rocsnr

Receiver operating characteristic curves by SNR

Syntax

[Pd,Pfa] = rocsnr(SNRdB)

[Pd,Pfa] = rocsnr(SNRdB,Name=Value)

rocsnr(___)

Description

[Pd,Pfa] = rocsnr(SNRdB) returns the single-pulse detection probabilities, Pd, and false-alarm probabilities, Pfa, for the SNRs in the vector SNRdB. By default, for each SNR, the detection probabilities are computed for 101 false-alarm probabilities between 1e–10 and 1. The false-alarm probabilities are logarithmically equally spaced. The ROC curve is constructed assuming a coherent receiver with a nonfluctuating target.

[Pd,Pfa] = rocsnr(SNRdB,Name=Value) returns detection probabilities and false-alarm probabilities with additional options specified by one or more name-value arguments.

rocsnr(___) plots the ROC curves.

Examples

ROC Curves for Different SNRs

Plot ROC curves for different SNRs for a single pulse.

SNRdB = [3 6 9 12];

[Pd,Pfa] = rocsnr(SNRdB,SignalType="NonfluctuatingCoherent");

semilogx(Pfa,Pd)
grid on
xlabel("P_{fa}")
ylabel("P_d")
legend("SNR "+SNRdB+" dB",Location="northwest")
title("Receiver Operating Characteristic (ROC) Curves")

ROC Curve for Fixed SNR

To achieve a probability of false alarm of 1e-6, the SNR threshold for Neyman-Pearson detection of a single pulse in real-valued Gaussian noise is approximately 13.5 dB. Plot an ROC curve at that SNR.

snrthreshold = npwgnthresh(1e-6,1,'real');
rocsnr(snrthreshold,'SignalType','real')

ROC Curves for Different Received Signal Types

Examine detector performance for different received signal types at a fixed SNR.

SNR = 13.54;
[Pd_real,Pfa_real] = rocsnr(SNR,'SignalType','real',...
    'MinPfa',1e-8);
[Pd_coh,Pfa_coh] = rocsnr(SNR,...
    'SignalType','NonfluctuatingCoherent',...
    'MinPfa',1e-8);
[Pd_noncoh,Pfa_noncoh] = rocsnr(SNR,'SignalType',...
    'NonfluctuatingNoncoherent','MinPfa',1e-8);
semilogx(Pfa_real,Pd_real)
hold on
grid on
semilogx(Pfa_coh,Pd_coh,'r')
semilogx(Pfa_noncoh,Pd_noncoh,'k')
xlabel('False-Alarm Probability')
ylabel('Probability of Detection')
legend('Real','Coherent','Noncoherent','location','southeast')
title('ROC Curve Comparison for Nonfluctuating RCS Target')
hold off

The ROC curves clearly demonstrate the superior probability of detection performance for coherent and noncoherent detectors over the real-valued case.

Input Arguments

SNRdB — Signal-to-noise ratios

vector

Signal-to-noise ratios in decibels, specified as a row or column vector.

Example: [3 6 9 12]

Data Types: double

Name-Value Arguments

Specify optional pairs of arguments as Name1=Value1,...,NameN=ValueN, where Name is the argument name and Value is the corresponding value. Name-value arguments must appear after other arguments, but the order of the pairs does not matter.

Example: MinPfa=1e-8,NumPoints=64,NumPulses=10

Before R2021a, use commas to separate each name and value, and enclose Name in quotes.

Example: 'MinPfa',1e-8,'NumPoints',64,'NumPulses',10

MaxPfa — Maximum false-alarm probability to include in the ROC calculation

1 (default) | positive scalar

Maximum false-alarm probability to include in the ROC calculation, specified as a positive scalar.

Data Types: double

MinPfa — Minimum false-alarm probability to include in the ROC calculation

1e-10 (default) | positive scalar

Minimum false-alarm probability to include in the ROC calculation, specified as a positive scalar.

Data Types: double

NumPulses — Number of pulses to integrate

1 (default) | positive integer

Number of pulses to integrate when calculating the ROC curves, specified as a positive integer. A value of 1 indicates no pulse integration.

Data Types: double

NumPoints — Number of SNR values to use when calculating the ROC curves

101 (default) | positive integer

Number of SNR values to use when calculating the ROC curves, specified as a positive integer. The actual values are equally spaced between MinSNR and MaxSNR.

Data Types: double

SignalType — Type of received signal

"NonfluctuatingCoherent" (default) | "NonfluctuatingNoncoherent" | "Real" | "Swerling1" | "Swerling2" | "Swerling3" | "Swerling4"

This property specifies the type of received signal or, equivalently, the probability density functions (PDF) used to compute the ROC. Valid values are: "Real", "NonfluctuatingCoherent", "NonfluctuatingNoncoherent", "Swerling1", "Swerling2", "Swerling3", and "Swerling4". Values are not case sensitive.

The "NonfluctuatingCoherent" signal type assumes that the noise in the received signal is a complex-valued, Gaussian random variable. This variable has independent zero-mean real and imaginary parts each with variance σ²/2 under the null hypothesis. In the case of a single pulse in a coherent receiver with complex white Gaussian noise, the probability of detection, P_D, for a given false-alarm probability, P_FA is:

$$P_D=\frac{1}{2}\text{erfc}({\text{erfc}}^{-1}(2P_{FA})-\sqrt{\chi })$$

where erfc and erfc-1 are the complementary error function and that function’s inverse, and χ is the SNR not expressed in decibels.

For details about the other supported signal types, see [1].

Data Types: char | string

Output Arguments

Pd — Detection probabilities

vector

Detection probabilities corresponding to the false-alarm probabilities, returned as a vector. For each SNR in SNRdB, Pd contains one column of detection probabilities.

Pfa — False-alarm probabilities

column vector

False-alarm probabilities, returned as a column vector. By default, the false-alarm probabilities are 101 logarithmically equally spaced values between 1e–10 and 1. To change the range of probabilities, use the optional MinPfa or MaxPfa input argument. To change the number of probabilities, use the optional NumPoints input argument.

References

[1] Richards, M. A. Fundamentals of Radar Signal Processing. New York: McGraw-Hill, 2005, pp 298–336.

Extended Capabilities

C/C++ Code Generation

Generate C and C++ code using MATLAB® Coder™.

Usage notes and limitations:

• Does not support variable-size inputs.

• Supported only when output arguments are specified.

Version History

Introduced in R2011a