DSP System Toolbox

FIR Halfband Filter Design

This example shows how to design FIR halfband filters. Halfband filters are widely used in multirate signal processing applications when interpolating/decimating by a factor of two. Halfband filters are implemented efficiently in polyphase form, because approximately half of its coefficients are equal to zero.

Halfband filters have two important characteristics, the passband and stopband ripples must be the same, and the passband-edge and stopband-edge frequencies are equidistant from the halfband frequency pi/2.

In this example we will design FIR halfband filters of a given order but also minimum-order and minimum-phase filters. Minimum-phase designs are particularly useful for perfect reconstruction 2 channel FIR filter banks.

Windowed-Impulse-Response Designs

Windowed-impulse-response designs are obtained by simply specifying an even order and a window of appropriate (odd) length. In this case, the passband-edge frequency cannot be specified, rather it is a result of both the order and the window specified.

N = 90;                            % Filter order must be even
d = fdesign.halfband('Type','Lowpass','N',N);
f = design(d,'window','Window','hamming','SystemObject',true);

One can verify that the passband/stopband ripples are the same

hfvt = fvtool(f, 'MagnitudeDisplay', 'Zero-phase','Color', [1 1 1]);

Impulse response showing every other coefficient equal to zero (except for the middle one)

hfvt.Analysis = 'Impulse';

Controlling the Transition Width

In addition to the even order, one can specify the transition width of the filter.

TW = 0.05; % Transition width
d = fdesign.halfband('Type','Lowpass','N,TW',N,TW);
f1 = design(d,'equiripple','SystemObject',true); % Equiripple design
f2 = design(d,'firls','SystemObject',true);      % Least-squares design
f3 = design(d,'kaiserwin','SystemObject',true);  % Kaiser-window design

One can verify that the passband/stopband ripples are the same

hfvt = fvtool(f1,f2,f3,'Color','white', ...
    'MagnitudeDisplay', 'Magnitude');
legend(hfvt, 'Equiripple design', 'Least-squares design', ...
    'Kaiser-window design')

Impulse response still zero for every other coefficient

hfvt.Analysis = 'Impulse';

Controlling the Stopband Attenuation

Alternatively, one can specify the order and the stopband attenuation. For example,

Ast  = 60; % Minimum stopband attenuation
d  = fdesign.halfband('Type','Lowpass','N,Ast',N,Ast);
f1 = design(d,'equiripple','SystemObject',true); % Equiripple design
f2 = design(d,'kaiserwin','SystemObject',true);  % Kaiser-window design
hfvt = fvtool(f1,f2,'Color','white');
legend(hfvt, 'Equiripple design', 'Kaiser-window design')

Passband-/stopband-edge frequencies are a result of design

hfvt.MagnitudeDisplay = 'Zero-phase';

Impulse response still zero for every other coefficient

hfvt.Analysis = 'Impulse';

Minimum-Order Designs

Another way of designing equiripple halfband filters is by specifying both the transition width and stopband ripple. The order of the filter cannot be specified in this case (because there are only two degrees of freedom), instead a filter of minimum-order that meets the design specifications is obtained.

TW = 0.02; % Transition width
Ast = 40;  % 40 dB of attenuation in the stopband
d  = fdesign.halfband('Type','Lowpass','TW,Ast',TW,Ast);
f1 = design(d,'equiripple','SystemObject',true); % Equiripple design
f2 = design(d,'kaiserwin','SystemObject',true);  % Kaiser-window design
hfvt = fvtool(f1,f2,'Color','white');
legend(hfvt, 'Equiripple Design', 'Kaiser-window design')

Equiripple Designs with Increasing Stopband Attenuation

If it is still desirable to have an increasing attenuation in the stopband, we can use design options for equiripple designs to achieve this.

f1 = design(d,'equiripple','StopbandShape','1/f','StopbandDecay',4,...
    'SystemObject',true);
f2 = design(d,'equiripple','StopbandShape','linear', ...
    'StopbandDecay',53.333,'SystemObject',true);
hfvt = fvtool(f1,f2,'Color','white');
legend(hfvt,'Stopband decaying as (1/f)^4','Stopband decaying linearly')

Highpass Halfband Filters

A highpass halfband filter can be obtained from a lowpass halfband filter by changing the sign of every second coefficient. Alternatively, one can directly design a highpass halfband by setting the 'Type' property to 'Highpass'.

d  = fdesign.halfband('Type','Highpass','TW,Ast',TW,Ast);
f1 = design(d,'equiripple','StopbandShape','linear', ...
    'StopbandDecay',53.333,'SystemObject',true);
hfvt = fvtool(f1,f2,'Color','white');
legend(hfvt, 'Highpass halfband filter', 'Lowpass halfband filter')

Minimum-Phase Equiripple Designs

Minimum-phase designs are useful for 2 channel filter banks with perfect reconstruction composed solely of FIR filters. For this case, the minimum-phase filter is not in itself a halfband filter. Rather, it is a spectral factor of a halfband filter.

N = 51;                       % Minimum-phase order must be odd
TW = 0.04;                    % Transition width
d = fdesign.halfband('Type','Lowpass','N,TW',N,TW);
f = design(d,'equiripple','MinPhase',true,'SystemObject',true);

Notice that the passband and stopband ripples are different in this case

fvtool(f,'Color','white', 'MagnitudeDisplay', 'Zero-phase');

All zeros of the minimum-phase filter are inside the unit circle

fvtool(f,'Color','white', 'Analysis', 'polezero');

The minimum-phase filter is not in itself a halfband filter since every other coefficient is not equal to zero

fvtool(f,'Color','white', 'Analysis', 'impulse');

The corresponding halfband filter can be obtained by convolving the impulse response with a reversed version of it (the maximum-phase spectral factor).

fmin = f.Numerator;           % Minimum-phase spectral factor
fmax = fliplr(fmin);          % Maximum-phase spectral factor
hhalf = conv(fmin,fmax);      % Halfband filter with non-negative zerophase

Impulse response is zero for every other coefficient

hfvt = fvtool(hhalf, 'Analysis', 'impulse','Color','white');

Halfband filter with non-negative zerophase

hfvt.Analysis = 'Magnitude';
hfvt.MagnitudeDisplay = 'Zero-phase';

See also Perfect Reconstruction Two-channel Filter Bank.