graphs, bar graphs, pie graphs, etc.

Things may get confused

array(x): define any value of a one-dimensional matrix array [x] with a variable x defines the
position of the element in the matrix.
array[x]: defines a one-dimensional matrix. Things may get remembered Any line in MATLAB
must end with a semicolon (;)

If you do not know any command or the functionality of a command, simply write help space that
command in the command window. Example: help disp

Note that MATLAB has a few built-in M-files, which will guide you during your simulation once
called upon them in the command window.

It is a good way to write the program in separate M-file and save that file as .m since writing
directly on the command window may get you lost the written program, if you either forget to save
or clear the command window.
MATLAB Code:

%+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
% Estimation of bit error probability for various modulation schemes in wireless communications
%

% Define Variables

Pb_M_PAM_dB=randint(1,21,[0 0]);
Pb_M_QAM_dB=randint(1,21,[0 0]);
Pb_M_OTH_dB=randint(1,21,[0 0]);

Pb_M_B_PSK_dB=randint(1,21,[0 0]);
Pb_M_4_PSK_dB=randint(1,21,[0 0]);
Pb_M_8_PSK_dB=randint(1,21,[0 0]);
%======================================================
% M-PAM signals
%======================================================

for i=1:1:21

Eb_N0_dB=i-1; % Eb/N0 in dB Eb_N0=10^(Eb_N0_dB/10);

M_PAM=4;
X_PAM=sqrt(((6*log2(M_PAM))/(M_PAM^2-1))*Eb_N0);
Pb_M_PAM=((2*(M_PAM-1))/(M_PAM*log2(M_PAM)))*((1/2)*(1-erf(X_PAM/sqrt(2)))); Pb_M_PAM_dB(i)=log10(Pb_M_PAM);

%============================================================
% M-QAM signals
%============================================================
M_QAM=sqrt(M_PAM); X_QAM=sqrt(((6*log2(M_QAM))/(M_QAM^2-1))*Eb_N0);
Pb_M_QAM=((2*(M_QAM-1))/(M_QAM*log2(M_QAM)))*((1/2)*(1-erf(X_QAM/sqrt(2)))); Pb_M_QAM_dB(i)=log10(Pb_M_QAM);

%============================================================
% M-orthogonal signals
%============================================================
M_OTH=M_PAM;
X_OTH=sqrt(log2(M_OTH)*Eb_N0);
Pb_M_OTH=(M_OTH/2)*((1/2)*(1-erf(X_OTH/sqrt(2))));
Pb_M_OTH_dB(i)=log10(Pb_M_OTH);

%============================================================
% M-PSK signals: B-PSK
%============================================================
M_B_PSK=2; % note 2-PAM is the same as B-PSK X_B_PSK=sqrt(2*Eb_N0);
Pb_M_B_PSK=(1/2)*(1-erf(X_B_PSK/sqrt(2)));
Pb_M_B_PSK_dB(i)=log10(Pb_M_B_PSK);

%============================================================
% M-PSK signals: Q-PSK
%============================================================
M_4_PSK=sqrt(M_PAM);
X_4_PSK=sqrt(((6*log2(M_4_PSK))/(M_4_PSK^2-1))*Eb_N0);
Pb_M_4_PSK=((2*(M_4_PSK-1))/(M_4_PSK*log2(M_4_PSK)))*((1/2)*(1-erf(X_4_PSK/sqrt(2))));
Pb_M_4_PSK_dB(i)=log10(Pb_M_4_PSK);
%============================================================
% M-PSK signals: 8-PSK
%============================================================
M_8_PSK=sqrt(M_PAM*2);
X_8_PSK=sqrt(1.5*(2-sqrt(2))*Eb_N0);
Pb_M_8_PSK=((2/3)*(1-erf(X_8_PSK/sqrt(2))));
Pb_M_8_PSK_dB(i)=log10(Pb_M_8_PSK);
%============================================================

end
% Output variables EN_dB=0:1:20;
plot(EN_dB,Pb_M_PAM_dB,EN_dB,Pb_M_QAM_dB,EN_dB,Pb_M_OTH_dB,EN_dB,Pb_M_B_PSK_dB,EN_dB,Pb_M_4_PSK_dB,EN_d
B,Pb_M_8_PS K_dB)
legend('4-PAM','4-QAM','4-point-OTH','B-PSK','Q-PSK','8-PSK');

%=============================================================

You might also like