Particle Swarm Optimization Algorithm Based Window Function Design

w w w . i j m r e t . o r g I S S N : 2 4 5 6 - 5 6 2 8 Page 6
International Journal of Modern Research in Engineering and Technology (IJMRET)
www.ijmret.org Volume 3 Issue 10 ǁ October 2018.
Particle Swarm Optimization Algorithm Based
Window Function Design
Ibrahim TUTAL, Duygu KAYA, Turgay KAYA
(Department of Electrical and Electronics Engineering, Fırat University, Turkey)
ABSTRACT:The window functions used for digital filter design are used to eliminate oscillations in
the FIR (Finite Impulse Response) filter design. In this work, the use of Particle Swarm Optimization
(PSO) algorithm is proposed in the design of cosh window function, in which has widely used in the
literature and has useful spectral parameters. The cosh window is a window function derived from the
Kaiser window. It is more advantageous than the Kaiser window because there is no power series
expansion in the time domain representation. The designed window function shows better ripple ratio
characteristics than other window functions commonly used in the literature. The results obtained
were presented in tables and figures and successful results were obtained
KEYWORDS -Cosh window, PSO, FIR filter
I. INTRODUCTION
Window functions are time-domain
functions that can be used to remove Gibbs'
oscillations in the FIR filter design. Window
functions are widely used in fields such as digital
filter design, signal analysis and prediction, sound
and image processing. Many window function
designs with different properties are proposed in
the literature. The window functions are two types,
fixed and adjustable window functions, according
to the variables they have.
In fixed windows, the window length (N
parameter) controls the mainlobe width of the
window function. Adjustable window functions
with two or more variables can provide a useful
amplitude spectrum. The adjustable window
functions are Dolph-Chebyshev [1], Kaiser [2] and
Saramaki [3] windows. Other window functions
developed based on the Kaiser window are given in
[4,5]. The two-parameter window functions are
insufficient to control the sidelobe roll-off ratio
window reduction which is window spectral
parameters. In the literature, a three-parameter
ultraspherical window function has been proposed
instead of these functions [6-8]. The proposed
window has spectral parameters such as mainlobe
width, ripple ratio, null-to-null width and side-lobe
pattern. PSOs, one of the heuristic computation
methods, is a method derived from the movement
of animals moving in swarms. With this aspect,
PSO can be represented as a social interaction
model [9-13].
In this work, the use of PSO in the design
of the cosh window function, which is developed
based on the Kaiser window and has better
properties, has been proposed. The results obtained
from the developed method are given by tables and
graphs and the method has been shown successful.
II. WINDOW FUNCTION
Window functions are used to eliminate
Gibbs' oscillations that occur in FIR filter design.
In the window function design with PSO, the
design of the window has been realized by an
alternative method without requiring the design
equations. The window functions are classified
according to their spectral characteristics and
compared the other windows according to these
characteristics. The frequency spectrum of a
window can be defined as follows.
( )
( ) ( )j T j
W e A e  

( 1) /2
0 ( ),j N T j T
e W e  

(1)

( 1) / 2
1
(0) 2 ( )cos
N
n
nT nT  


  
where 0 ( )j T
W e 
is called the amplitude function
of the window, N is the window length and T is the
sampling period. In the equation 1, A(w) = |W0ejwT
|
represents the window's magnitude and θ(w)=-w(N-
1)T/2 represents the window's angle. The
normalized amplitude spectrum of a window can be
obtained by the following equation [5].
𝑊𝑁 𝑒 𝑗𝑤𝑇
| = 20𝑙𝑜𝑔10(|𝐴 𝑤 |/|𝐴 𝑤 | 𝑚𝑎𝑥 ⁡) (2)
The spectral parameters are the mainlobe width
(wM), the null-to-null width (wN), the ripple ratio
(R) and the sidelobe roll off ratio (S), which
determine the window performance. From these
parameters, the mainlobe width determines the
width of the transition band between the pass and
stop band. The ripple ratio determines the ripple at
the pass and stop band, the sidelobe roll of ratio
determines the distribution of the energy in the stop
band.
A. Kaiser Window
At the discrete time, the Kaiser window can be
defined as follows
2
0
0
2
( 1 )
1 1[ ]
( ) 2
0
k
k
n
I
N Nw n n
I
otherwise



       



(3)
whereαk is the adjustable parameter, I0(x) is the first
kind zero-order Bessel function, and the power
series expansion is as follows.
2
0
1
1
( ) 1
! 2
k
k
x
I x
k


  
    
   
 (4)
B. Cosh Window
The cosh window, which is another two parameter
window function, is obtained by writing a cosh
function with similar characteristics instead of I0
(x) function [5].
2
2
cosh( 1 )
1 1[ ]
cosh( ) 2
0
c
c
n
N Nw n n
otherwise



       



(5)
III. PARTICLE SWARM OPTIMIZATION
PSO is an optimization algorithm inspired
by birds and fish moving in flocks. In the PSO
algorithm, each individual is called a particle and
these particles constitute a swarm. Individuals in
the swarm are constantly interacting with other
individuals to achieve the result and with this
interaction they update their current position and
speed thus creating a social model. Each individual
adjusts their position to the position of the best
individual in the herd, taking advantage of previous
experience. PSO algorithm is also an evolutionary
algorithm like Genetic Algorithm (GA). However,
PSO is faster than GA because there are no
operators such as crossover and mutation.
The Basic PSO algorithm:
Every individual in the swarm can be a solution
and every individual is represented by the
dimension vector [9,10].
𝑥𝑖 = 𝑥𝑖1, 𝑥𝑖2, 𝑥𝑖3, … . , 𝑥𝑖𝐷 ∈ 𝑆 (6)
The speed of each individual in the herd is
randomly generated. Each individual has the same
speed as in Equation 7.
𝑣𝑖 = 𝑣𝑖1, 𝑣𝑖2, 𝑣𝑖3,… . , 𝑣𝑖𝐷 (7)
The best local and global positions are
determined. Here, the position of each individual is
defined as follows.
𝑝𝑖 = 𝑝𝑖1, 𝑝𝑖2, 𝑝𝑖3, … . , 𝑝𝑖𝐷 ∈ 𝑆 (8)
Each individual in the PSO adjusts its position
around the individual to Pbest, global and gbest.
The speed and position information of the
individuals are given in equations 9 and 10.
𝑣𝑖
𝑡+1
= 𝑣𝑖
𝑡
+ 𝑐1 𝑟𝑖1 ∗ 𝑝𝑏𝑒𝑠𝑡𝑖
𝑡
− 𝑥𝑖
𝑡
+𝑐2 𝑟𝑖2 ∗ (𝑔𝑏𝑒𝑠𝑡 − 𝑥𝑖
𝑡
)
(9)
𝑥𝑖
(𝑡+1)
= 𝑥𝑖
(𝑡)
+ 𝑣𝑖
(𝑡+1)
, 𝑖 = 1, … … , 𝑃 (10)

Here c1 and c2 are two social and cognitive
acceleration parameters. r1 and r2 are random
numbers between [0,1]. The general PSO algorithm
is as given below.
Algorithm 1. Particle Swarm optimization
algorithm
Set the initial value of P (swarm size) and c1,c2
(acceleration constants)
Set t=0
Generate 𝑥𝑖
(𝑡)
and 𝑣𝑖
(𝑡)
randomly
Evaluate the fitness function 𝑓(𝑥𝑖
𝑡
)
Set 𝑔𝑏𝑒𝑠𝑡(𝑡)
(where gbest is the best glocal
solution)
Set 𝑝𝑏𝑒𝑠𝑡𝑖
𝑡
(where pbestis the best local solution)
repeat
𝑣𝑖
(𝑡+1)
= 𝑣𝑖
(𝑡)
+ 𝑐1 𝑟𝑖1 ∗ 𝑝𝑏𝑒𝑠𝑡𝑖
𝑡
− 𝑥𝑖
𝑡
+𝑐2 𝑟𝑖2 ∗ (𝑔𝑏𝑒𝑠𝑡 − 𝑥𝑖
𝑡
)
𝑥𝑖
(𝑡+1)
= 𝑥𝑖
(𝑡)
+ 𝑣𝑖
(𝑡+1)
, 𝑖 = 1, … … , 𝑃
Evaluate the fitness function 𝑓(𝑥𝑖
𝑡+1
)
if𝑓(𝑥𝑖
𝑡+1
) ≤ 𝑓(𝑝𝑏𝑒𝑠𝑡𝑖
𝑡
)then
𝑝𝑏𝑒𝑠𝑡𝑖
𝑡+1
= 𝑥𝑖
𝑡+1
else
𝑝𝑏𝑒𝑠𝑡𝑖
𝑡+1
= 𝑝𝑏𝑒𝑠𝑡𝑖
𝑡
end if
if𝑥𝑖
𝑡+1
≤ 𝑓(𝑔𝑏𝑒𝑠𝑡 𝑡
)then
𝑔𝑏𝑒𝑠𝑡(𝑡+1)
= 𝑥𝑖
𝑡+1
else
𝑔𝑏𝑒𝑠𝑡(𝑡+1)
= 𝑔𝑏𝑒𝑠𝑡 𝑡
end if
t=t+1
until termination criteria are satisfied.
Show the best particle
IV. DESIGN RESULTS
The cosh window function designed with PSO
for N = 15 and alphac= 1.73 is given in Figure 1
and the graphical data for graph is given in Table 1.
Error variation is given in Figure 2.
Figure 1: Amplitude response of cosh window for N=15
and alphac=1.73
Table 1: Obtaıned Results
Cosh Window WR S R
N=15, alphac=1.73 0.478 11.23 -20.97
Figure 2: Error variation
V. CONCLUSION
In this study, the cosh window function is
designed by using PSO algorithm. The success of
the developed method has been made for the cosh
window which is preferred in the literature in terms
of its properties. The obtained results are shown in
graphical form. The results showed that the method
used was successful in response to the window
amplitude.

REFERENCES
[1] P. Lynch, “The Dolph-Chebyshev window:
a simple optimal filter,” Monthly Weather
Review, vol. 125, pp. 655–660, 1997.
[2] J.F., Kaiser, “Nonrecursive digital filter
design using I0-sinh window function”,
1974, in proc. IEEE Int. Symp. Circuits and
systems (ISCAS’74). pp. 20-23.
[3] T., Saramaki, “A class of window functions
with nearly minimum sidelobe energy for
designing FIR filters”, 1989, in proc. IEEE
Int. Symp. Circuits and systems (ISCAS’89).
pp. 359-362.
[4] K. Avci, A. Nacaroğlu, “A new window
based on exponential function”, IEEE Ph.D.
Research in Microelectronics and
Electronics (PRIME 2008), June Istanbul,
Turkey, pp. 69-72, 2008.
[5] K. Avci, A. Nacaroğlu, “Cosh window
family and its application to FIR filter
design”, International Journal of
Electronics and Communications-AEU, vol.
63, pp. 906-917, 2009.
[6] S.W.A. Bergen, A. Antoniou, “Generation
of Ultraspherical window functions”, 2002,
in XI European Signal Processing
Conference, Toulouse, France, September,
vol. 2, pp. 607-610.
[7] S.W.A. Bergen, A. Antoniou, “Design of
Ultraspherical Window Functions with
Prescribed Spectral Characteristics”,
EURASIP Journal on Applied Signal
Processing, vol. pp.13 2053-2065, 2004.
[8] S.W.A. Bergen, A. Antoniou, “Design of
Nonrecursive Digital Filters Using the
Ultraspherical Window Function”,
EURASIP Journal on Applied Signal
Processing, vol. 12, pp. 1910-1922, 2005.
[9] M. Y.Ozsaglam, M.Cunkas, “Particle
Swarm Optimization Algorithm for Solving
Optimızation Problems”, Politelnikdergisi,
vol. 11, 2008.
[10] R.J. Kuo, Y.S. Han, “A hybrid of genetic
algorithm and particle swarm optimization
for solving bi-level linear programming
problem – A case study on supply chain
model”, Applied Mathematical Modelling,
vol. 3,5 pp. 3905–3917, 2011.
[11] A. Aggarwal, T. K. Rawat, D. K. Upadhyay,
“Design of optimal digital FIR filters using
evolutionary and swarm optimization
techniques”, AEU - International Journal of
Electronics and Communications, vol.70, 4,
pp. 373-385, 2016.
[12] F. Javidrad, M. Nazari, “A new hybrid
particle swarm and simulated annealing
stochastic optimization method”, Applied
Soft Computing, vol. 60, pp. 634-654, 2017.
[13] J. Cervantes, F. Garcia-Lamont,
L.Rodriguez, A. López, J. R. Castilla, A.
Trueba, “PSO-based method for SVM
classification on skewed data sets”,
Neurocomputing, vol. 228, pp. 187-197,
2017.

Particle Swarm Optimization Algorithm Based Window Function Design

More Related Content

What's hot (20)

Similar to Particle Swarm Optimization Algorithm Based Window Function Design (20)

More from International Journal of Modern Research in Engineering and Technology (20)

Recently uploaded (20)

Particle Swarm Optimization Algorithm Based Window Function Design