Bit Error Rate Analysis

Analyze BER performance of communications systems

Description

The Bit Error Rate Analysis app calculates the bit error rate (BER) as a function of the energy per bit to noise power spectral density ratio (E_b/N₀). Using this app, you can:

• Generate BER data for a communications system and analyze performance using:

  • Monte Carlo simulations of MATLAB^® functions and Simulink^® models.

  • Theoretical closed-form expressions for selected types of communications systems.

  • Run systems contained in MATLAB simulation functions or Simulink models. After you create a function or model that simulates the system, the Bit Error Rate Analysis app iterates over your choice of E_b/N₀ values and collects the results.

• Plot one or more BER data sets on a single set of axes. You can graphically compare simulation data with theoretical results or simulation data from a series of communications system models.

• Fit a curve to a set of simulation data.

• Plot confidence levels of simulation data.

• Send BER data to the MATLAB workspace or to a file for further processing.

For more information, see Use Bit Error Rate Analysis App.

×

Open the Bit Error Rate Analysis App

• MATLAB Toolstrip: On the Apps tab, under Signal Processing and Communications, click the app icon.

• MATLAB command prompt: Enter bertool.

Examples

Compute BER Using Theoretical Tab

Generate a theoretical estimate of BER performance for a 16-QAM link in AWGN.

Open the Bit Error Rate Analysis app.

bertool

On the Theoretical tab, set these parameters to the specified values: E_b/N₀ range to 0:10, Modulation type to QAM, and Modulation order to 16.

Plot the BER curve by clicking Plot.

Compute BER Using Monte Carlo Tab and MATLAB Function Simulation

Simulate the BER by using a custom MATLAB function. By default, the app uses the viterbisim.m simulation.

Open the Bit Error Rate Analysis app.

bertool

On the Monte Carlo tab, set the E_b/N₀ range parameter to 1:.5:6. Run the simulation and plot the estimated BER values by clicking Run.

On the Theoretical tab, set E_b/N₀ range to 1:6 and set Modulation order to 4. Enable convolutional coding by selecting Convolutional. Click Plot to add the theoretical upper bound of the BER curve to the plot.

Prepare MATLAB Function for Use in Bit Error Rate Analysis App

Add code to the simulation function template given in the Template for Simulation Function topic to run in the Monte Carlo tab of the Bit Error Rate Analysis.

Prepare Function

Copy the template from the Template for Simulation Function topic into a new MATLAB file in the MATLAB Editor. Save the file in a folder on your MATLAB path, using the file name bertool_simfcn.

Place lines of code that initialize parameters or create objects used in the simulation in the template section marked Set up initial parameters. This code maps simulation variables to the template input arguments. For example, snr maps to EbNo.

% Set up initial parameters.
siglen = 1000; % Number of bits in each trial
M = 2;         % DBPSK is binary
snr = EbNo;    % Because of binary modulation

% Create an ErrorRate calculator System object to compare 
% decoded symbols to the original transmitted symbols.
errorCalc = comm.ErrorRate;

Place the code for the core simulation tasks in the template section marked Proceed with simulation. This code includes the core simulation tasks, after all setup work has been performed.

    msg = randi([0,M-1],siglen,1); % Generate message sequence
    txsig = dpskmod(msg,M);        % Modulate
    hChan.SignalPower = ...        % Calculate and assign signal power
        (txsig'*txsig)/length(txsig); 
    rxsig = awgn(txsig,snr,'measured'); % Add noise
    decodmsg = dpskdemod(rxsig,M);      % Demodulate
    berVec = errorCalc(msg,decodmsg);   % Calculate BER
    totErr = totErr + berVec(2);
    numBits = numBits + berVec(3);

After you insert these two code sections into the template, the bertool_simfcn function is compatible with the Bit Error Rate Analysis app. The resulting code resembles this code segment.

function [ber,numBits] = bertool_simfcn(EbNo,maxNumErrs,maxNumBits,varargin)
%
%   See also BERTOOL and VITERBISIM.

% Copyright 2020 The MathWorks, Inc.

% Initialize variables related to exit criteria.
totErr = 0;  % Number of errors observed
numBits = 0; % Number of bits processed

% --- Set up the simulation parameters. ---
% --- INSERT YOUR CODE HERE.
% Set up initial parameters.
siglen = 1000; % Number of bits in each trial
M = 2;         % DBPSK is binary.
snr = EbNo;    % Because of binary modulation
% Create an ErrorRate calculator System object to compare
% decoded symbols to the original transmitted symbols.
errorCalc = comm.ErrorRate;

% Simulate until the number of errors exceeds maxNumErrs
% or the number of bits processed exceeds maxNumBits.
while((totErr < maxNumErrs) && (numBits < maxNumBits))

    % Check if the user clicked the Stop button of BERTool.
    if isBERToolSimulationStopped(varargin{:})
      break
    end
  
    % --- Proceed with the simulation.
    % --- Update totErr and numBits.
    % --- INSERT YOUR CODE HERE.
    msg = randi([0,M-1],siglen,1); % Generate message sequence
    txsig = dpskmod(msg,M);        % Modulate
    hChan.SignalPower = ...        % Calculate and assign signal power
        (txsig'*txsig)/length(txsig); 
    rxsig = awgn(txsig,snr,'measured'); % Add noise
    decodmsg = dpskdemod(rxsig,M);      % Demodulate
    berVec = errorCalc(msg,decodmsg);   % Calculate BER
    totErr = totErr + berVec(2);
    numBits = numBits + berVec(3);
end % End of loop

% Compute the BER.
ber = totErr/numBits;

The function has inputs to specify the app and scalar quantities for EbNo, maxNumErrs, and maxNumBits that are provided by the app. The Bit Error Rate Analysis app is an input because the function monitors and responds to the stop command in the app. The bertool_simfcn function excludes code related to plotting, curve fitting, and confidence intervals because the Bit Error Rate Analysis app enables you to do similar tasks interactively without writing code.

Use Prepared Function

Run bertool_simfcn in the Bit Error Rate Analysis app.

Open the Bit Error Rate Analysis app, and then select the Monte Carlo tab.

Set these parameters to the specified values: E_b/N₀ range to 0:10, Simulation environment to MATLAB, Function name to bertool_simfcn, Number of errors to 5, and Number of bits to 1e8.

Click Run.

The Bit Error Rate Analysis app computes the results and then plots them. In this case, the results do not appear to fall along a smooth curve because the simulation required only five errors for each value in EbNo.

Fit a curve to the series of points in the BER Figure window, by selecting the Fit parameter in the data viewer.

The Bit Error Rate Analysis app plots the fitted curve, as shown in this figure.

Compute Error Rate Simulation Sweeps Using Bit Error Rate Analysis App

Use the Bit Error Rate Analysis app to compute the BER as a function of ${\mathit{E}}_{\mathrm{b}}/{\mathit{N}}_0$. The app analyzes performance with either Monte Carlo simulations of MATLAB® functions and Simulink® models or theoretical closed-form expressions for selected types of communications systems. The code in the mpsksim.m function provides an M-PSK simulation that you can run from the Monte Carlo tab of the app.

Open the Bit Error Rate Analysis app from the Apps tab or by running the bertool function in the MATLAB® command window.

On the Monte Carlo tab, set the ${\mathit{E}}_{\mathbf{b}}/{\mathit{N}}_0$ range parameter to 1:1:5 and the Function name parameter to mpsksim.

Open the mpsksim function for editing, set M=2, and save the changed file.

Run the mpsksim.m function as configured by clicking Run on the Monte Carlo tab in the app.

After the app simulates the set of ${\mathit{E}}_{\mathrm{b}}/{\mathit{N}}_0$ points, update the name of the BER data set results by selecting simulation0 in the BER Data Set field and typing M=2 to rename the set of results. The legend on the BER figure updates the label to M=2.

Update the value for M in the mpsksim function, repeating this process for M = 4, 8, and 16. For example, these figures of the Bit Error Rate Analysis app and BER Figure window show results for varying M values.

Parallel SNR Sweep Using Bit Error Rate Analysis App

The default configuration for the Monte Carlo processing of the Bit Error Rate Analysis app automatically uses parallel pool processing to process individual ${\mathit{E}}_{\mathrm{b}}/{\mathit{N}}_0$ points when you have the Parallel Computing Toolbox™ software but for the processing of your simulation code:

• Any parfor function loops in your simulation code execute as standard for loops.

• Any parfeval (Parallel Computing Toolbox) function calls in your simulation code execute serially.

• Any spmd (Parallel Computing Toolbox) statement calls in your simulation code execute serially.

Prepare Simulink Model for Use with Bit Error Rate Analysis App

Use a Simulink simulation model to run in the Monte Carlo tab of the Bit Error Rate Analysis app. Compare the BER performance of the Simulink simulation results with theoretical BER results.

Prepare Model

Open the model by entering doc_bpsk at the MATLAB command prompt.

doc_bpsk

Initialize parameters in the MATLAB workspace to avoid using undefined variables as block parameters.

EbNo = 0;
maxNumErrs = 100;
maxNumBits = 1e8;

Ensure that the Bit Error Rate Analysis app uses the correct amount of noise each time it runs the simulation, by opening the dialog box for the AWGN Channel block and verifying that the Es/No parameter is set to EbNo.

Note

For BPSK modulation, E_s/N₀ is equivalent to E_b/N₀.

Ensure that the Bit Error Rate Analysis app uses the correct stopping criteria for each iteration by:

• Opening the dialog box for the Error Rate Calculation block and verifying that Target number of errors is set to maxNumErrs and that Maximum number of symbols is set to maxNumBits.

• Verifying that the simulation stop time is set to Inf.

Enable the Bit Error Rate Analysis app to access the BER results that the Error Rate Calculation block computes, by ensuring that the BER variable name parameter in the app matches the Variable name parameter set in the To Workspace (Simulink) block that connects to the output of the Error Rate Calculation block.

Tip

More than one To Workspace block exists. Select the To Workspace block from the DSP System Toolbox™ / Sinks sublibrary.

Use Prepared Model

Run the doc_bpsk model in the Bit Error Rate Analysis app.

Open the Bit Error Rate Analysis app, and then select the Monte Carlo tab.

Set these parameters to the specified values: E_b/N₀ range to 0:9, Simulation environment to Simulink, Function name to doc_bpsk, Number of errors to 100, and Number of bits to 1e8.

Click Run.

The Bit Error Rate Analysis app computes the results and then plots them.

Compare these simulation results with the theoretical results, by clicking the Theoretical tab in the Bit Error Rate Analysis app and setting E_b/N₀ range to 0:9.

Click Plot.

The Bit Error Rate Analysis app plots the theoretical curve in the BER Figure window along with the earlier simulation results.

Parameters

Theoretical

Eb/N0 range — Range of E_b/N₀ values

0:18 (default) | scalar | vector

Range of E_b/N₀ values over which the BER is evaluated, specified as a scalar or vector. Units are in dB.

Example: 5:10 specifies the evaluation of E_b/N₀ values over the range [5, 10] at 1 dB increments.

Channel type — Type of channel over which BER is evaluated

AWGN (default) | Rayleigh | Rician

Type of channel over which the BER is evaluated, specified as AWGN, Rayleigh, or Rician. The Rayleigh and Rician options correspond to flat fading channels.

Modulation type — Modulation type of communications link

PSK (default) | DPSK | OQPSK | PAM | QAM | FSK | MSK | CPFSK

Modulation type of the communications link, specified as PSK, DPSK, OQPSK, PAM, QAM, FSK, MSK, or CPFSK.

Modulation order — Modulation order of communications link

2 (default) | 4 | 8 | 16 | 32 | 64

Modulation order of the communications link, specified as 2, 4, 8, 16, 32, or 64.

Differential encoding — Differential encoding of input data

off (default) | on

Select this parameter to enable differential encoding of the input data.

Correlation coefficient — Correlation coefficient

0 (default) | real scalar in the range [-1, 1]

Correlation coefficient, specified as a real scalar in the range [-1, 1].

Dependencies

To enable this parameter, set Modulation type to FSK.

Modulation index — Modulation index

0.5 (default) | positive real scalar

Modulation index, specified as a positive real scalar.

Dependencies

To enable this parameter, set Modulation type to CPFSK.

Demodulation type — Coherent demodulation of input data

on (default) | off

• Select this parameter to enable coherent demodulation of the input data.

• Clear this parameter to enable noncoherent demodulation of the input data.

Dependencies

To enable this parameter, set Modulation type to FSK or MSK.

Channel coding — Channel coding type used when estimating theoretical BER

None (default) | Convolutional | Block

Channel coding type used when estimating the theoretical BER, specified as None, Convolutional, or Block.

Synchronization — Synchronization error

Perfect synchronization (default) | Normalized timing error | RMS phase noise level

Synchronization error in the demodulation process, specified as Perfect synchronization, Normalized timing error, or RMS phase noise (rad).

• When you set Synchronization to Perfect synchronization no synchronization errors are encountered in the demodulation process.

• When you set Synchronization to Normalized timing error, you can set the normalized timing error as a scalar in the range [0, 0.5].

• When you set Synchronization to RMS phase noise (rad), you can set the RMS phase noise level as a nonnegative scalar. Units are in radians

Dependencies

To enable this parameter, set Modulation type to PSK, Modulation order to 2, and Channel coding to None.

Decision method — Decoding decision method

Hard (default) | Soft

Decoding decision method used to decode the received data, specified as Hard or Soft.

Dependencies

To enable this parameter, set Channel coding to Convolutional or set Channel coding to Block and set Coding type to General.

Trellis — Convolutional code trellis

poly2trellis(7,[171 133]) (default) | structure

Convolutional code trellis, specified as a structure variable. You can generate this structure by using the poly2trellis function.

Dependencies

To enable this parameter, set Channel coding to Convolutional.

Coding type — Block coding type

General (default) | Hamming | Golay | Reed-Solomon

Block coding type used in the BER evaluation, specified as General, Hamming, Golay, or Reed-Solomon.

Dependencies

To enable this parameter, set Channel coding to Block.

N — Codeword length

positive integer

Codeword length, specified as a positive integer.

Dependencies

To enable this parameter, set Channel coding to Block and set Coding type to General.

K — Message length

positive integer

Message length, specified as a positive integer such that K is less than N.

Dependencies

To enable this parameter, set Channel coding to Block and set Coding type to General.

dmin — Minimum distance of (N,K) block code

positive integer

Minimum distance of the (N,K) block code, specified as a positive integer.

Dependencies

To enable this parameter, set Channel coding to Block and set Coding type to General.

Monte Carlo

Eb/N0 range — Range of E_b/N₀ values

1:0.5:5 (default) | scalar | vector

Range of E_b/N₀ values over which the BER is evaluated, specified as a scalar or vector. Units are in dB.

Example: 4:2:10 specifies evaluation of E_b/N₀ over the range [4, 10] at 2 dB increments.

Simulation environment — Simulation environment

MATLAB (default) | Simulink

Simulation environment, specified as MATLAB or Simulink.

Function name — Name of MATLAB function

viterbisim (default)

Name of the MATLAB function for the app to run for the Monte Carlo simulation.

Dependencies

To enable this parameter, set Simulation environment to MATLAB.

Model name — Name of Simulink model

commgraycode (default)

Name of the Simulink model for the app to run for the Monte Carlo simulation.

Dependencies

To enable this parameter, set Simulation environment to Simulink.

BER variable name — Name of variable containing BER simulation data

grayBER (default)

Name of the variable containing the BER simulation data. To output the BER simulation data to the MATLAB workspace, you can assign this variable name as the Variable name parameter value in a To Workspace (Simulink) block.

Tip

More than one To Workspace block exists. Select the To Workspace block from the DSP System Toolbox / Sinks sublibrary.

Dependencies

To enable this parameter, set the Simulation environment to Simulink.

Number of errors — Number of errors to be measured before simulation stops

100 (default) | positive integer

Number of errors to be measured before the simulation stops, specified as a positive integer. Typically, to produce an accurate BER estimate,100 measured errors are enough.

Number of bits — Number of bits to be processed before simulation stops

1e8 (default) | positive integer

Number of bits to be processed before the simulation stops, specified as a positive integer. This parameter is used to prevent the simulation from running too long.

Note

The Monte Carlo simulation stops when either the number of errors or number of bits threshold is reached.

Tips

• You can stop the simulation by clicking Stop on the Monte Carlo Simulation dialog box.

  

Compatibility Considerations

Semianalytic tab in the Bit Error Rate Analysis has been removed

Behavior changed in R2020b

The Semianalytic tab and functionality in the Bit Error Rate Analysis app has been removed. To generate semianalytic BER results, you can still use the semianalytic function.

For example, this code shows you how to use the semianalytic function to programmatically generate semianalytic BER results for a BPSK-modulated signal.

data = [0 1 1 0 0 1 1 1 1 0 1 1 0 0 0 0].';
bpskmod = comm.BPSKModulator
txSig = rectpulse(bpskmod(data),16); 
rxSig = rectpulse(bpskmod(data),16); % Before receive filter 
modType = ‘psk’; 
modOrder = 2; 
sps = 16; % samples per symbol
num = ones(16,1) / 16; % Filter numerator 
den = 1 % Filter denominator 
EbNo = 0:18; % dB 
BER = semianalytic(txSig,rxSig,modType,modOrder,sps,num,den,EbNo); 
semilogy(EbNo,BER)