prony

Prony method for filter design

Syntax

[b,a] = prony(h,bord,aord)

Description

[b,a] = prony(h,bord,aord) returns the numerator and denominator coefficients for a causal rational transfer function with impulse response h, numerator order bord, and denominator order aord.

Examples

Filter Responses Using the Prony Method

Fit a 4th-order IIR model to the impulse response of a lowpass filter. Plot the original and Prony-designed impulse responses.

d = designfilt('lowpassiir','NumeratorOrder',4,'DenominatorOrder',4, ...
    'HalfPowerFrequency',0.2,'DesignMethod','butter');

h = filter(d,[1 zeros(1,31)]);
bord = 4;
aord = 4;
[b,a] = prony(h,bord,aord);

subplot(2,1,1) 
stem(impz(b,a,length(h)))
title 'Impulse Response with Prony Design'

subplot(2,1,2)
stem(h)
title 'Input Impulse Response'

Fit a 10th-order FIR model to the impulse response of a highpass filter. Plot the original and Prony-designed frequency responses. The responses match to high precision.

d = designfilt('highpassfir','FilterOrder',10,'CutoffFrequency',0.8);

h = filter(d,[1 zeros(1,31)]);
bord = 10;
aord = 0;
[b,a] = prony(h,bord,aord);

[hp,f] = freqz(b,a);
[hd,fd] = freqz(d);

figure
plot(f/pi,mag2db(abs(hp)),fd/pi,mag2db(abs(hd)))
legend('Prony','Original')

Input Arguments

h — Impulse response

vector

Impulse response, specified as a vector.

Example: impz(fir1(20,0.5)) specifies the impulse response of a 20th-order FIR filter with normalized cutoff frequency π/2 rad/sample.

Data Types: single | double
Complex Number Support: Yes

bord, aord — Numerator and denominator orders

positive integer scalars

Numerator and denominator orders, specified as positive integer scalars. If the length of h is less than max(bord,aord), the function pads the impulse response with zeros.

• If you want an all-pole transfer function, specify bord as 0.

• If you want an all-zero transfer function, specify aord as 0.

Data Types: single | double

Output Arguments

b, a — Transfer function coefficients

vectors

Transfer function coefficients, returned as vectors. b has length bord + 1 and a has length aord + 1.

More About

Transfer Function

The transfer function is the Z-transform of the impulse response h[n]:

$$H(z)={\displaystyle \sum _{n=-\infty }^{\infty }h\left(n\right)}{z}^{-n}.$$

A rational transfer function is a ratio of polynomials in z^–1. This equation describes a causal rational transfer function of numerator order q and denominator order p:

$$H(z)=\frac{B\left(z\right)}{A\left(z\right)}=\frac{{\displaystyle \sum _{k=0}^{q}b\left(k\right)z^{-k}}}{1+{\displaystyle \sum _{l=1}^{p}a\left(l\right)z^{-l}}},$$

where a[0] = 1.

References

[1] Parks, Thomas W., and C. Sidney Burrus. Digital Filter Design. New York, NY, USA: Wiley-Interscience, 1987.

Extended Capabilities

C/C++ Code Generation
Generate C and C++ code using MATLAB® Coder™.

Version History

Introduced before R2006a