Professional Documents
Culture Documents
Keywords: Wavelets, wavelet transforms, Fourier transform, utilized for characterizing waveforms the frequency content of
Gabor transform which changes with time. One of the main features that distin-
guishes the three transforms is the choice of the window function.
The purpose of the window function is to localize the information
1. Abstract about the signal, in both space and time. To this end, the T-pulse
technique provides a very flexible window function, which can be
The wavelet transform is described from the perspective of a generated a priori, utilizing computerized optimization techniques.
Fourier transform. The relationships among the Fourier transform, Once such methodologies are known, one can then utilize special
the Gabor transform (windowed Fourier transform), and the wave- pulse shapes to enhance detection performance. Hence, waveform
let transform are described. The differences are also outlined, to shaping and analysis of wave shapes is the main theme of this
bring out the characteristics of the wavelet transform. The limita- paper.
tions of the wavelets in localizing responses in various domains are
also delineated. Finally, an adaptive window is presented that may 3. Mathematical background
be optimally tailored to suit ones needs, and hence, possibly, the
scaling functions and the wavelets. This section summarizes the various analytical tools that per-
form time-frequency analysis. The three techniques that are most
popular are the Fourier transform, the Gabor transform, and the
2. Introduction wavelet transform. These techniques are presented both from a
continuous and a discrete-time signal point of view.
n nature, one does not encounter pure single-tone signals, but
signals the frequency content of which varies with time. Gen- The basic difference that distinguishes the three transfonn
erally, frequency is defined as a phenomenon where the period of techniques as time-frequency localization tools is the choice OP a
zero crossings of a signal has a fixed duration for all times, and window function. The window function is essentially an arti-
frequency is related to the period. However, one may define the fact through which we observe the data. For each of the three trans-
term instantaneous frequency by defining signals the frequency forms, the window function has certain fixed properties. As a
content of which changes as a function of the time or duration of result, each of these analysis techniques has preferred domains
the signal. A good example of this is speech, where the instantane- over which their application provides good results. One of the
ous frequency may change from 20 Hz to 20 kHz,depending on objectives of this paper is to generalize these analysis techniques,
the system. Hence, it is interesting to develop methodologies that and to describe a window function that is very flexible and can deal
can be introduced to analyze such signals. with a very broad class of practical target-identification problems.
In modern times, such concepts have proven to be useful in 4. Continuous transforms
modem radar-system analysis. Conventionally, radar has dealt with
pure CW signals, which are either frequency modulated or tumed 4.1 Fourier transform
off and on to generate pulses. However, there are other radar sys-
tems that deal with wideband pulses. Hence, in understanding how The classical Fourier-transform technique is utilized to find
such radar systems work, it is necessary to understand the time- the frequency content of a particular wave shape, p ( t ) ,which has
frequency representation of waveforms, and how they are charac-
occurred just once, has existed for a time interval [O,T],and is
terized and analyzed.
nonexistent for any other times. It is well known that the frequency
In this study, we look at the classical Fourier transforms, the content of such a signal p ( t ) is given by the Fourier transform
Gabor transforms, and the wavelet transforms, which have been P ( w ):
36 1045-9243/98/$10.0001998 IEEE IEEE Antennas and Propagation Magazine, Vol. 40, No. 6, December 1998
Equation (1) can be interpreted as modulating the function p ( t ) by By observing Equation (5), it is clear that the short-time Fou-
e-wf and then integrating it. The Fourier transform is defined for rier transform is the convolution of the signal p ( t ) with a filter
all values of the variable t [Le., from --co < t < a ] , whereas the having an impulse response of the form h(t) = w(-t)e@, so that
Fourier series of the function p ( t ) is defined to be periodic. That
is, p ( t ) repeats itself after every period (e.g., 27~).Hence, p F ( t ) ,
the Fourier series of p ( t ),is defined as
In the short-time Fourier transform, the function p ( t ) is mul- Altemately, one can also recover p ( t ) from Equation (7) by
tiplied by a window function w ( t ) , and the Fourier transform is multiplying both sides by G(t - p) , and integrating with respect to
calculated. The window function is then shifted in time, and the p . The over-bar denotes the complex conjugate. Then,
Fourier transform of the product is computed again. So, for a fixed
shift P of the window w ( t ) , the window captures the features of
the signal p ( t ) around P . The window helps to localize the time-
domain data, before obtaining the frequency-domain information.
Hence, the short-time Fourier transform is given by
m
IEEE Antennas and Propagation Magazine, Vol. 40, No. 6, December 1998 37
Thus, ,B is a shift parameter, and now the functions ~ , , ~ ( tcan
)
-m
span any function p ( t ) for all possible choices of the parameter a
and ,B. Now, if we look at the Fourier transform of the product of
An attempt was made by Gabor [8],in 1946, to develop a the two functions p ( t ) ~ ~ , ~which
( t ) ,is called the Gabor transform
methodology where a function can be simultaneously localized in G,,p(w) 1 3 then
time and frequency. If that is possible, then the frequency con-
tent of any signal can easily be obtained, by observing the
response in certain narrow frequency bands. Hence, it is possible to
track the instantaneous frequency of a signal.
then Equation (13), when substituted into Equation (1 I), defines Since ( p ; q ) = -1( P ; Q ) , andif q ( t ) = ~ , , ~ ( t ) e @ then
,
2n
the forward Fourier transform.
co m
In the next section, we see how Gabor, through the choice of J p ( t ) q ( t ) d t= jp(t)w,,p(t)e-jox dt
certain window functions, made it possible not only to localize a -m -m
signal in the time domain, but was also able to localize its fre-
quency content in a narrow hequency band. - 1
= G,,p(w) = -P(rrl)Q(Q)dQ.
-m
2z
4.2 Gabor transform [l,8,9]
And since ~ , , ~ ( tis) real, we have
The objective of the Gabor transform is to expand p ( t ) into a
set of functions that are simultaneously limited in both time and
frequency. This is in contrast to the Fourier transform, where the
expansion is done by functions e-Jwt that are not time limited, but
highly localized in frequency.
t2
1 --
q,(t)=-e 4a
2&
Also, since
Then the product p(t)q,(t) can be localized in time from a practi-
cal standpoint if a>O. This is because, beyond a certain value
t = q , the function qa(t) decays down to practically zero, and so
will the product p(t)q,(t). Next, we introduce another parameter,
,L?, so that the function ~ ~ , (which
~ ( t is) real) is defined by we have
38 IEEE Antennas and Propagation Magazine, Vol. 40, No. 6, December 1998
4.3 Wavelet transform
It is clear from Equation (22) that the Gabor transform also local-
izes the Fourier transform, P(w) of p ( t ) ( i t in fact, the window
function wa,B(t) is highly localized), to give its local spectral namely, the average value of the window is zero. This property
gives an extra degree of freedom for introducing a dilation (or
information. Hence, not only is the function p(t) localized in time
scale) parameter in the window, in order to make the time-fre-
by the function ~ ~ , ~ but ) , its transform P(w) is localized in
( talso quency window flexible. With this dilation parameter, the integral
frequency by the function The
W ~ / ( ~ ~ ) , ~ ( Wwidth
) . of the window wavelet transform provides a flexible time-frequency window,
which automatically narrows when observing high-frequency com-
function in the time domain, A t , is then obtained as
ponents, and widens when studying low-frequency components.
Hence, it is in tune with our auditory and visual sense perceptions,
which, at least in the first step, process signals in this fashion. This
explains the evolution of the musical scale in the West. For exam-
ple, Figure 1 shows the location of the notes C and G in the major
diatonic scale, for several octaves [4]. On a logarithmic scale, they
We have would appear to be nearly equally spaced. Thus, the notes C and G
become sparser and sparser as the frequency increases.
and
wa,$(t) = -w
& __ .
and the width of the spectrum of the window function in the fre-
quency domain is given by A. : with a 8 0 .
1
AtAW= -.
2
(28)
p
111 1 1
C Q C 0 c 0
'
c
1
i
So, the width of the time-frequency window is unchanged for
le!
aw H I
wz :,.cu.nn - 1 2 2 1 HI B 116 I 2
628 H t
observing the spectrum of p(t) at all frequencies. In fact, it is seen
that the Gabor transform is essentially a short-time Fourier trans- Figure 1. The pitch frequencies corresponding to the keys of C
form with the smallest time-frequency window. It is the smallest and G on a piano. These correspond to the major diatonic scale
time-frequency window because the principle of uncertainty is of Western music. The spacing is very nonuniform, and will
exactly equal to 1/2, as given by Equation (28). appear to be almost uniform on a logarithmic scale.
IEEE Antennas and Propagation Magazine, Vol. 40, No. 6,December 1998 39
decrease: so that the window adjusts its width according to fre- From the wavelet transform given by Equations (31) or (35),
quency. This is accomplished by the wavelet transform, as outlined the original function can be recovered, utilizing
above. For example, if
(33) (39)
Thus, all the possible states of the window are obtained by fre- Thus, Equation (39) implies that only certain classes of window
quency scaling the response of the window function. This is in functions can be utilized in the wavelet transform, namely, those
contrast to the STFT (short-time Fourier transform), where the 1
responses that decay at least as fast as
various window functions were obtained as W(w)eJpW, Le., by
modulating the spectrum of the window function, which is fixed.
The convergence of the integral in Equation (38) is defined in
Hence, if the center of the window function w(t) is given by
a weak sense [2], Le., taking the inner product of both sides of
t * , and the width of the window in the time domain is given by Equation (38) with any function g ( x ) L 2 and commuting the
A,: then the function ~ , , ~ ( tis) a window function with its center inner product with the integral over a,P on the right-hand side
at p + at*,and its width is aA7. In addition, its spectrum is given leads to the true formula. Since for any absolutely integrable func-
by tion W ( o )-the Fourier transform of w(t)-is continuous, Equation
(39) can only be evaluated provided
w(0)= 0 , (40)
m
where W ( o ) is the Fourier transform of w(t) or, equivalently, j w ( t ) d t = 0 , i.e., the wavelet w(t) has no dc
-m
From Parseval's relation in Equation (19), and utilizing value, as mentioned in Equation (29).
Equation (34), the wavelet transform WTp o f p is given by
Implementation of the wavelet transform is equivalent to fil-
tering the signal p ( t ) by a bank of bandpass filters that are of con-
(35) stant Q. The center hequencies of the filters are offset from one
another by an octave and, in the limit of Equation (39), this guar-
antees that the center frequencies of the bandpass filters will never
The window function in the frequency domain is centered at w *, be zero.
2 A:
and has a width -, with the exception of the multiplicative In the continuous domain, all three techniques (Fourier,
a
Gabor, and wavelet) have good theoretical properties. The question
a
factor ~ , and the phase factor eJpW, The wavelet transform is, what happens in the discrete domain? Do all these properties
274&/ cany over to the discrete domain, or do certain additional con-
straints need to be imposed?!
provides local information in the frequency window
5. Discrete transforms
(44)
and
Hence, one can recover p ( n ) as long as w(0) # 0 . If w(0)= 0 ,
then one chooses a value of /3 for which w(m) = w(n - p) z 0 , and
the procedure continues.
where Tis the sampling interval.
An altemative representation can be made, provided we have
One of the objectives of this representation is to look at the
cell centered at the point ( t k ,f!) in the time-frequency plane, and
how its surrounding regions are distributed. The main objectives of
the representation of Equation (48) is that if the signal within the
For Equation (44), p ( n ) is recovered by T T
window NT - - 5 t 5 NT + - contains a pure sinusoid of fie-
2 2
quency fo,then we would hope that the expression for the time-
limited spectrum would be
m
and that IP(NT,f,T)I2 would have a significant magnitude, pri-
&(z) = C"(k)z-k, (47)
k=-m marily near f0 . However, if one looks at the representation of p ( t )
given by (48) and (49), then it is not at all clear that there would
then if we replace P'~TFT(z,~) in Equation (46) by not be any leakage from nearby cells, where k = N + 1 , N + 2 , or
2
P'sTFT(z,m) + z: , then Equation (46) is still satisfied. This is in N - 1, N - 2 , and so on, to IP(NT,f,T)I . Indeed, the leakage
contrast to the conventional, discrete Fourier transform (or, from neighboring cells can produce erroneous interpretations, as
equivalently, the Fourier series), which provides a unique inverse. described in [9]. This is because neither Equation (49) nor (SO) has
a finite support in the time domain. They extend for all times. One
would expect the signals centered around the cell NT to contribute
5.2 Discrete Gabor transform to P(NT,t , f) if and only if the time-domain support in the repre-
sentation in Equation (48) is finite. Otherwise, the results are not
In the discrete Gabor transform, the objective is to represent relevant. This is why Lemer [ 111 extended the representation in
the signal, p ( t ),by a series, Equation (SO) to have the basis signals of the general form
IEEE Antennas and Propagation Magazine, Vol. 40, No. 6, December 199% 41
modified by Roach [SI, who demonstrated that unless the function Equation (62) guarantees that the coefficient ckf for a given
v ( t ) is of finite support in the time domain, representations of the segment centered at t = tk is completely independent of the signal
82 outside the Mh segment. To confirm this, observe that
form in Equation (48) produce an energy spectrum IP(NT,f,T)I
that has no clear relationship to the energy of the signal concen-
trated in the corresponding interval in time. Note that this pre-
cludes the Gaussian fimction or the prolate-spheroidal functions as
possible window functions, since they are of infinite duration!
Now note that, always, k = m : otherwise, the window functions
It has been shown that for a proper time-hequency represen-
tation, the expansion must be of the form [9] w(t - t k ) and w(t - tm) are disjoint. We have
-m tk -Ti2
where
and finally
Ti2
WT(W)= ~ ( t ) dt , (57) Utilizing the property of Equations (58) and (59) of the window
-Ti2 function, it is seen that Equation (65) becomes unity for e = n and
zero otherwise, satisfying Equation (62), and hence Equation (61)
and holds. The interesting point in Equation (61) is that the coefficient
for a given segment in time is completely independent of the signal
outside the kth cell, since the coefficients in different time seg-
ments are orthonormal, Le., the total energy in the signal is then the
sum of all the individual coefficients squared.
with
It is interesting to observe that Equations (56) and (61) have
(59) close resemblance to the windowed Fourier transform, presented in
the earlier sections, when the window function becomes a rectan-
gular window. However, any window function that results from the
The class of window functions that possesses this property is convolution of any rectangular pulse with a symmetric window
formed by convolving a rectangular pulse of duration (T - p) with function will provide the mathematical requirements for ~ ( t in)
a symmetric positive pulse a ( t ) having duration p and area Equations (56) to (59).
__1
5.3 Discrete wavelet transform
T-p'
If, in the continuous wavelet transform, one uses integer val-
Here,
ues for some integers k,n in Equation (32), and usually assumes
p = 2knT (one can assume T = 1 without loss of generality for the
discrete case) and a = 2 k , then the discrete wavelet transform of
p ( t ) is given by [4]
where Q is the integer frequency-spacing factor. The coefficients
eke in Equation (55) can be given by
c n m
provided p ( t )= 2 WT,(k,n)2-k'2w(2-kt-n~), (67)
The + ( x ) are called the scaling functions, and are generated by the
dilation equation
The shift integers n are chosen in such a way that ~ ( 2 - -~ ntT )
covers the whole line for all values of t. The wavelet transform (73)
thus separates the object into different components in its trans-
form domain, and studies each component with a resolution
matched to its scale. Another interesting property to observe is that where, in each case, only a finite number of coefficients cm and
the original signal in Equation (67) is recovered from the discrete- dm are different from zero.
wavelet-transform coefficients by filtering with the window
~ , , ~ ( tof) Equation (68). Equivalently, the discrete wavelet trans- Here, 4 does not have integral zero, but w does, and q5 is
form is obtained by filtering the signal p ( t ) by ~ ~ , ~ ( -ort ) , normalized such that
2-kZw(-2-kt). m
J #(x)dx = 1. (74)
The wavelet series of Equation (67) amounts to expanding -m
the function p ( t ) in terms of wavelets wk,n(t) or, equivalently, by
the shifted-dilated versions of the same window function, so that We define (bk,n even though 4 is not a wavelet, i.e.,
IEEE Antennas and Propagation Magazine, Vol. 40, No. 6,December 1998
possibly, sine functions. Even though they generate orthogonal
bases with some localization, there is a problem, as with the Gabor
basis. Unless the windows have the specific properties as outlined
The functions ~ ~ , ~are( orthonormal
t ) because their bandwidths by Equations (56) to (59) the windowed localized bases of Equa-
tion (61), which are similar to the Malvar wavelets, do not always
are non-overlapping, namely, for a fixed k, P k ( 0 ) has the band-
provide the correct localization picture in the transformed domain!
width A m k , which is [ 2 k n , 2k+n]. So, the wavelet expansion of
a function is complete in the sense that it makes an approximation
by orthogonal functions that have non-overlapping bandwidth. 6. A discussion of possible choices of the window function
As concluded by Vaidyanathan [4],even though the continu- The claim that a particular transform provides simultaneous
ous wavelet transform has a wider scope with deeper mathematical localization, strictly in time and frequency, is preposterous. What
issues, the discrete wavelet transform is quite simple, and can be we have seen is that if a function is limited in time, it cannot
explained in terms of basic filter theory. Even before the develop- simultaneously be limited in frequency, Le., bandlimited. Hence,
ment of wavelets, nonuniform filter banks had been used in speech resolution in time is accomplished at the expense of the frequency,
processing by [12, 131. The motivation was that the nonuniform such that for the best window, we have the situation that the
bandwidths could be used to exploit the nonuniform frequency Heisenberg principle of uncertainty, A t A o 5 112 is satisfied with
resolution of the human ear [14]. So if the wavelet application is the equality. The equality is achieved only for a Gaussian window
already in the digital domain, it is really not necessary to under- function. However, as we have seen for the Gabor transform, the
stand the deeper results of the scaling function 4(t) and wavelets Gaussian window provides nonphysical power spectral density for
~ ( t. In
) this case, all we need to focus on is the dilation equation a region of the function localized in time. The wavelet transform
and the shift principle to generate a complete basis. cannot do any better! For all other choices of the window function,
A t A , is > 112.
5.3.1 First word of caution. Often, over-zealous researchers, Other possible choices for the window may be the prolate-
in order to push the advantages of this newly developed tool spheroidal functions. Unfortunately the prolate-spheroidal func-
(which, of course, is quite useful for selective problems), overlook tions are not limited in time, but they are strictly bandlimited. An
two important factors. altemate choice is to deal with the truncated prolate-spheroidal
functions, which are strictly time limited, but one does not have
The first one is termed a wavelet crime by Strang and any control over the spectrum of such functions [15-171. That is
Nguyen. The problem has to do with the selection of sample values why a flexible methodology has been developed, where one can
of the function as the wavelet coefficients. As Strang and Nguyen design a window shape that it is finite in time and, in addition, can
point out, ...start with a function ~ ( t )Its. samples x ( n ) are often focus the energy into a pre-specified band [18, 191.
the input to the filter bank (to generate the wavelet decomposition).
Is this legal? NO. IT IS A WAVELET CRIME. Some cant imag- However, what one may try to do is to have a finite time
ine doing it, others cant image not doing it. Is this crime conven- window for localizing a process in time. However, its shape can be
manipulated in such a way that it is approximately bandlimited for
ient? YES. We may not know the whole function n(t), it may not
most practical purposes, Le., 99.9% of its energy is within a very
be a combination of 4(t - k ) , and computing the true coefficients narrow prescribed band. Such a construct is called a T pulse. A T
in x a ( k ) 4 ( t - k ) may take too long. But the crime cannot go pulse is a pulse limited in time T. Moreover, its spectral content,
unnoticed-we have to discuss it [20,p. 2321. Le., 99.9% of its energy, is focussed within a few band times (i.e.,
44 IEEE Antennas and Propagation Magazine, Vol. 40, No. 6,December 1998
assume there are N,. samples per baud rate (the inverse of baud
time), then
where 6 ( k ) is the impulse function. This guarantees that if the Step 1. Choose an initial guess for f (m)and an initial guess
waveform is shifted by a baud time or its multiples, then the wave for the weights we, wm, and w p , p = 0, 1, ..., Ne - 1 .
shape is orthogonal to itself. This is termed zero inter-symbol inter-
ference. Note that when k = 0, it is the square of the function Step 2. Compute the gradient of the functional Jwith respect
itself, and no constraint needs to be put on that. In addition, we to f (m) . This is given by
need to put in a dc constraint, i.e., the wave shape should have no
dc component. Hence,
N -1
p=o
where w e , wm, and wp are various weights to the errors E,,,, em,
assuming f ( m ) is real.
ep . The weights should be adjusted in a search procedure that has
been designed to minimize J. Next, an optimum step length to update the signal sequence
In addition, f ( m) is chosen, through one-dimensional searches.
and
m=O
-0.4
N,-1
e, = C f ( m ) f ( m + p N s ) - 6 ( p )for p = l , 2, ..., N c - l . (90) -0.61
m=O -0.8
1
Equivalently, 0 10 20 30 40 50 60 70 BO 90 100
Time Index
Figure 2. A baseband T pulse with zero mean and uon inter-
m=O symbol interference.
-1.5
0 10 20 30 0 50
Time (light-meters)
Time Index
Figure 6. A T pulse with a dc value that is not zero.
-30
-40
Frequency Index
If the errors are small enough, stop the process. If the errors are 0 IO 20 30 0 50 SO 0
-1.0
a 70
Frequency Index Finally, we conclude this section by pointing out the differ-
ences between a T pulse and a wavelet. In the T pulse, a window
Figure 9. The spectral content of the signal in Figure 8. has been created that is not only limited in time, but 99% of the
energy is focused in a narrow band and, in this way, an attempt is
made to satisfy the equality in the Heisenberg principle of uncer-
tainty, i.e.,
A,A, 4 0.5.
0.017
0 015 The expansion in Equation (70) can be done in terms of a dilated
0.013.
O.OI,
and shifted version of the T pulse in an approximate fashion. For
0.00s. the wavelets. the decomposition by Equation (70) is essentially
0.007.
exact, and perfect reconstruction can be done, even after the func-
Y 0.00S.
B 0.003.
tions have been down-sampled [4]. Also, in Equation (70), there is
0.001. a desire to limit the number of nonzero coefficients in Equation
-0.001.
(70), to improve the efficiency of the decomposition. This is
-0.003.
-0.005 achieved by enforcing that the pth derivative and all its lower
-0.007. derivatives are zero at co = 0 , Le.,
-0.009.
-0.0,1.
-0.013'
-0.O,S,
(93)
-0.017 \-.-.-.-~,-~.-- , .,.. . . . ... , . . ,.. . .. . ,
0 IO 20 30 0 HI
Time (light-meters)
Depending on the nature of the requirements for the problem of
Figure 10. The current at the center of the dipole antenna, interest, one can use either of the constraints on the window fiinc-
when excited by the incident pulse of Figure 8. tions.
IEEE Antennas and Propagation Magazine, Vol. 40, No. 6, December 1998 47
7. Conclusion 13. R. W. Schafer, L. R. Rabiner, and 0. Herrmann, FIR Digital
Filter Banks for Speech Analysis, Bell System Technical Journal,
This paper provided a survey of the short-time Fourier tech- 54, March 1975, pp. 531-544.
niques, the Gabor transform, and the wavelet transform. The
strengths and the weaknesses of each method have been identified. 14. J. L. Flanagan, Speech Analysis, Synthesis and Perception,
The unity in diversity between the three transforms is the choice of New York, Springer-Verlag, 1972.
the appropriate window function. In addition, it has been shown
how to apply numerical-optimization techniques to design a flexi- 15. I. Gerst and J. Diamond, The Elimination of Intersymbol
ble window function, such that it is strictly time limited, yet 99.9% Interference by Input Signal Shaping, Proceedings of the IRE, 19,
of its energy can be focused in a narrow band. 1963, pp. 1195-1203.
2. I. Daubechies, Ten Lectures on Wavelets, Philadelphia, Society 19. T. K. Sarkar, S. M. Naryana, H. Wang, M. Wicks, and M.
of Industrial and Applied Mathematics (CBMS, 61), 1992. Salazar-Palma, Wavelets and T-pulses, in L. Carin and L. B.
Felsen (eds.), Ultra- Wideband Short-Pulse Electromagnetics 2,
3. R. Gopinath and C. S. Bums, Wavelet Transforms and Filter New York, Plenum Press, 1995, pp. 475-485.
Banks, in C. K. Chui (ed.), Wavelets-A Tutorial in Theoly and
Applications, New Nork, Academic Press, 1992, pp. 603-654. 20. G. Strang and T. Nguyen, Wavelets and Filter Banks,
Wellesley Cambridge Press, 1996.
4. P. P. Vaidyanathan, Multirate Systems and Filter Banks, New
Jersey, Prentice Hall, 1993. 21. T. K. Sarkar, N. Radhaknshna, and H. Chen, Survey of Vari-
ous z-Domain to s-Domain Transformations, IEEE Transactions
5. F. I. Tseng, T. K. Sarkar, and D. D. Weiner, A Novel Window on Instrumentation and Measurements, IM-35, December 1986,
for Harmonic Analysis, IEEE Transactions on Acoustics, Speech pp. 508-520.
andsignal Processing, April 1981, pp. 177-188.
22. M. Vettereli and J. Kovacevic, Wavelets and Subband Coding,
6. W. F. Walker, T. K. Sarkar, F. I. Tseng, andD. D. Weiner, Car- New York, Prentice Hall, 1995.
rier Frequency Estimation Based on the Location of the Spectral
Peak of a Windowed Sample of Carrier Plus Noise, IEEE Trans-
actions on Instrumentation and Measurements, IM-31, December
1982, pp. 239-249.
Introducing Feature Article Authors
7. W. F. Walker, T. K. Sarkar, F. I. Tseng, and J. Cross, Optimum
Windows for Carrier Frequency Estimation, IEEE Transactions Tapan Kumar Sarkar received the B. Tech. degree from the
on Geoscience Electronics, GE-21, October 1983, pp. 446-454. Indian Institute of Technology, Kharagpur, India, the MScE degree
from the University of New Brunswick, Fredericton, Canada, and
8. D. Gabor, Theory of Communications, Journal of the Institute the MS and PhD degrees from Syracuse University, Syracuse, New
for Electrical Engineers, November 1946, pp. 429-457. York in 1969, 1971, and 1975, respectively.
9. J. E. Roach, A Vector Space Approach to Time-Variant Energy From 1975 to 1976, he was with the TACO Division of the
Spectral Analysis, PhD thesis, Syracuse University, Syracuse, New General Instruments Corporation. From 1976 to 1985, he was with
York, 1982. the Rochester Institute of Technology, Rochester, N Y . From 1977
to 1978, he was a Research Fellow at the Gordon McKay Labora-
10. M. Bastiaans, Gabors Expansion of a Signal into Gaussian tory, Harvard University, Cambridge, MA. He is now a Professor
Elementary Signals, Proceedings ofthe IEEE, 68, April 1980, pp. in the Department of Electrical and Computer Engineering, Syra-
538-539. cuse University; Syracuse, N Y . He has authored or co-authored
more than 170 joumal articles, and has written chapters in ten
11. R. Lemer, Representation of Signals, in E. Baghdady (ed.), books. His current research interests deal with numerical solutions
Lectures on Communrcation Svstem Theow, New York, McGraw- of operator equations arising in electromagnetics and signal proc-
Hill, 1961, Chapter 10. essing, with application to system design.
12. G. A. Nelson, L. L. Pfeiffer, and R. C. Wood, High Speed Dr. Sarkar is a registered Professional Engineer in the State
Octave Band Digital Filtering, IEEE Transactions on Audio and of New York. He was an Associate Editor for Feature Articles of
Electroacoustics, AU-20, March 1972, pp. 8-65. the IEEE Antennas and Propagation Society Newsletter, and he
48 IEEE Antennas and Propagation Magazine, Vol. 40, No. 6 , December 1998
was the Technical Program Chairman for the 1988 IEEE Antennas ~ ~~
So why settle for a limited planning tool? for indoor and outdoor microcell stu&s
Some planning tools are great at cellular EDX SignalPro includes SignalMXTM
or PCS hut useless otherwise. EDX (pat.pend.) making it the worlds only
SignalProTMis a ful-featured planning planning tool with multimedia
tool for Windows@95/NT that provides simulations of system performance. EDX
comprehensive, accurate designs for also offers an extensive collection of
any system in the 30 MHz to 100 GHz terrain, groundcover,demographic,map
frequency range. image, and vector map data for most
Cellular Wireless local loop areas of the world.
PCS Microcell For 13 years EDX has been the leader in
Pqng Indoor wireless PC-based wireless planning tools. We
MDS Landmobile provlde features and performance that are
LMDS Microwavelink not availahle from any other tool-at any
AM, FM, and TV broadcast (DTV) price-on any platform.
EDX SignalPro and its modules offer 21 Contact us today for a fully operational
different propagation models, including demo CD (not just a slideshow) Youll
transnutter-specilkmeasurement-derived discover why EDX is the name to know
models, and full 3D ray-tracing models in wreless system planning tools. Tools for Wireless Design