Main Content

Generate Code for an Averaging Filter

This example shows the recommended workflow for generating C code from a MATLAB® function using the codegen command. These are the steps:

1. Add the %#codegen directive to the MATLAB function to indicate that it is intended for code generation. This directive also enables the MATLAB code analyzer to identify warnings and errors specific to MATLAB for code generation.

2. Generate a MEX function to check that the MATLAB code is suitable for code generation. If errors occur, you should fix them before generating C code.

3. Test the MEX function in MATLAB to ensure that it is functionally equivalent to the original MATLAB code and that no run-time errors occur.

4. Generate C code.

5. Inspect the C code.

Prerequisites

There are no prerequisites for this example.

About the averaging_filter Function

The averaging_filter.m function acts as an averaging filter on the input signal; it takes an input vector of values and computes an average for each value in the vector. The output vector is the same size and shape as the input vector.

type averaging_filter
% y = averaging_filter(x)
% Take an input vector signal 'x' and produce an output vector signal 'y' with
% same type and shape as 'x' but filtered.
function y = averaging_filter(x) %#codegen
% Use a persistent variable 'buffer' that represents a sliding window of
% 16 samples at a time.
persistent buffer;
if isempty(buffer)
    buffer = zeros(16,1);
end
y = zeros(size(x), class(x));
for i = 1:numel(x)
    % Scroll the buffer
    buffer(2:end) = buffer(1:end-1);
    % Add a new sample value to the buffer
    buffer(1) = x(i);
    % Compute the current average value of the window and
    % write result
    y(i) = sum(buffer)/numel(buffer);
end

The %#codegen compilation directive indicates that the MATLAB code is intended for code generation.

Create Some Sample Data

Generate a noisy sine wave and plot the result.

v = 0:0.00614:2*pi;
x = sin(v) + 0.3*rand(1,numel(v));
plot(x, 'red');

Figure contains an axes object. The axes object contains an object of type line.

Generate a MEX Function for Testing

Generate a MEX function using the codegen command. The codegen command checks that the MATLAB function is suitable for code generation and generates a MEX function that you can test in MATLAB prior to generating C code.

codegen averaging_filter -args {x}
Code generation successful.

Because C uses static typing, codegen must determine the properties of all variables in the MATLAB files at compile time. Here, the -args command-line option supplies an example input so that codegen can infer new types based on the input types. Using the sample signal created above as the example input ensures that the MEX function can use the same input.

By default, codegen generates a MEX function named averaging_filter_mex in the current folder. This allows you to test the MATLAB code and MEX function and compare the results.

Test the MEX Function in MATLAB

Run the MEX function in MATLAB

y = averaging_filter_mex(x);
% Plot the result when the MEX function is applied to the noisy sine wave.
% The 'hold on' command ensures that the plot uses the same figure window as
% the previous plot command.
hold on;
plot(y, 'blue');

Figure contains an axes object. The axes object contains 2 objects of type line.

Generate C Code

codegen -config coder.config('lib') averaging_filter -args {x}
Code generation successful.

Inspect the Generated Code

The codegen command with the -config coder.config('lib') option generates C code packaged as a standalone C library. The generated C code is in the codegen/lib/averaging_filter/ folder. The files are:

dir codegen/lib/averaging_filter/
.                              ..                             _clang-format                  averaging_filter.a             averaging_filter.c             averaging_filter.h             averaging_filter.o             averaging_filter_data.c        averaging_filter_data.h        averaging_filter_data.o        averaging_filter_initialize.c  averaging_filter_initialize.h  averaging_filter_initialize.o  averaging_filter_rtw.mk        averaging_filter_terminate.c   averaging_filter_terminate.h   averaging_filter_terminate.o   averaging_filter_types.h       buildInfo.mat                  codeInfo.mat                   codedescriptor.dmr             compileInfo.mat                examples                       interface                      rtw_proj.tmw                   rtwtypes.h                     

Inspect the C Code for the averaging_filter.c Function

type codegen/lib/averaging_filter/averaging_filter.c
/*
 * File: averaging_filter.c
 *
 * MATLAB Coder version            : 23.2
 * C/C++ source code generated on  : 19-Aug-2023 11:10:49
 */

/* Include Files */
#include "averaging_filter.h"
#include "averaging_filter_data.h"
#include "averaging_filter_initialize.h"
#include <string.h>

/* Variable Definitions */
static double buffer[16];

/* Function Definitions */
/*
 * Use a persistent variable 'buffer' that represents a sliding window of
 *  16 samples at a time.
 *
 * Arguments    : const double x[1024]
 *                double y[1024]
 * Return Type  : void
 */
void averaging_filter(const double x[1024], double y[1024])
{
  double dv[15];
  int i;
  int k;
  if (!isInitialized_averaging_filter) {
    averaging_filter_initialize();
  }
  /*  y = averaging_filter(x) */
  /*  Take an input vector signal 'x' and produce an output vector signal 'y'
   * with */
  /*  same type and shape as 'x' but filtered. */
  for (i = 0; i < 1024; i++) {
    double b_y;
    /*  Scroll the buffer */
    memcpy(&dv[0], &buffer[0], 15U * sizeof(double));
    /*  Add a new sample value to the buffer */
    b_y = x[i];
    buffer[0] = b_y;
    /*  Compute the current average value of the window and */
    /*  write result */
    for (k = 0; k < 15; k++) {
      double d;
      d = dv[k];
      buffer[k + 1] = d;
      b_y += d;
    }
    y[i] = b_y / 16.0;
  }
}

/*
 * Use a persistent variable 'buffer' that represents a sliding window of
 *  16 samples at a time.
 *
 * Arguments    : void
 * Return Type  : void
 */
void averaging_filter_init(void)
{
  memset(&buffer[0], 0, 16U * sizeof(double));
}

/*
 * File trailer for averaging_filter.c
 *
 * [EOF]
 */