International Conference on Communication and Signal Processing, April 6-8, 2016, India

Design of FIR Filter for Burst mode

Demodulator of Satellite Receiver

Jayshri J. Patel, Dr. Kiran R. Parmar and Hardik N. Mewada

 correspondence, radar, picture, programmed control, voice,

Abstract—This paper proposes a satellite receiver filter design aerospace, medical and so on [2]. Contrasted with the simple
using FIR digital filtering technique. We present various design analog filters, advanced digital filters have higher accuracy,
methods like windowing, least squares and equiripple for satellite better solidness and unwavering quality.
burst demodulator application and compare their performance. On the basis of unit impulse response function, digital
Various designs of FIR filter are compared from the view point filters can be classified into two parts namely an IIR (Infinite
of hardware complexity, frequency response characteristics and
Impulse Response) digital filter and FIR (Finite Impulse
implementation strategies. The filter is designed for band pass of
the frequency range of 100 MHz to 500 MHz suitable for the Response) digital filter [3]. Comparing to the Infinite Impulse
entire bandwidth of satellite transponder. The burst mode Response, the Finite Impulse Response is efficient to fulfill
detector requires narrow passband to increase SNR for preamble the needs of the amplitude and phase characteristics to avoid
portion. When acquisition phase is complete, the bandpass is the drift and noise. IIR filter has an advantage of
increased to full bandwidth of the signal. programmability, but the operating frequency of the system is
bound by its feedback loops. FIR computerized digital filters
Index Terms— FIR filter; Kaiser Window method; Equiripple has entirely direct stage and adequacy recurrence trademark
method; Least square method; FDA tool in MATLAB and high strength, and so on. In some high speed wideband
communication system, e.g. high-speed broadband satellite
receiver, the transmission data rate is getting higher and
I. INTRODUCTION higher, so the FIR filter operation speed requirements are also

T HIS paper briefly introduces the general facet of

designing FIR filters for burst mode demodulator of a
satellite receiver. The Time Division Multiple Access
increasing, typically, greater than 1 GHz [4].
Advanced digital filters can be accomplished by equipment
circuits, as well as by PC software programming. For
(TDMA) is a highly efficient method for satellite access. The simulation and real time implementation, MATLAB tool is the
receiver needs to have precise filtering mechanism for most widely popular which also allows the conceptual design
wideband signal. Designing of an FIR digital filter using FDA and modeling [5]. The design and simulation analysis of
tool in MATLAB for the design specifications selected for digital filter is efficiently achieved by using powerful
burst mode applications is described in this paper. The signal computing capabilities of MATLAB. As one of the MATLAB
that is received at the input of the burst demodulator is in the signal processing boxes, Filter Design and Analysis tool (FDA
digital form. This signal contains noise in the actual data. So, tool) has powerful features to design digital filters [5].
digital filtering is done on this baseband signal. Its purpose is Designing of digital filters can be achieved by calculation of
to maximize the signal to noise ratio at the output at some the filter coefficients on the basis of filter order, sampling
instant [1]. frequency, passband frequency, stopband frequency etc. As
Digital filters are commonly expanded on large satellites to order of the filter increases, power consumption and the
reduce noise or disturbances. Broadly, filters can also be computational complexity will be increased. MATLAB FDA
explained as a system which selects an applicable signal. In tool gives simple and easy calculation of digital filter
the development computerized data age, the advanced digital coefficients [6].
filters has pulled in individuals' profound consideration to an The association of this paper is as per the following. Section
ever increasing extent, has been widely connected to I describes the introduction. The section II describes basic
principles and theoretical formulation of FIR filter design
technology. The section III describes the design of FIR filter
for given specifications of satellite receiver using three
Jayshri J. patel pursuing her post-graduation with the Department of
Electronics and Communication at Parul Institute of Engineering &
different techniques viz. Kaiser Windowing method, Least
Technology(PIET),Limda,Gujarat,India (e-mail: [email protected]). square method and Equiripple method. All these methods are
compared for frequency response characteristics, hardware
Dr. Kiran R. Parmar is technical consultant at eiTRA (e-infochips Training complexity and structures for implementation. The last section
and Research Academy),Gujarat, India. (e-mail: [email protected]).
discusses about order, passband and stopband attenuations and
Hardik N. Mewada is assistant professor at Parul Institute of Engineering multipliers/adders required for filter structures followed by
and Technology, Gujarat, India. (e-mail: [email protected]). conclusion.

978-1-5090-0396-9/16/$31.00 ©2016 IEEE

0686
 II. DESIGN PRINCIPLE the filter is equivalent to 1 for all passband frequencies and
equivalent to 0 for all stopband frequencies [10]. To generate a
A. Basics of FIR filter
Finite Impulse Response (FIR) digital filter, the time domain
The most basic characteristic of FIR filter is that it has filter coefficients must be blocked in number of multiplying
only zeros and no poles. Hence, FIR filters are known as all- by a window function of a finite width. In this method,
zero filters. It can also be known as feed-forward or non- consider desired frequency response specification Hd(ω) and
recursive or transversal filters. If x(n) is an input of M length from that the corresponding unit sample response hd(n) can be
FIR filter and y(n) is its output, then it can be described by determined. Absolutely, hd(n) is related to Hd(ω) by the
following difference equation [7]: Fourier transform relation [10]:
M 1 f

¦ b k x( n  k ) ¦h
 jZ n
y ( n) (1) H d (Z ) n 0
d
(n)e (4)
k 0

Where, {bk} shows the filter coefficients set. we can also Where,
denote the output of the filter by doing convolution of the unit S
1 jZ n
(5)
sample response h(n) of the system and the input signal x(n). h d
( n)
2S ³S H d
(Z )e
Thus, we have, 

M 1 Thus, the unit sample response hd(n) can be determined by

y(n)= ¦ h  k  x(n  k ) (2) evaluating the integral in above equation (5).
k 0
Different window functions that are generally used can be
Causality and finite-duration characteristics of the filter are listed as Triangular windows, Hann window, Hamming
dependent on the lower and upper limits of the convolution window, Kaiser window and Blackman window and so on. In
sum. Here M represents the tap coefficients of the filter. h(k) this paper, we shall discuss about only Kaiser window. The
acts as the k level tap coefficients of the FIR filter [8]. Clearly, Kaiser window permits separate control of width of the
equation (1) and (2) are similar in form and hence it concluded primary lobe and constriction of the side lobe [11]. A simple
that bk= h(k), Where, k = 0,1,…..,M-1. approximation of the window using zero-th order modified
Bessel function of the first kind is as follows,
The digital filter can be described by its system function:
M 1 k
H  z ¦ h(k ) z
k 0
(3) ( )= [ 1− ], 0 ≤ ≤ (6)
0, ℎ
Which is viewed as a polynomial of degree M-1 in the
Here, I0 is the zero-th order modified Bessel function,
variable z-1. The foundations of this polynomial constitute the
usually α=M/2. α is stopband attenuation in decibels. The
zeros of the filter.
‘shape’ of filter can be modified by β . Kaiser analytically
B. Digital Filter coefficient determination method found formula going from δ, ωs, ωp to M and β [11].
The function of designing of an digital filter depends on the
calculation of the filter coefficients. The iterative method and 'Z Z  Z s p

the window method are the most popular methods for

determining filter coefficients. Constant equiripple
A 20log G
10
characteristic in the passband and the stopband can be
designed by an iterative method [9]. Low complexity and easy 0.1102( − 8.7), > 50
designing are noticeable advantages of window method over 0.5842( − 21) . + 0.07886( − 21), 21 ≤ ≤ 50
other traditional methods. The way that express mathematical 0.0, < 21
statements are regularly accessible for figuring the window And
coefficients has made this strategy prevalent. Window method
is basically useful for the design of model filters like lowpass, A8
highpass, bandpass, etc [7]. This makes its need in speech and M
image processing applications much simpler. 2.285'Z
This makes it extremely easy to do filter design instantly.
III. DESIGN METHODS OF THE DIGITAL FIR FILTER Actual filter response can be obtained by multiplying an ideal
Different design techniques for the FIR filters are the impulse response hd(n) with window w[n] shown as [12]:
frequency sampling technique, Windowing technique ,Optimal
filter design techniques like Equiripple , Least- square error h(n) w> n@ h d  n  (7)
etc [7].
M does not change ripple error, this is achieved by sidelobe
A. Windowing method for FIR filter design amplitude.
The essential thought behind the window technique for
filter configuration is that the perfect recurrence reaction of

0687
B. Least mean squares error method The filter design is a bandpass design and it occupies the
The least mean squares error method utilizes the use of a bandwidth of 400 MHz. The digital filter works with digital
minimum mean square error rule given by [3]: data and we convert analog filter specifications into equivalent
digital specifications using Nyquist sampling rate of 1.2 GHz.
¦^W Z  ª¬ H Z   D Z º¼`
k 2
H i i i
(8)
i 1

Where, H (ωi) is the amplitude response of the designed

filter, D(ωi) is the desired amplitude response, W(ωi) is the
weighting function. The amplitude response for a linear phase
FIR filter is expressed as [13]:
L
H (Z i ) Q(Z )¦ a(k ) cos(Zk ) (9)
k 0

This is utilized to get a reliable transfer function.

C. Equiripple method for FIR filter design Fig. 1. Burst mode demodulator for satellite receiver
As the name recommends, equiripple channel has measures
the same in the passband and stopband, which implies the
signal distortion that happens at the edge of the passband We use MATLAB FDA Tool to design different types of
because of nearness of huge swell is stayed away from in the digital FIR filters. FDA tool is a digital filter design and
equiripple design. The equiripple strategy approaches filter analysis tool in a MATLAB signal processing toolbox. FDA
outline as a streamlining issue in which the coefficient apparatus can plan every customary filter, including different
qualities are acclimated to create a brilliant filter with swells configuration techniques for FIR and IIR and it has
that are of equivalent height [7]. A least square approach to straightforward and ordinary operation.
minimize error is not entirely satisfactory. So, superior ways A. Filter Specifications
to deal with minimize the greatest error at the band is as
The burst mode demodulator is equipped for synchronous
underneath:
demodulation of N channels in a 20 MHz band, where N relies
H E (Z ) (10) on upon burst type and symbol rate and is hard constrained to
max Z 200 channels. The overall bandpass filter bandwidth is kept as
This technique is excellent in the view of minimising the 400 MHz. The other specifications of the bandpass filters are
maximum magnitude of the ripple in the all bands of the as follows:
interest. Order of the filter is fixed,
Filter order: 40
min max E Z  (11) Passband attenuation (αp): 1 dB
Hd Z
Stopband attenuation (αs): 60 dB
It can be demonstrated that this prompts an equirippple
filter, a filter which sufficiency reaction sways uniformaly Passband frequency: 100 MHz
between the resilience bound of every band [7]. Stopband frequency: 500 MHz
IV. DESIGNING OF FIR FILTER BY MATLAB Sampling rate: 1.2 GHz
The burst mode demodulator for TDMA satellite access is Density factor for the equiripple method: 20
shown in fig.1. The input to the filter is from the ADC
connected at the input of the receiver. The overall Weighting factor for the least square method: 1
specifications for the receiver are as follows: β for the Kaiser window method: 0.5
x Multi frequency-multi channel burst mode
demodulator B. Design of FIR filter for satellite receiver
x Applicable to star and mesh networks for satellite
application The filters are designed by three approaches viz. Kaiser
Window based design, Least square method and Equiripple
x DVB-RCS standard is used
method. The filter responses are reproduced here for
x TDMA multiple access
comparing their characteristics with ideal filter characteristics.
x Maximum of 200 Simultaneous demodulation of
bursts channels, bandwidth of 20MHz. 1) Analysis of Magnitude and Phase response of Kaiser
x Symbol rates varies from 64 Ksps to 2 Msps Windowing bandpass FIR filter
x Capable of Viterbi/Turbo Decoding

0688
 Fig. 2. Magnitude and Phase response of the Kaiser windowing FIR
bandpass filter

Fig. 6. Impulse response of the Equiripple FIR bandpass filter

Fig. 3. Impulse response of the Kaiser windowing FIR bandpass filter

Fig. 7. Pole/Zero plot of the Equiripple FIR bandpass filter

3) Analysis of Magnitude and Phase response of least

square error bandpass FIR filter.

Fig. 4. Pole/Zero plot of the Kaiser windowing FIR bandpass filter

2) Analysis of Magnitude and Phase response of

Equiripple bandpass FIR filter

Fig. 8. Magnitude and phase response of the Least square error FIR
bandpass filter

Fig. 5. Magnitude and Phase response of the Equiripple FIR bandpass

filter
Fig. 9. Impulse response of the Least square error FIR bandpass filter

0689
 Table III: Structure of filters

Sr. Method Number of Number Filter

No. of Adders Length
Multipliers
1 Kaiser 35 34 41
windowing
2 Equiripple 41 40 41

3 Least 41 40 41
square
Fig. 10. Pole/Zero plot of the Least square error FIR bandpass filter

V. RESULTS VI. CONCLUSION

The FIR filters are implemented with selected order of 40 for
comparison of passband ripple and stopband attenuation as The FIR filters designed using various methods for burst
shown in table I. They are also designed under the constraint mode satellite receiver are compared for their impulse
of minimum order for given specifications as shown in table response, frequency response, ripple characteristics, order of
II. The designed filter structures are also compared for the filter and hardware complexity in terms of number of
hardware complexity in terms of no. of adders and multipliers adders and multipliers. Kaiser window based design is
as shown in table III. These results are reproduced in the optimum and can be further implemented onto FPGA for final
following tables. We can see that the equiripple design is the integration with adjacent communication system blocks.
best in terms of stopband attenuation. Under the constraints of
the order, again equiripple design performs much better REFERENCES
compared to other two types. Kaiser window based design [1] G. Eas K. S. D. Deepak Mishra and S.Jit, “Efficient qpsk burst
proves to be the optimum design in terms of hardware demodulator for onboard application.”
complexity viz. no. of multipliers. [2] L. Jieshan and H. Shizhen, “An design of the 16-order fir digital filter
based on FPGA,” in Information Science and Engineering (ICISE), 2009
1st International Conference on. IEEE, 2009, pp. 489–492.
Table I: Passband ripple and stopband attenuation [3] T. W. Parks and C. S. Burrus, “Digital filter design,” 1987.
[4] B. HOU, Y. YAO, and M. QIN, “Design and FPGA implementation of
Sr. Method Passband Stopband Order high-speed parallel fir filters,” 3rd International Conference on
No. Ripple attenuation Mechatronics, Robotics and Automation (ICMRA), 2015.
[5] Y. Gao and L.-l. Zhang, “Simulation study of fir filter based on matlab,”
(dB) (dB) in Wireless Communications Networking and Mobile Computing
(WiCOM), 2010 6th International Conference on. IEEE, 2010, pp. 1–4.
1 Kaiser 1 -25 40 [6] S. Yu and F. Yisi, “16-order fir filter design based on matlab and its
windowing quartus ii simulation,” in Computer, Mechatronics, Control and
Electronic Engineering (CMCE), 2010 International Conference on, vol.
5. IEEE, 2010, pp. 44–46.
2 Equiripple 0 -42 40 [7] G. Proakis John and G. Manolakis Dimitris, “Digital signal processing:
principles, algorithms, and applications,” Pentice Hall, 1996.
[8] Z. Sun, W. Yu, and C. He, “Fir digital filter design and matlab
3 Least square 0 -36 40
simulation,” in Measurement, Information and Control (MIC), 2012
International Conference on, vol. 2. IEEE, 2012, pp. 677–680.
[9] M. B. I. Reaz, M. T. Islam, M. Sulaiman, M. A. M. Ali, H. Sarwar, and
S. Rafique, “Fpga realization of multipurpose fir filter,” in Parallel and
Table II: comparison with different orders Distributed Computing, Applications and Technologies, 2003.
PDCAT’2003. Proceedings of the Fourth International Conference on.
Sr. Method Passband Stopband Min. IEEE, 2003, pp. 912–915.
No. Ripple attenuation [10] A. A. Eleti and A. R. Zerek, “Fir digital filter design by using windows
Order method with matlab,” in Sciences and Techniques of Automatic Control
(dB) (dB) and Computer Engineering (STA), 2013 14th International Conference
on. IEEE, 2013, pp. 282–287.
1 Kaiser 0 -59 88 [11] J. Bilmes and M. Song, “Fir design by windowing, kaiser window &
windowing optimal approximation,” Nov 28, 2001.
[12] T. Kaur, “Approach for design of fir filter using kaiser window,” 1st
2 Equiripple 1 -59 47 International Conference on Research in Science, Engineering &
Management, (IOCRSEM 2014).
[13] B. Kwaha, B. Gyang, and P. Amalu, “A least mean square based method
3 Least square 0 -41 47 of lowpass fir filter design using the matlab toolbox,” 2011.

0690