International Journal of Emerging Technologies and Engineering (IJETE) ISSN: 23488050
ICRTIET-2014 Conference Proceeding, 30th -31st August 2014
Analysis of Power Spectrum Estimation Using Welch Method
for Various Window Techniques
Pranay Kumar Rahi1, Rajesh Mehra2
1
ME Scholar, 2Associate Professor,
1,2
Department of Electronics & Communication Engineering
National Institute of Technical Teachers Training & Research
Chandigarh, UT, India
ABSTRACT
In this paper, Power Spectral Estimation (PSE) scheme
for variable data length using Rectangular, Blackman,
Hanning, Bartlett and Hamming window with Welch
Method. Welch method includes the periodogram
having advantage of possible implementation using the
Fast Fourier Transform (FFT). This method gives good
resolution when optimal selection of data length
samples is done. The PSE based on both Rectangular
as well as Hamming window has been designed and
simulated using MATLAB. It can be observed that the
Rectangular and Hamming give better results than
other windows like Bartlett, Hanning and Blackman
window.
Keywords- DSP, FFT, FIR, MAC, PSE, PSD, Welch
Method
1. INTRODUCTION
Power Spectrum estimation (PSE) is one of the most
important areas of research and applications in Digital
Signal Processing (DSP). PSE techniques and
algorithm that nonparametric or classical methods
based on the periodogram. PSE is the spectral
characteristics of signals characterized as random
processes. The autocorrelation function of a random
signal is the appropriate statistical average that will use
for the characterizing random signals in the time
domain, and the power density spectral is the Fourier
transform of the autocorrelation function, provides the
transformation from the time domain to the frequency
domain [1].
A power spectrum describes the energy distribution of
a time series in the frequency domain. Energy is a realvalued quantity, so the power spectrum does not
contain phase information. Because a time series may
contain non-periodic or asynchronously-sampled
periodic signal components, the power spectrum of a
time series typically is considered to be a continuous
function of frequency.
The nonparametric Welch method in which the power
of any input is guesstimation at different frequencies.
It is an improvement on the Periodogram (a method of
estimating the autocorrelation of finite length of a
signal) spectrum estimation method where signal to
noise ratio is high and reduces noise in the estimated
power spectra in exchange for reducing the frequency
resolution. Periodogram is easy to compute and have
limited ability to produce accurate power spectrum
estimation. It is biased when dealing with finite
windows. As data length increase, the rate of
fluctuation in this is also increase [2]. An improved
version of the periodogram is Welch's method. A more
modern nonparametric technique is the multitaper
method (MTM).
The goal of spectral estimation is to describe the
distribution (over frequency) of the power contained in
a signal, based on a finite set of data. Estimation of
power spectra is useful in a variety of applications,
including the detection of signals buried in wide-band
noise. Spectrum analysis has found use in the study of
communication engineering signals, event-related or
stimulated
responses
of
the
human
electroencephalogram (EEG) in the diagnosis of brain
diseases, other biological signals, meteorological
industrial process control and the measurement of
noise spectra for optimal linear filters [3].
The deleterious effects of spectral leakage and
smearing may be minimized by windowing the data by
a suitable window function. The sampled data value
are multiplied point to point by the sampled value of
selected window function. The equivalent noise
bandwidth, processing gain, worst-case processing
loss, and minimum resolution are considered for
choosing a suitable window. In overlap correlation
averaging the spectrum of the windowed data directly
leads to a significantly improved estimate of the
spectrum [4].
2. POWER SPECTRUM ESTIMATION
PSE is most important application area in Digital
Signal Processing. There are mainly two types of
power spectrum estimation (PSE) method: Parametric
and nonparametric. Parametric or non-classical
methods an analyzed process is replace by an
appropriate model with known spectrum. Non-
Divya Jyoti College of Engineering & Technology, Modinagar, Ghaziabad (U.P.), India
106
International Journal of Emerging Technologies and Engineering (IJETE) ISSN: 23488050
ICRTIET-2014 Conference Proceeding, 30th -31st August 2014
parametric or classical methods do not make any
assumption on the data generating process.
Parametric methods are based on parametric models
of a time series, such as AR models, moving average
(MA) models, and autoregressive-moving average
(ARMA) models. Therefore, parametric methods also
are known as model-based methods. To estimate the
PSD of a time series with parametric methods, you
need to obtain the model parameters of the time series
first.
In Nonparametric methods, which include
the periodogram method, Welch method and the
blackman-Turkey methods, are based on the discrete
Fourier transform. In these methods no need to obtain
the parameters of the time series. All these methods
have the advantage of possible implementation using
the FFT, but with the disadvantage in the case of short
data lengths of limited frequency resolution. Also,
considerable care has to be exercised to obtain
meaningful result. Parametric methods on the other
hand can provide high resolution in addition to being
computationally efficient. The most common
parametric approach is to derive the spectrum from the
parameters of an autoregressive model of the signal
[4]. In Bartlett method, divide the signal into blocks,
find their periodograms and average to get the Power
spectrum. (The data segments are non-overlapping).
The final effect is true power spectrum convolved with
a window. Due to windowing (leakage frequency due
to side lobes) the frequency resolution is low.
In Welch Method, data segments can be overlapping
and Window the data (signal) before computing
Periodogram (we may use different windows for each
segment). This method has better precision but less
frequency resolution than Bartlett method. In
Blackman-Tukey Method, windowed the autocorrelation sequence and take Fourier transform to get
power spectrum estimate (Periodogram) in effect we
smooth out the Periodogram. It has better variance
(even at large lags) and better precision than above two
methods, but frequency resolution is less than the
others [5].
The estimates are based entirely on a finite record of
data, the frequency resolution is equal to the spectral
width of rectangular window of length N, which is
approximately 1/N at the
-3dB points. The
estimation techniques decrease the frequency
resolution in order the variance in the spectral
estimate. The estimates are computed at discrete
frequencies [1].
two respects. First, the data segments in the Welch
method are allowed to overlap. Second, each data
segment is windowed prior to computing the
periodogram.
The method consists of dividing the time series data
into (possibly overlapping) segments, computing a
modified periodogram of each segment, and then
averaging the PSD estimates. The result is Welch's
PSD estimate. Welch's method is implemented in the
Signal Processing Toolbox function. By default, the
data is divided into eight segments with 50% overlap
between them. A Hamming window is used to
compute the modified periodogram of each segment.
The averaging of modified periodograms tends to
decrease the variance of the estimate relative to a
single periodogram estimate of the entire data record.
Although overlap between segments tends to introduce
redundant information, this effect is diminished by the
use of a nonrectangular window, which reduces the
importance or weight given to the end samples of
segments (the samples that overlap).
In the Welch method L data sections of length M are
overlapped and the periodograms are computed from
the L windowed data sections. The periodograms are
normalized by the factor U to compensate for the loss
of signal energy owing the windowing procedure.
(1)
The Welch power density spectral estimate, PWE(f), is
(2)
The expected value of the Welch estimate is
(3)
For 50% overlap (L=2K)
(4)
Which is less than for the Bartlett by the factor
9/16=0.56 [4].
4. RESULT
In this paper, we are trying to show the Data length
effect on resolution with data sequence of 312
samples, fs = 1000 and Overlay 50%.
3. WELCH METHOD
The Welch method [Welch 1967] is an improved
estimator obtained by chaning the Bartlett method in
Divya Jyoti College of Engineering & Technology, Modinagar, Ghaziabad (U.P.), India
107
International Journal of Emerging Technologies and Engineering (IJETE) ISSN: 23488050
ICRTIET-2014 Conference Proceeding, 30th -31st August 2014
Fig. 1. Data sequence in Hanning window
Fig. 4. Data sequence in Blackman window
Fig. 5. Data sequence in Rectangular window
Fig. 2. Data sequence in Hamming window
Fig. 3. Data sequence in Bartlett window
5. CONCLUSION
The Welch method provides the result on power
spectrum estimation for various windows in window
method. This method is used to calculate the PSD of a
signal with reducing the effect of noise in DSP. In this
paper the window techniques like Hamming, Hanning,
Bartlett, Blackman and Rectangular window are used
to extract the unwanted noise from the signal. The
sampling frequency 1000Hz fixed and the number of
sample are 312 with 50% overlay. We can see the
differences / variations in the different windows. The
purposed algorithms operate in frequency domain,
where the calculation of samples is done from the
frequency domain using sine waveforms. The quality
of the estimate increase as the length N of the data
increase, which means that the consistence. When data
length is short Blackman tuckey method is better than
Welch method but as the data length increase Welch
Method gives excellent results. The rectangular and
Divya Jyoti College of Engineering & Technology, Modinagar, Ghaziabad (U.P.), India
108
International Journal of Emerging Technologies and Engineering (IJETE) ISSN: 23488050
ICRTIET-2014 Conference Proceeding, 30th -31st August 2014
Bartlett window have the clear peak in the graph
showing the power spectrum estimation.
REFERENCES
[1] J.G.Proakis and D.G.Manolakis, Digital Signal
Processing Principles, Algorithms and Applications
Forth Edition, Prentice-Hall, pp.908-913, 2009.
[2] Hansa Rani Gupta, Sushila Batan and Rajesh
Mehara, Power Spectrum Estimation using Welch
Method
for
various
Window
Techniques,
International Journal of Scientific Research
Engineering & Technology (IJSRET), Vol. 2, Issue 6,
pp 389-392, September 2013.
[3] Sanjit k Mitra, Digital signal Processing, Tata
Mcgraw Hill, Second Edition, pp. 523-524, 2006
[4] Emmanual C. Ifeachor, Barrie W. Jervis,Digital
Signal Processing A practical Approach, Person
Education, Second Edition, pp. 864-868, 2005.
[5] http://www.oneirix.com/www.udayankanade.org
//adsp/jan2004/lecture017.pdf
of Technical Teachers Training &
Research,
Punjab
University,
Chandigarh, India in 2008. He is
pursuing Doctor of Philosophy
degree
in
Electronics
and
Communication Engineering from
National Institute of Technical
Teachers Training & Research,
Punjab University, Chandigarh,
India. He is an Associate Professor
with the Department of Electronics
& Communication Engineering,,
National Institute of Technical
Teachers Training & Research,
Ministry of Human Resource
Development, Chandigarh, India.
His current research and teaching
interests are in Signal and
Communications Processing, Very
Large Scale Integration Design. He
has authored more than 175 research
publications including more than 100
in Journals. Mr. Mehra is member of
IEEE and ISTE.
[6] Mathworks, Users Guide Filter Design Toolbox
4, March-2007.
AUTHORS
Pranay Kumar Rahi received the
Bachelors of Technology degree in
Electronics and Telecommunication
Engi- neering from Government
Engineering College, Guru Gasidas
University, Bilaspur, Chhattisgarh,
India in 2004, and pursuing Masters
of Engineering in Electronics and
Communication
Engineering
from National Institute of Technical
Teachers Training & Research,
Punjab University, Chandigarh,
India.
Rajesh Mehra
received
the
Bachelors of Technology
degree
in Electronics and Communication
Engineering from National Institute
of Technology, Jalandhar, India in
1994,
and
the
Masters of
Engineering
degree
in
Electronics and Communication
Engineering from National Institute
Divya Jyoti College of Engineering & Technology, Modinagar, Ghaziabad (U.P.), India
109
