2017 2nd International Conference on Telecommunication and Networks (TEL-NET 2017)

IIR Based Digital Filter Design and Performance

Analysis
Shapna Rani Sutradhar, Nazmus Sayadat,
Ashfiqur Rahman, Sirajum Munira, A.K.M. Fazlul Haque and Syed Nazmus Sakib
Department of Electronics and Telecommunication Engineering
Daffodil International University
Dhaka,Bangladesh
Email: [email protected]

Abstract—Digital filters are mandatory for digital signal Some selected audio signals have been used to observe
processing. This paper presents digital filter dispelling the different digital filters response and also plotted and played
unwanted signals or noise from the required signal and enhances
the better performances of the signal. The extracted features of the the output of the filters. The aims of this research are to find
digital filter have been analyzed to acquire the better output of the out the better performance of digital filters. To achieve this
signal by using IIR Butterworth filter. It provides different designed goal, it is needed to optimize the different parameters of
parameters of IIR filter to achieve the desired result. MATLAB digital filters. Parameters like magnitude response, phase
FDAtool is considered to find out the different responses of a delay, phase response, group delay, impulse response,
digital filter. About eight parameters like- Phase response, pole/zero plots and step response have been used to optimize
Magnitude response, Magnitude & Phase response, Step the result and determine the better output. To acquire the
response, Group delay, Pole/Zero Plot, Phase delay, Impulse better output, IIR Butterworth filter has been used. The
response are used to analyze the filter responses. Some selected cooperation between attenuation and phase response is called
audio signals are used for observing the empirical response of Butterworth filter. Once in a while this is called a maximally
high pass, low pass, band stop filter and band pass filter. A flat filter because; it has no flow in the stop band or the pass
special tool is developed for this observation purpose. band. The Butterworth filter acquires its flatness at the
expenditure of a comparatively wide evolution section from
Keywords— Digital Filters, Signal Processing, IIR Butterworth
Filter, Responses, Impulse, Delay. pass band to stop band. The estimations of the mechanism of
the Butterworth filter are more realistic and less vital than
I. INTRODUCTION several filter types. Butterworth filters have practically better
amplitude and transient manners [2]. Another main concern of
For any kind of communication system, signal processing this research is to use this designed filters in real and practical
is the leading part. To avoid unwanted signals, signal execution. All the simulation and results for the analyzing of
processing is mandatory. Analog and digital signal processing different parameters have been analyzed and verified using
can be used to process information. More flexibility and FDA tool in MATLAB. A tool is developed for filtering audio
better control of accuracy are provided by digital signal signals in practical.
processing compared to analog signal processing. Filters are
applied to process the signal. To process the raw signals, II. LITERATURE REVIEW
digital filters are very essential. The system which performs
mathematical operations on a sampled, discrete-time signal to Many works have been done on digital filter for
decrease or increase certain feature of that signal is called signal processing. A.K.M. Fazlul Haque worked on enviable
digital signal [1]. Different statistical methods may be parameters extraction and enhanced performance of digital
introduced to observe the performance of a digital filter. filters using FDATool [3]. Mohammad Saiful Islam, Shaila
Several observation methods may also be employed in Shanjada and Mithun Banik worked on improved feature
designs, and often form the origin of a filter detail. Basically, extraction of digital filters using FDATool [4]. Yaduvir Sing,
one symbolizes filters by computing how they will react to a Sweta Tripathi and Manoj Pandey performed an analysis of
simple input such as an impulse. Another can then encompass digital IIR filters with LabVIEW [5]. ZHANG Chengliang and
this data to subtract the filter's response to more composite WANG Aihong worked on IIR digital filter design research
signals. In this research, different forms of filters like high- and simulation on MATLAB [6]. Ethan Elenberg et.al.
pass filter, low-pass filter, band-stop filter and band-pass filter worked on digital filter design for audio processing [7]. Er.
have been used to analyze the performance of digital filter. Daljit Singh Bajwa, Er.Karamjeet Singh and Navpreet Kaur
Chahal worked on survey paper of digital IIR filter design [8].
 2017 2nd International Conference on Telecommunication and Networks (TEL-NET 2017)

Ekta Yadav and Rupali worked on digital IIR filter design [9]. button groups are selected from the toolbar and renamed them
In the above research papers, all of them worked on analysis to “Input” and “Filters”. Then some push buttons are selected
and design of IIR digital filter. Different types of design for those two groups. “Select Audio” and “Play Audio” button
methods have been used for those analysis purposes like- for the “Input” button group and “High Pass”, “Low Pass”,
window function method, impulse invariance method and “Band Stop”, “Band pass” button for the “Filter” button group.
some of them used LabVIEW software for their research. In Four axes, axes1, axes2, axes 3, axes 4 are selected to plot the
this paper, diverse forms of filters like high-pass filter, low- input and output in time domain and frequency domain. Axis1
pass filter, band-stop filter and band-pass filter have been used and axis2 are used for the input and axis3 and axis 4 for
to examine the performance of digital filter. IIR Butterworth plotting filter output in time and frequency domain.
design method is used for this design purpose. A tool is also
developed for filtering audio signal. When this GUI has been saved, it saves a “. Fig” and
generates an “.m” file. The “.m” is the editable MATLAB file
III. METHODOLOGY which can be modified to customize the app. In this MATLAB
file, all the commands have been written to run the GUI. The
In this paper, the actions of different digital filters work of every push buttons and the figures have also been
and the responses have been studied and analyzed based on defined to plot in the axes number.
different parameters. Also, the filter output of a selected audio
signal has been examined. 2) Working Process:

MATLAB FDA (Filter design and analysis) tool has been used The block diagram shown in the Fig.3 shows a
for analysis part and MATLAB GUIDE (Graphical user typical working process of the designed GUI. First of all, an
interface development environment) tool for the experiment of audio has been selected from the computer. Then the audio has
filtering audio signal. been played with MATLAB and plotted the audio in time
domain and frequency domain. One of the four designed
The working procedure of this research has been illustrated in digital filter is then selected to pass the audio. The output of
the following Fig.2 and Fig.3. the filter is also played and the filtered audio is also plotted in
time and frequency domain.
A. Why IIR filter instead of FIR filter:
3) Step by step working of the GUI:
Mainly, there are two categories of digital filters. Infinite
impulse response (IIR) and Finite impulse response (FIR). Following steps are involved while running the GUI in
Filters which usually have an impulse response containing MATLAB.
only a finite number of values are typically called finite
impulse response (FIR) filters [10]. FIR filters are involved in a) Selecting audio:
filtering problems where there is a condition for linear phase
personality within the passband of the filter. The design of FIR Then an audio has been selected from the computer. “Select
filters is centered on a direct approximation of the particular Audio” button opens a popup window to browse and select a
magnitude response, with the frequently added obligation that “.wav” file. Then our desired audio filter has been selected
the phase response be linear [11]. On the other hand, the filters from a predefined directory. Fig.5 shows the pop-up window
which have impulse response of infinite dimension are called to select the audio.
infinite impulse response (IIR). In this paper only IIR filters
have been applied for analyzing different parameters. Because,
IIR filters have more efficient filtering characteristics
compared to FIR filters. IIR filters are normally used in the
applications that don’t need linear phase response. That means
IIR filters usually have nonlinear phase response [12]. IIR
filters require less memory as it needs fewer coefficients to
perform similar filtering action and it is quicker than FIR
filters. For this reason, IIR filters have been used.

B. Designing digital filter with MATLAB:

1) Creating filters in MATLAB GUI with GUIDE:

In order to create a GUI with GUIDE tool for

Fig. 1. Editing GUI in MATLAB GUIDE tool
selecting an audio signal, a digital filtering has been applied
from high pass, low pass, band stop or band pass filters. Two
 2017 2nd International Conference on Telecommunication and Networks (TEL-NET 2017)

b) Play and plot the audio:

Filter
Selection Pressing the “Play Audio” button plays back the
selected audio and plot the audio in frequency domain and
time domain. For simplification, .wav file is only used. To plot
Parameters
Selection the audio, a time vector and a frequency vector have been
created from the selected audio. The selected length of the fft
was 2^16, which is also the length of the frequency of the
Order audio. Fig.6 shows the time and frequency domain
selection
representation of that selected audio.

Extraction of IV. SIMULATION & ANALYSIS:

the output
A. Input data table:

Fig. 2. Working procedure for finding the better output of filters for some The input values that have been used to analyze the
selected parameters using different orders. performance of different filters is shown in TABLE I.
Different orders like- 400, 600, 800, 1000 and 1500 are
Select Audio randomly selected to find out the better output for different
filters.

Play & Plot

Select a Filter

Play & Plot the

output

Fig. 3. Working block diagram of GUI for filtering audio signal.

Fig. 6. Play & plot the selected audio

TABLE I: Input Data

Filter Frequency Specifications(Hz) Magnitude

Type Specification
Fs Fc

Low Pass 48000 10800 3 dB

High Pass 48000 10800 3 dB

Fig. 4. Running the GUI.
Band Pass 48000 4800 13200 3 dB

(Fc1) (Fc2)

Band Stop 48000 4800 13200 3 dB

(Fc1) (Fc2)

B. Performance analysis of different filters:

Fig. 5. Selecting audio file from computer. 1) Low pass filter:

 2017 2nd International Conference on Telecommunication and Networks (TEL-NET 2017)

a) Magnitude Response: high pass filters. Butterworth takes various orders such as,
400, 600, 800, 1000, 1500 and it gives different result. In
From the magnitude response, IIR Butterworth can FDAtool, IIR Butterworth high pass filters have been used and
easily find out the basic properties of filters. The Fig.7 shows found that, 1500 magnitude response’s quality is better than
that, the filter passes only low frequencies while reject higher others.
frequency. IIR Butterworth used different orders like 400, 600,
800, 1000 and 1500 to find out which gives the better output. b) Phase Response:
Then it has been found out that, order 1500 gives more sharp
output. It passes lowest frequency below cutoff frequency. When Phase response has been discussed, it can be
said that, it is a part of a signal processing. If IIR Butterworth
b) Phase Response: filter wants efficient phase response in high pass filter, then it
measures some orders like- 400, 600, 800, 1000 and 1500
Phase response is one of the most important shows in the Fig.10 which contains lower properties of high
parameters of digital filters. It shows the relationships between pass filter and the response of order 1500 is stronger.
phases of the input and output signal. The phase response of
low pass filters is incorporated with some applications such as 3) Band Pass filters:
time delay simulation, cascaded filter stages and particularly
for process-control loops. IIR Butterworth low pass filter used a) Magnitude Response:
different orders such as 400, 600, 800, 1000 and 1500 to
evaluate the output which is shown in Fig.8. Magnitude response in IIR Butterworth band pass
filter is used to find out the basic characteristics of digital
filter. The best magnitude response in band pass filter will be
found out. For this, different filter orders like- 400, 600, 800,
1000 and 1500 have been used which showed in the Fig.11
and it is definitely high filter order delivered the better output.
So order 1500 is sharper and magnitude response takes the
best output.

Fig. 7. Magnitude Response of low pass filter for different orders

Fig. 9. Magnitude Response of high pass filter for different orders

Fig. 8. Phase Response of low pass filter for different orders

2) High pass filter:

a) Magnitude Response:

This is based on high pass filter and high pass filter

can pass high frequency signal. Now IIR Butterworth is
Fig. 10. Phase Response of high pass filter for different orders
discussing about high pass filter in magnitude response and
trying to know in what situation high pass filter can gives
better result. Here the Fig.9 shows the magnitude response of
 2017 2nd International Conference on Telecommunication and Networks (TEL-NET 2017)

Fig. 11. Magnitude Response of band pass filter for different orders Fig. 13. Magnitude Response of band stop filter for different orders

b) Phase response:

In IIR Butterworth band pass filter, the connection

between the phase of a sinusoidal input and the output signal
passing through any expedient is known as phase response.
Here, in the Fig.12, IIR Butterworth uses three filter orders to
get the best response where 1500 order gives the beneficial
result.

4) Band stop filter:

a) Magnitude Response:

The magnitude response of a band stop filter is a

filter that passes most frequency response is unchanged. A
general ideal band stop filter can display both positive and
negative angular frequency. In IIR Butterworth band stop filter Fig 14. Phase Response of band stop filter for different orders
in the Fig.13 takes different filter order, order 400, order 600,
order 800, order 1000 where higher order 1000 gets more All Parameter’s values have been taken accordingly.
priority and gives accurate output.
C. Filtering audio signal:
b) Phase Response:
1) Low pass filtering:
In IIR Butterworth band stop filter, phase response is
the stage of the output with the input as reference. In Fig.14, When "Low Pass" button has been selected, it
IIR Butterworth method gets different filter order for taking executes a callback function in background, which generates a
favorable output. Higher filter order deployed much better low pass filter output and plays back the filtered audio. To
output. design a low pass filter, some parameters like, sampling
frequency 44100 Hz, cutoff frequency 10000 Hz have been
used. After filtering the audio signal, the change in the time
and frequency domain could be visualized. Fig.15 shows the
output of the low pass filter. The change in the audio sound
quality is also noticeable. It has been found out that, for a high
frequency audio signal with noise, applying the low pass filter
improves the sound quality by cancelling high frequency
noise.
2) High pass filtering:
For filtering audio signal with high pass filter, a
callback function is considered with high pass filter
Fig. 12. Phase Response of band pass filter for different orders coefficient, which can be executed by pressing the "High
Pass" button. For high pass filter, the sampling frequency and
cutoff frequencies are the same as low pass filter 44100 Hz
 2017 2nd International Conference on Telecommunication and Networks (TEL-NET 2017)

and 10000 Hz. In this research finding, high pass filter was not V. CONCLUSION:
a better choice for audio signal. Because, it cancels out all low
frequencies while allows passing high frequencies with noise, In this paper, different design parameters have been
though it could be useful for some specific applications. considered for IIR Butterworth filter to analyze the filter
Fig.16 shows the high pass filter output of the audio signal. performance. Four categories of filters like high pass, low
The change in the amplitude level can be seen in this figure. pass, band stop and band pass filters have been used for IIR
Butterworth filter. To analyze the performance, MATLAB
3) Band pass filtering: FDA tool has been used. To simulate the result in FDA tool,
For band pass filtering the sampling frequency is different orders like order 400, 600, order 800, order 1000 and
same as for low and high pass filtering. The pass band order 1500 have been considered. From the performance of
frequency is set from 1000 Hz to 2000 Hz. different order, it has been analyzed that the better output can
be found from higher order. That means, better result has been
found from order 1500 comparing order 1000, 800, 600 and
order 400. The selected audio signal which passed through
Fig.15. Audio signal after passing high pass filter. different filters can give better output in low pass filter than
the other filters. MATLAB GUIDE tool is used to analyze the
performance of digital filters in terms of audio signal.

REFERENCES
[1] James D. Broesch, Dag Stranneby and William Walker, “Digital Signal
Processing: Inastant access”, 2009, pp.3.
[2] Hank Zumbahlen, “Basic linear design”, 2007, pp. 8.21
[3] A.K.M. Fazlul Haque, “Enviable Parameters Extraction and Enhanced
Performance of Digital Filters using FDA Tool”, Journal of Bangladesh
Electronic Society 9 (1), 2009.
[4] Mohammad Saiful Islam, Shaila Shanjada and Mithun Banik, “Improved
Feature Extraction of Digital Filters using FDA Tool”, Thesis
paper,Department of Electronics & Telecommunication Engineering,
Daffodil International University, 2010.
Fig. 14. Audio signal after passing band pass filter. [5] Yaduvir Sing, Sweta Tripathi and Manoj Pandey, “Analysis of Digital
IIR Filters with LabVIEW”, International Journal of Computer
Applications, vol 10, 2010.
[6] ZHANG Chengliang and WANG Aihong, “IIR digital filter design
research and simulation on MATLAB”, 4th International Conference on
Signal Processing Systems (ICSPS), 2012.
[7] Ethan Elenberg, Stephanie Ng, Anthony Hsu, Alaap Parikh, Michelle
Yu, Marc L’Heureux and E.J. Thiele, “Digital Filter Design for Audio
Processing”, American journal of physics, 1965.
[8] Er. Daljit Singh Bajwa, Er.Karamjeet Singh and Navpreet Kaur Chahal,
“Survey Paper of Digital IIR Filter Design”, International Journal of
Advance Research in Computer Science and Software Engineering
(IJARCSSE), 2014.
[9] Ekta Yadav and Rupali, “Digital IIR Filter Design”, International
Journal of Innovative Research in Technology (IJIRT), vol 1,2015
Fig. 15. Audio signal after passing band stop filter.
[10] Les Thede, “Practical Analog and Digital Filter Design”, 2004, pp. 141.
[11] Sanjit K Mitra, “Digital Signal Processing: A Computer Based
4) Band stop filtering: Approach”, 3e, pp.523.
[12] J. S Chotode, “Digital signals processing”, 3e, 2008, pp.4.1.
A call back function “Band Stop” button is also
created to perform band stop filtering. The stop band
frequency is set from 2000 Hz fast cutoff to 4000 Hz second
cutoff for band stop filtering. Selecting the “Band Stop”
button the output is generated as Fig.18 and an audio is also
played of that selected audio which is the band stop filer
output.