Techniques - Review - Krishna Gopal Soni

advertisement
International Journal of Engineering Trends and Technology (IJETT) – Volume 9 Number 4 - Mar 2014
Design of Digital Filters using Different
Techniques - Review
Krishna Gopal Soni1 , Mr.Manish Gurjar2
1
M.tech.(student), TIT college, Bhopal
2
ABSTRACT
- Digital filters
Asst. Professor , TIT College, Bhopal
measure wide employed
in completely different areas. they're easier in storage
and
maintenance.
As
there
is
an
excellent
advancement in style techniques for varied digital
filters. style techniques when the early windowing
methodology,
approaches
there
square
conjointly that
measure
suffer
from
another
some
reasonably drawbacks i.e. a number of them couldn't
provide optimum style in any sense, some square
measure lacking of generality, and a few wants long
computing then on. In this paper we are showing
ripples within the passband and stopband,stopband
attenuation and therefore the transition breadth.
These varied windows limit the infinite length
impulse response of ideal filter into a finite window
to style AN actual response. moreover, windowing
strategies don't enable sufficient management of the
frequency response within the varied frequency
bands and different filter parameters such as
transition
breadth.
Linear-phase
filters
square
measure sometimes designed as non-recursive (FIR)
filters which may have constant cluster delay over the
complete base-band. However, once extremely
some survey based on design of digital filters.
selective filters square measure needed, a awfully
high filter order is required that makes these filters
Keywords— ANN, DIGITAL FILTER, SIOM
uneconomical or impractical. To eliminate this
I.
downside, tries are created to develop strategies to
INTRODUCTION
style
algorithmic
(IIR)
filters
whose
delay
Digital filters exist in 2 types: Finite impulse
characteristics approximate a relentless worth within
response (FIR) and Infinite impulse response (IIR) or
the passband. This includes IIR filter style approach
algorithmic.FIR filters suffer from the matter of high
that
order (hence implementation and performance issues)
characteristics at the same time.in next part of this
if strict necessities are obligatory at the planning
paper we are going to discuss some different design
stage. Different techniques exist for the planning of
techniques.
may
satisfy
each
magnitude
and
part
digital filters. Windowing method; during which the
best impulse response is increased by a window
II.
ARTIFICIAL NEURAL NETWORKS
operate, is that the preferred. There square measure
ANN has been wide utilized in the appliance of
varied styles of window functions (Butterworth,
communication systems. The ANN is networks of
Chebyshev, Kaiser etc.), counting on the wants on
simple process components known as neurons.
they're connected to every different by weights. every
ISSN: 2231-5381
http://www.ijettjournal.org
Page 155
International Journal of Engineering Trends and Technology (IJETT) – Volume 9 Number 4 - Mar 2014
vegetative cell multiplies the incoming signals by the
and harmonic parts elimination) that area unit the
corresponding weights and sums then up. If this add
magnitude response at intervals tiny information
or the activation price is over threshold, the
measure as well as sharp frequency edges in addition
vegetative cell changes its output. The network may
as AN about constant cluster delay during this band
be trained to adjust its weights within the learning
area unit needed too. Usually the best optimum price
section. Still, the network is in a position to perform
of all the target functions of this filter style is
some task additional simply than a traditional pc due
obtained for a few values of style variables. A
to huge property and parallel operations of all the
compromise or a trade-off between the target
weather. It resembles brain in 2 respects: A neural
functions should be created to realize a satisfactory
network acquires information through learning. A
filter style. The thought-about algorithmic digital
neural network's information is hold on inside inter-
filter should satisfy 3 multi-objective functions.
neuron association strengths called conjugation
These functions are:
weights. Artificial Neural Networks area unit being
counted because the wave of the longer term in
1) Meet a nominative or a desired magnitude
response specification;
computing. they're so self-learning mechanisms that
2) AN about constant cluster delay; and
do not need the standard skills of a technologist.
3) A minimum time response or sinking time
Currently, only a few of those neuron-based
that involves a minimum section or a gaggle
structures,
delay.
paradigms
really,
area
unit
being
employed commercially. The power and quality of
The optimization approach considers the discrete-
artificial neural networks are incontestable in many
time transfer perform that is developed on the
applications including speech synthesis, diagnostic
premise of some desired amplitude response and a
issues, medicine, business and finance, robotic
stability margin parameter. A norm of the weighted
management, signal processing, pc vision and lots of
error perform is then decreased with relevance the
different issues that constitute the class of pattern
transfer-function coefficients with a prescribed most
recognition. for a few application areas, neural
pole radius named as stability margin. the steadiness
models show promise in achieving human-like
margin parameter is varied to optimize the filter
performance over additional ancient AI techniques.
coefficients that minimizes principally the magnitude
response, satisfies the simplest about constant cluster
III.
DIGITAL FILTERS TECHNIQUE
delay and therefore the lowest cluster delay that ends
A digital filter primarily based answer is planned to
up in minimum sinking time or time delay of the
get rid of unwanted disturbances exploitation digital
system dynamic response.
filter style techniques. The filter time response should
be enclosed within the necessities. the current
filtering
application
imposes
different
IV.
THE SPIRAL INSPIRED
OPTIMIZATION METHOD
quite
specifications. On one hand, the time domain demand
Compared with ancient improvement techniques and
wherever each a high speed and correct system
different world optimizers, the spiral improvement
response area unit required. On the opposite hand, the
methodology is easy to implement and really
frequency domain necessities (DC, sub-synchronous
economical in reaching optimum solutions. Spiral
ISSN: 2231-5381
http://www.ijettjournal.org
Page 156
International Journal of Engineering Trends and Technology (IJETT) – Volume 9 Number 4 - Mar 2014
improvement
methodology
has
been
recently
V.
CONCLUSION
developed supported the analogy to spiral phenomena
This paper suggests the neural network technique for
(Tamura and Yasuda 2011a; Tamura and Yasuda
designing linear phase FIR filter. Based on the
2011b).Patterns in nature are visible regularities of
various algorithms of neural network we concluded
type found within the plants. These patterns recur in
that the designed model of FIR filters using neural
numerous contexts and can generally be modelled
network are have better performance than the
mathematically.
conventional design method of FIR filter.
Natural
patterns
embrace
symmetries, trees, spirals, meanders, waves, foams,
arrays, cracks and stripes. Arithmetic, physics and
REFERENCES
chemistry will justify patterns in nature at totally
different levels. Patterns in living things are
[1] A. V. Oppemheim, and R.W. Schafer, “Discrete –Time Signal
Processing”, Englewood Cliffs: Prentice-Hall Press, 1989.
explained
by
the
biological
processes
of
[2] V. R. Alagzi, and M. Suk, “On the Frequency Weight Least
choice|survival|survival of the fittest|selection|natural
Squares Design of Finite Duration Filters”, IEEE Trans. Circuits
process|natural action|action|activity} and sexual
Syst., vol. 22, no. 12, pp. 943-953, Dec. 1975.
selection. Studies of pattern formation build use of pc
[3] G. C. Goodwin, and K. S. Sin, Adaptive Filtering Prediction
and Control, Englewood Cliffs:
models to simulate a good vary of patterns.Among
Prentice-Hall Press, 1989.
the natural patterns, spirals are common in plants and
[4] X. P. Lai, “A Random Sampling Recursive Least-Squares
in some animals. For instance, within the nautilus,
Approach to the Design of FIR Digital Filter”, China Journal of
every chamber of its shell is AN approximate copy of
Signal Processing, vol. 15, no. 3, pp. 260-264, Mar. 1999.
[5] Simon Haykin, “Neural Networks”, Second edition by, Prentice
following one, scaled by a continuing issue and
Hall of India, 2005.
organized in a exponent spiral. The spiral phenomena
[6] S N Sivanandam and S Sumathi, “Introduction To NEURAL
occurring in nature are approximated to exponent
NETWORKS using MATLAB 6.0”, McGraw –Hill, 2006.
spirals as in. Samples of natural spiral dynamics
[7] Martin T. Hagan, “Neural Network Design”, CENGAGE
Learning, Fourth Indian reprint 2011.
embrace whirling currents, air mass fonts, nautilus
[8] Robert J Schalkoff, “Artificial Neural Networks”, McGraw-
shells and arms of spiral galaxies. Exponent spirals
Hill International Editions, 1997.
separate processes to come up with spirals which will
[9] A. Antoniou and D.Bhattacharya, “Design of 2-D FIR Filters
type a good behavior in metaheuristics. A two-
by Feedback Neural Networks”, IEEE Trans.,1995.
[10] Hui Zhao and Juebang Yu, “ A Novel Neural Network –
dimensional formula has been 1st projected (Tamura
and
Yasuda
2011a),
and
then,
a
additional
Based Approach for Designing 2-D FIR Filter”, IEEE Trans., vol.
44, no. 11, 1997.
generalized n-dimensional version has been recently
[11] Xiaohua Wang & Xianzhi Meng, “ A Novel Neural Networks
urged (Tamura and Yasuda 2011b). Within the gift
– Based Approach for Designing FIR Filters”, IEEE Trans.,2006.
[12] A. Antoniou and D.Bhattacharya, “Real-time design of FIR
work, the employment of the spiral improvement
filters by feedback neural networks”, IEEE Trans. ,vol. 3 ,no.
technique is given and wont to solve the multi
5,1996.
objective IIR filter style. First, supported sensible
[13] Hui Zhao & Juebang Yu, “A Novel Neural Network-based
necessities concerned in installation application, an in
Approach for Designing Digital Filters”, IEEE International
Symposium on Ciruits and Systems, 9-12 June 1997. 44
depth mathematical IIR filter style formulation is
given.
[14] Y. D. Jou and F. K. Chen, “Least-Squares Design of FIR
Filters Based on a Compacted Feedback Neural Network”, IEEE
Trans.,vol. 54, No. 5, 5 May. 2007.
ISSN: 2231-5381
http://www.ijettjournal.org
Page 157
International Journal of Engineering Trends and Technology (IJETT) – Volume 9 Number 4 - Mar 2014
[15] Zhe-Zhao Zeng, Ye Chen & Yao-Nan Wang, “Optimal
Design Study of High-Order FIR Digital Filters Based On Neural
Network Algorithm”, IEEE International conference ,Dahan, 13-16
Aug. 2006.
[16] S. Russell, D. Edwards, P. Canny, “ Artificial intellengence: a
modern approach”. Prentice hall Englewood Cliffs, NJ (1995).
[17] Yuichi Kida & Takuro, “ Discrete FIR filter banks minimizing
various measures of approximation error at the same time”, IEEE
Trans., 2010
[18] J. A. Freeman and D. M. Skapura, Neural NetworkAlgorithms, in Applications and Programming Techniques”, ch. 4,
pp. 144-156, Addison-Wesley, 1991.
ISSN: 2231-5381
http://www.ijettjournal.org
Page 158
Download