Académique Documents
Professionnel Documents
Culture Documents
In the previous three lessons, we discussed the Fourier Series, which is for periodic signals. This lesson will cover the Fourier Transform which can be used to analyze aperiodic signals. (Later on, we'll see how we can also use it for periodic signals.) The Fourier Transform is another method for representing signals and systems in the frequency domain
On
the -axis, distance between two consecutive aks is now 0=2/T, the fundamental frequency.
-T
-T1 0 T1
t T
k =0 k 0
-20 -0
0 0 20
On the -axis, distance between two consecutive aks is now 0=2/T, the fundamental frequency.
-T1 0 T1
This means the Fourier Transform can represent an aperiodic signal on the frequency-domain.
ak =
1 x(t )e jk 0t dt T X ( j ) =
x(t ) = lim x p (t )
T
ak =
1 X ( j 0 ) T
x(t )e
j t
dt
Since xp(t) is periodic, we can write Fourier series representation for it: T /2
x p (t ) =
k =
a e
k
jk 0 t
1 ak = x p (t )e jk 0t dt T T/ 2
x p (t ) =
where 0=2/T.
k =
T X ( j
)e jk 0t
x p (t ) =
1 2
k =
X ( j
)e jk 0t 0
As T, this sum will approach to an integral because there will a continuum of Fourier terms in the Fourier series. Finally, the Fourier Transform of x(t):
X ( j ) =
x (t ) = 1 2
x(t )e
j t
dt
d
X ( j ) e
j t
x(t ) dt <
Over a finite interval of time, the signal must have finite number of maxima and minima (or variations) Over a finite interval of time, the signal must have finite number of discontinuities. Also, those discontinuities must be finite.
Linearity
Because the Fourier Transform is linear, we can write: F[a x1(t) + bx2(t)] = aX1() + bX2()
Time Scaling
where X1() is the Fourier Transform of x1(t) and X2() is the Fourier Transform of x2(t).
Time Shifting
Duality
In general, the Duality property is very useful because it can enable to solve Fourier Transforms that would be difficult to compute directly (such as taking the Fourier Transform of a sinc function). The Duality Property tells us that if x(t) has a Fourier Transform X(), then if we form a new function of time that has the functional form of the transform, X(t), it will have a Fourier Transform x() that has the functional form of the original time function (but is a function of frequency). Mathematically, we can write:
Example 1 Using the Fourier Transform integral equation, directly find the Fourier Transform of x(t) = e-at u(t), a> 0
Notice that the second term in the last line is simply the Fourier Transform integral of the function X(t), i.e.
Example 2 Using the results of Example 1 and the Duality Property, find the Fourier Transform of
The convolution property states that: Let us show that. To start, let
Then we can take the Fourier Transform of y(t) and plug in the convolution integral for y(t) (notice how we've marked the integrals with dt and d to keep track of them):
Therefore,
We've just shown that the Fourier Transform of the convolution of two functions is simply the product of the Fourier Transforms of the functions. This means that for linear, time-invariant systems, where the input/output relationship is described by a convolution, you can avoid convolution by using Fourier Transforms. This is a very powerful result.
Multiplication of Signals
It states that the Fourier Transform of the product of two signals in time is the convolution of the two Fourier Transforms.
Proof
Example 2 Find the Fourier Transform of x(t) = sinc2(t) (Hint: use the Multiplication Property).
Therefore,
This tells us that modulation (such as multiplication in time by a complex exponential, cosine wave, or sine wave) corresponds to a frequency shift in the frequency domain.
DC Level Example Find the Fourier Transform of the constant 1. Use duality and the fact that the transform of (t) is 1
Differentiation in time
Transform of derivatives
Suppose that f(n) is piecewise continuous, and absolutely integrable on R. Then
We know ejt2(-0) =0
e
F { f ( n ) ( t )} = ( iw ) n F ( w )
In particular
F { f ' ( t )} = iwF ( w )
and
F { f '' ( t )} = w 2 F ( w )
but we can not directly calculate this integral because it does not converge. For many periodic signals, the integral for Fourier transform may not work. Luckily, there is an easier way of determining Fourier transforms of periodic signals So we will generalize the Fourier Transform to include impulses in the frequency domain. We can use either the Duality or Modulation Properties to show that. So we will generalize the Fourier Transform to include impulses in the frequency domain. We can use either the Duality or Modulation Properties to show that: We can check this by taking Inverse Fourier Transform (and using the sifting property)
Now let's consider a general periodic signal x(t) that we can represent as a Fourier Series:
Pulsed Cosine
Example Find the Fourier Transform of cos(0t)
Fourier Transforms of Sampled Signals In this lesson, we will discuss sampling of continuous time signals. Sampling a continuous time signal is used, for example, in A/D conversion, such as would be done in digitizing music for storage on a CD, digitizing a movie for storage on a DVD, or taking a digital picture.
Now, x(t) is the continuous time signal we wish to sample. We will model sampling as multiplying a signal by p(t Let xs(t) = x(t)p(t) be the sampled signal. Then
As you just saw, p(t) is an infinite train of continuous time impulse functions, spaced Ts seconds apart.
Now, we will derive the Sampling Theorem . To do this, we will examine our signals in the frequency domain. To start, let p(t) have a Fourier Transform P(), x(t) have a Fourier Transform X(), and xs(t) have a Fourier Transform Xs(). Then, because xs(t) = x(t)p(t), by the Multiplication Property,
Now let's find the Fourier Transform of p(t). Because the infinite impulse train is periodic, we will use the Fourier Transform of periodic signals:
Let's find the Fourier Series coefficients Ck for the periodic impulse train p(t):
Therefore
Now, to finish our derivation of the Sampling Theorem, we will go back and determine Xs(). We saw that: Therefore,
From this development and observing the above figure, we can derive our Sampling Theorem. This is one of the most important results.
We can recover x(t) from its sampled version xs (t) by using a low pass filter to recover the center island:
As you can see from the figure if 0 - c < c, we would get overlap of the replicates of X() in frequency. This is known as "aliasing." Therefore, to avoid aliasing, we require 0 - c > c or 0 > 2c. If we avoid aliasing, we can recover x(t) from its samples. (Usually, we choose a sampling rate a bit higher than twice the highest frequency since filters are not ideal.) We hear music up to 20 kHz and CD sampling rate is 44.1 kHz. (Dogs would need a higher quality CD since they hear higher frequencies than humans.)
Aliasing (Under-sampling)
What happens when sampling frequency is less than Nyquist Frequency, i.e. s<2M ? The original signal x(t) cannot be recovered from xs(t) since there will be unwanted overlaps in Xs().
The second condition is also known as Nyquist Criterion. s is referred as Nyquist Frequency, i.e. the smallest possible sampling frequency in order to recover the original analog signal from its samples.
This will cause the recovered signal xr(t) to be different than x(t), also called as aliasing or under-sampling.
Example 1 Given a signal x(t) with Fourier Transform with cutoff frequency ωc as shown:
Draw the sampled spectrum in each case. Which case(s) experiences aliasing?
Example 2 The inverse Fourier Transform of the signal in the previous example is
Draw the sampled signals using the sampling trains of the previous example
Example 1 Given two functions x1(t) = cos(500t) and x2(t) = cos(1000t), form a new signal x3(t) = x1(t) x2(t). Draw the frequency response of a low pass filter that passes the low frequency component of the signal and blocks the high frequency component of the signal.
Sinusoidal Amplitude Modulation Modulation is multiplying a signal in time by complex exponentials (such as sine waves and cosine waves) to shift the signal to a desired frequency band. This is done, for example, by radio (or television) to assign different parts of the frequency spectrum to different radio (or television) stations. Remember that the Modulation Theorem told us that: We will use the Modulation Theorem in an example of AM radio.
Your radio demodulates the signal by multiplying the received signal y(t) by another cosine wave to form a third signal z(t):
To recover the original music signal x(t) from the demodulated signal z(t), you filter z(t) with a low-pass filter by multiplying by a rect function in frequency. This corresponds to convolving with a sinc function in time.