Detection Performance for DC Level

in White Gaussian Noise (WGN)

Submitted By:
Aatika Arshad
aa07993
Stats And Inferencing

Date: th November 2024

Brief Description of the Code
This MATLAB code plots the probability of detection (PD ) versus the energy-to-noise ratio (ENR) in
dB for various false alarm probabilities (PF A ) in a white Gaussian noise (WGN) environment. The code
uses the built-in ‘qfunc‘ and ‘qfuncinv‘ functions in MATLAB to compute the Q-function and its inverse,
which are essential for calculating the detection probability.
The PD is calculated for a range of ENR values, and for each PF A value, a separate curve is generated.
The resulting plot helps visualize how the probability of detection improves as the ENR increases for
different values of PF A .

1
 MATLAB Code
1 % MATLAB Code to plot Probability of Detection ( P_D ) vs . SNR for various P_FA
2

3 % Parameters
4 PFA_values = [1 e -1 , 1e -2 , 1e -3 , 1e -4 , 1e -5 , 1e -6 , 1e -7]; % Values of P_FA
5 SNR_dB = linspace (0 , 20 , 100) ; % Range of SNR values in dB
6 N = 1; % Assuming N = 1 for simplicity
7

8 % Plotting
9 figure ;
10 hold on ;
11 colors = lines ( length ( PFA_values ) ) ; % Generate unique colors for each line
12

13 for i = 1: length ( PFA_values )

14 PFA = PFA_values ( i ) ;
15 threshold = qfuncinv ( PFA ) ; % Calculate threshold using built - in qfuncinv
16

17 % Calculate P_D for each SNR

18 PD = arrayfun ( @ ( snr ) qfunc ( threshold - sqrt ( N * 10^( snr / 10) ) ) , SNR_dB ) ;
19

20 % Plot the result with legend entry

21 plot ( SNR_dB , PD , ’ LineWidth ’ , 1.5 , ’ Color ’ , colors (i ,:) , ...
22 ’ DisplayName ’ , sprintf ( ’ P_ { FA } = 10^{% d } ’ , log10 ( PFA ) ) ) ;
23 end
24

25 % Customize the plot

26 xlabel ( ’ Energy - to - noise - ratio ( dB ) 10 log_ {10} ( NA ^2 / \ sigma ^2) ’ , ’ FontSize ’ ,
12) ;
27 ylabel ( ’ Probability of detection , P_D ’ , ’ FontSize ’ , 12) ;
28 title ( ’ Detection performance for DC level in WGN ’ , ’ FontSize ’ , 14) ;
29 legend ( ’ show ’ , ’ Location ’ , ’ southeast ’) ; % Show legend in the bottom right
30 grid on ;
31 set ( gca , ’ FontSize ’ , 12) ; % Set axis font size for readability
32 hold off ;

Listing 1: MATLAB code for plotting PD vs ENR for different PF A values

2
Resulting Plot

Figure 1: Detection performance plot showing PD vs. ENR for various PF A values.

3
Code Explanation
The MATLAB code consists of the following main steps:
1. Parameters Setup: We define an array of values for PF A (False Alarm Probability) and a range
for ENR in dB.
2. Q-function and Threshold Calculation: The threshold for each PF A is calculated using ‘qfuncinv‘.
For each ENR value, the probability of detection, PD , is computed using ‘qfunc‘ and stored.

Task 2
We are tasked with determining the necessary number of samples N for a detection problem in white
Gaussian noise, given the following parameters:

• False Alarm Probability PF A = 10−4

• Detection Probability PD = 0.99

• Signal-to-Noise Ratio (SNR) in dB:

A2
 
10 log10 = −32 dB
σ2

Approach
The formula relating PD , PF A , and N for detection in white Gaussian noise is:
r !
−1 N A2
PD = Q Q (PF A ) −
σ2
Rearranging to solve for N :
r
−1 −1 N A2
Q (PD ) = Q (PF A ) −
σ2
Then,
r
N A2
= Q−1 (PF A ) − Q−1 (PD )
σ2
Squaring both sides,

N A2
2
2
= Q−1 (PF A ) − Q−1 (PD )
σ
Finally, we have:

σ2
2
N= 2
Q−1 (PF A ) − Q−1 (PD )
A

Converting SNR from dB

 
A2
Given 10 log10 σ2 = −32 dB, we can convert this to a linear ratio:

A2 −32

2
= 10 10 = 10−3.2
σ

Substituting Values
Now, we need to compute Q−1 (PF A ) and Q−1 (PD ) using the inverse Q-function and substitute these
values into the equation for N .

4
MATLAB Code
The following MATLAB code can be used to compute the necessary number of samples N :

% Given values
P_FA = 10e-4;
P_D = 0.99;
SNR_dB = -32;

% Convert SNR from dB to linear scale

A2_over_sigma2 = 10^(SNR_dB / 10);

% Calculate Q-inverse values

Q_inv_P_FA = qfuncinv(P_FA);
Q_inv_P_D = qfuncinv(P_D);

% Calculate necessary number of samples N

N = (1 / A2_over_sigma2) * (Q_inv_P_FA - Q_inv_P_D)^2;

% Display result
fprintf(’The necessary number of samples N is: %.2f\n’, N);

Explanation of Code
• **Convert SNR from dB to linear scale**: This step converts the SNR value from dB to a linear
2
ratio to get A
σ2 .

• **Use ‘qfuncinv‘**: The MATLAB function ‘qfuncinv‘ is used to compute the inverse Q-function
values for PF A and PD .
• **Substitute into the equation**: The values of Q−1 (PF A ) and Q−1 (PD ) are substituted into the
equation to compute N .
• **Display result**: The result is displayed as the necessary number of samples N .

Results
After executing the MATLAB code, the result for the necessary number of samples N is calculated. For
the given parameters, the value of N is approximately:

N ≈ 46499.72
Thus, the necessary number of samples required is N ≈ 46499.72.