Handout 5: Course Notes Were Prepared by Dr. R.M.A.P. Rajatheva and Revised by Dr. Poompat Saengudomlert
Handout 5: Course Notes Were Prepared by Dr. R.M.A.P. Rajatheva and Revised by Dr. Poompat Saengudomlert
Handout 5: Course Notes Were Prepared by Dr. R.M.A.P. Rajatheva and Revised by Dr. Poompat Saengudomlert
1
P∞
From S(f ) = T k=−∞ δ(f − kfs ) and Vs (f ) = V (f ) ∗ S(f ),
∞
1 X
Vs (f ) = V (f − kfs )
T k=−∞
NOTE: The Fourier transform of the sampled signal vs (t) consists of periodically repeated
copies at V (f ) equally spaced apart by fs , as illustrated in figure 1.23.
1
Course notes were prepared by Dr. R.M.A.P. Rajatheva and revised by Dr. Poompat Saengudomlert.
1
Conversion from
impulse train
to sequence
0 0
0 1 2 0 1 2
Figure 1.22: Sampling process
2
Aliasing and Nyquist rate
Let W be the bandwidth of v(t), i.e. W is the smallest number such that V (f ) = 0 for
|f | > W . From figure 1.23, the following facts are evident.
1. When fs > 2W , the copies of V (f ) do not overlap. Therefore, when they are added
together, there remains a copy of V (f ) around each integer multiple of fs .
2. When fs ≤ 2W , the copies of V (f ) overlap. Therefore, when they are added
together, V (f ) may not be recoverable. This overlapping phenomenon is referred
to as aliasing.
NOTE: The maximum signal frequency W is referred to as the Nyquist frequency. The
frequency 2W that must be exceeded by the sampling frequency is called the Nyquist
rate.
Figure 1.24 illustrates aliasing in the time domain. Note that there are two sinusoidal
signals with frequencies f1 = fs /8 and f2 = 7fs /8. These two signals have exactly the
same samples.
1.5
fs=8,f1=1,f2=7,v1(t)=sin(2π t),v2(t)=-sin(14π t)
0.5
-0.5
-1
-1.5
0 0.2 0.4 0.6 0.8 1
t
Figure 1.24: Simple illustration of aliasing in the time domain.
where Ω denotes the digital angular frequency expressed in radians. The corresponding
discrete-time inverse Fourier transform is defined as
Z π
1
v[n] = V (ejΩ )ejΩn dΩ
2π −π
3
A discrete-time Fourier transform is always periodic in Ω with period 2π. To see this,
note that we can write for any nonnegative integer k,
∞
X ∞
X
j(Ω+2πk) jΩ
V (e )= v[n]e −j(Ω+2πk)n
= {z } = V (e ).
v[n]e−jΩn e|−j2πkn
n=−∞ n=−∞ =1
If v[n] is sampled from v(t) with sampling period T , i.e. v[n] = v(nT ), we can relate
V (ejΩ ) and V (f ) as follows. Recall that the sampled signal can be expressed using the
unit impulse as
∞
X
vs (t) = v(t)s(t) = v(nT )δ(t − nT )
n=−∞
Alternatively, we can use the Fourier transform pair δ(t − nT ) ↔ e−j2πf nT to write
∞
X
Vs (f ) = v(nT )e−j2πf nT .
n=−∞
By setting
Ω = 2πf T
we can write
∞
jΩ 1 X Ω
V (e ) = V − kfs
T k=−∞ 2πT
NOTE: V (ejΩ ) is the frequency-scaled version of Vs (f ) with the frequency scaling specified
by Ω = 2πf T . This frequency scaling can be seen as a normalization of the frequency
axis so that the analog frequency ω = 2πfs (in rad/s) in V (f ) is normalized to the digital
frequency Ω = 2π (in rad) for V (ejΩ ).
4
1.2 1.2
T=1
T 1
1
0.8
0.8
V(f)=Trect(fT)
v(t)=sinc(t/T)
0.6
0.6
0.4
0.4
0.2
0.2
0
0 -0.2
-1/2T 1/2T
-0.2 -0.4
T
-1.5 -1 -0.5 0 0.5 1 1.5 -6 -4 -2 0 2 4 6
f t
Figure 1.25: Low pass filter and its Fourier transform.
If we reconstruct v(t) from the discrete-time signal v[n], then we can write
∞
X t
v(t) = v[n]sinc −n
n=−∞
T
The overall process of reconstructing v(t) from v[n] is illustrated in figure 1.26.
Figure 1.27 illustrates how the reconstruction of v(t) can be viewed as interpolating
v[n] by the sinc functions. In summary, the overall discussion in this section can be
summarized in the following sampling theorem.
Sampling theorem: Suppose that v(t) is bandlimited such that V (f ) = 0 for |f | > fN .
If v[n] is obtained from sampling v(t) with sampling rate fs = 1/T , i.e. v[n] = v(nT ),
such that fs > 2fN , then v(t) is uniquely defined by its samples v[n] as follows.
∞
X t
v(t) = v[n]sinc −n
n=−∞
T
5
1.2
1
0.8
0.6
v[n]
0.4
0.2
0
-0.2
-0.4
-4 -3 -2 -1 0 1 2 3 4 5
n
v(t)=Σn v[n]sinc(t/T-n)
1.2
1
0.8
0.6
0.4
0.2
0
-0.2
-0.4
-4 -3 -2 -1 0 1 2 3 4 5
t
Figure 1.27: Reconstruction by interpolating using sinc pulses.