Finite Impulse Response filters are those filters present a non-zero finite length response when excited with a very brief (ideally an infinite peak) input signal. A linear causal FIR filter can be described by the following difference equation

This operation describe a nonrecursive system, i.e. a system that only depends on current and past samples of the input data stream x[n]

Creating a FIR filter in the LTPDA

The LTPDA Toolbox allows the implementation of FIR filters by means of the mfir class.

Creating from a plist

The following example creates an order 64 highpass filter with high frequency gain 2. Filter is designed for 1 Hz sampled data and has a cut-off frequency of 0.2 Hz.


        >> pl = plist('type', 'highpass', ...
                      'order', 64,         ...
                      'gain',  2.0,       ...
                      'fs',    1,        ...
                      'fc',    0.2);
        >> f = mfir(pl)
  

Creating from a difference equation

The filter can be defined in terms of two vectors specifying the coefficients of the filter and the sampling frequency. The following example creates a FIR filter with sampling frequency 1 Hz and the following recursive equation:



        >> b = [-0.8 10];
        >> fs = 1;
        >> f = mfir(b,fs)

Creating from an Analysis Object

A FIR filter can be generated based on the magnitude of the input Analysis Object or fsdata object. In the following example a fsdata object is first generated and then passed to the mfir constructor to obtain the equivalent FIR filter.


        >> fs  = 10;                      $ sampling frequency
        >> f = linspace(0, fs/2, 1000);
        >> y = 1./(1+(0.1*2*pi*f).^2);    $ an arbitrary function
        >> fsd = fsdata(f,y,fs);          $ build the fsdata object
        >> f = mfir(ao(fsd));


Available methods for this option are: 'frequency-sampling' (uses fir2), 'least-squares' (uses firls) and 'Parks-McClellan' (uses firpm)

Importing an existing model

The mfir constructor also accepts as an input existing models in different formats:

  • LISO files:

            >> f = mfir('foo_fir.fil')
    
  • XML files:

            >> f = mfir('foo_fir.xml')
    
  • MAT files:

            >> f = mfir('foo_fir.mat')
    
  • From repository:

            >> f = mfir(plist('hostname', 'localhost', 'database', 'ltpda', 'ID', []))