5.6: Discrete Time Fourier Transform (DTFT)
 Page ID
 1625
Learning Objectives
 Discussion of Discretetime Fourier Transforms.
 Topics include comparison with analog transforms and discussion of Parseval's theorem.
The Fourier transform of the discretetime signal is defined to be
\[S(e^{i2\pi f})=\sum_{n=\infty }^{\infty }s(n)e^{(i2\pi fn)}\]
Frequency here has no units. As should be expected, this definition is linear, with the transform of a sum of signals equaling the sum of their transforms. Realvalued signals have conjugatesymmetric spectra:
\[S(e^{(i2\pi f)})=\overline{S(e^{j2\pi f}})\]
Exercise \(\PageIndex{1}\)
A special property of the discretetime Fourier transform is that it is periodic with period one:
\[S(e^{i2\pi (f+1)})=S(e^{i2\pi f})\]
Derive this property from the definition of the DTFT.
Solution
\[S(e^{i2\pi (f+1)})=\sum_{n=\infty }^{\infty }s(n)e^{(i2\pi (f+1)n)}\]
\[S(e^{i2\pi (f+1)})=\sum_{n=\infty }^{\infty }e^{(i2\pi n)}s(n)e^{(i2\pi fn)}\]
\[S(e^{i2\pi (f+1)})=\sum_{n=\infty }^{\infty }s(n)e^{(i2\pi fn)}\]
\[S(e^{i2\pi (f+1)})=S(e^{i2\pi f})\]
Because of this periodicity, we need only plot the spectrum over one period to understand completely the spectrum's structure; typically, we plot the spectrum over the frequency range
\[\left [ \frac{1}{2},\frac{1}{2} \right ]\]
When the signal is realvalued, we can further simplify our plotting chores by showing the spectrum only over
\[\left [ 0,\frac{1}{2} \right ]\]
the spectrum at negative frequencies can be derived from positivefrequency spectral values.
When we obtain the discretetime signal via sampling an analog signal, the Nyquist frequency corresponds to the discretetime frequency To show this, note that a sinusoid having a frequency equal to the Nyquist frequency has a sampled waveform that equals
\[\cos \left ( 2\pi \times \frac{1}{2T_{s}}nT_{s} \right )=\cos (\pi n)=(1)^{n}\]
The exponential in the DTFT at frequency
\[e^{\frac{i2\pi n}{2}}=e^{(i\pi n)}=(1)^{n}\]
meaning that discretetime frequency equals analog frequency multiplied by the sampling interval
\[f_{D}=f_{A}T_{s}\]
f_{D} and f_{A} represent discretetime and analog frequency variables, respectively. The aliasing figure provides another way of deriving this result. As the duration of each pulse in the periodic sampling signal p_{Ts}(t) narrows, the amplitudes of the signal's spectral repetitions, which are governed by the Fourier series coefficients of p_{Ts}(t), become increasingly equal. Examination of the periodic pulse signal reveals that as Δ decreases, the value of the largest Fourier coefficient, decreases to zero:
\[\left  c_{0} \right =\frac{A\Delta }{T_{s}}\]
Thus, to maintain a mathematically viable Sampling Theorem, the amplitude must increase as Δ, becoming infinitely large as the pulse duration decreases. Practical systems use a small value of Δ, say 0.1.and use amplifiers to rescale the signal. Thus, the sampled signal's spectrum becomes periodic with period Thus, the Nyquist frequency corresponds to the frequency
Example \(\PageIndex{1}\):
Let's compute the discretetime Fourier transform of the exponentially decaying sequence
\[s(n)=a^{n}u(n)\]
where is the unitstep sequence. Simply plugging the signal's expression into the Fourier transform formula:
\[S(e^{i2\pi f})=\sum_{n=\infty }^{\infty }a^{n}u(n)e^{(i2\pi fn)}\]
\[S(e^{i2\pi f})=\sum_{n=\infty }^{\infty }\left ( ae^{(i2\pi fn)}\right )^{n}\]
This sum is a special case of the geometric series.
\[\sum_{n=0}^{\infty }=\forall \alpha ,\left  \alpha \right < 1:\left ( \frac{1}{1\alpha } \right )\]
Thus, as long as a<1, we have our Fourier transform.
\[S(e^{i2\pi f})=\frac{1}{1ae^{(i2\pi f)}}\]
Using Euler's relation, we can express the magnitude and phase of this spectrum.
\[\left  S(e^{i2\pi f})\right =\frac{1}{\sqrt{\left ( 1a\cos (2\pi f) \right )^{2}+a^{2}\sin ^{2}(2\pi f)}}\]
\[\angle \left ( S(e^{i2\pi f})\right )=\tan^{1}\left ( \frac{a\sin 2\pi f}{1a\cos (2\pi f)} \right )\]
No matter what value of we choose, the above formulae clearly demonstrate the periodic nature of the spectra of discretetime signals. Fig. 5.6.1 below shows indeed that the spectrum is a periodic function. We need only consider the spectrum between ½to unambiguously define it. When we have a lowpass spectrum—the spectrum diminishes as frequency increases from 0 to ½ —with increasing a leading to a greater low frequency content; for we have a highpass spectrum as shown in Fig. 5.6.2 below.
Fig. 5.6.1 The spectrum of the exponential signal (a = 0.5) is shown over the frequency range [2,2], clearly demonstrating the periodicity of all discretetime spectra. The angle has units of degrees.
Fig. 5.6.1 The spectra of several exponential signals are shown. What is the apparent relationship between the spectra for
Example \(\PageIndex{1}\):
Analogous to the analog pulse signal, let's find the spectrum of the length N pulse sequence.
\[s(n)=\begin{cases} 1 & \text{ if } 0\leq n\leq N1 \\ 0 & \text{ if } otherwise \end{cases}\]
The Fourier transform of this sequence has the form of a truncated geometric series.
\[S(e^{i2\pi f})=\sum_{n=0}^{N1}e^{(i2\pi fn)}\]
For the socalled finite geometric series, we know that
\[\sum_{n=n_{0}}^{N+n_{0}1}\alpha ^{n}=\alpha ^{n_{0}}\frac{1\alpha ^{N}}{1\alpha }\]
for all values of α.
Exercise \(\PageIndex{1}\)
Derive this formula for the finite geometric series sum. The "trick" is to consider the difference between the series' sum and the sum of the series multiplied by α.
Solution
\[\alpha \sum_{n=n_{0}}^{N+n_{0}1}\alpha ^{n}\sum_{n=n_{0}}^{N+n_{0}1}\alpha ^{n}=\alpha ^{N+n_{0}}\alpha ^{n_{0}}\]
which, after manipulation, yields the geometric sum formula.
Applying this result yields to Fig. 5.6.3,
\[S(e^{i2\pi f})=\frac{1e^{(i2\pi fN)}}{1e^{(i2\pi f)}}\]
\[S(e^{i2\pi f})=e^{(i\pi f(N1))}\frac{\sin (\pi fN)}{\sin (\pi f)}\]
The ratio of sine functions has the generic form of
\[\frac{\sin (Nx)}{\sin (x)}\]
which is known as the discretetime sinc function . Thus, our transform can be concisely expressed as
\[S(e^{i2\pi f})=e^{(i\pi f(N1))}dsinc(\pi f)\]
The discretetime pulse's spectrum contains many ripples, the number of which increase with the pulse's duration.
Fig. 5.6.3 The spectrum of a lengthten pulse is shown. Can you explain the rather complicated appearance of the phase?
The inverse discretetime Fourier transform is easily derived from the following relationship:
\[\int_{\frac{1}{2}}^{\frac{1}{2}} e^{(i2\pi fm)} e^{i2\pi fn}df=\begin{cases} 1 & \text{ if } m=n \\ 0 & \text{ if } m\neq n \end{cases}\]
\[\int_{\frac{1}{2}}^{\frac{1}{2}} e^{(i2\pi fm)} e^{i2\pi fn}df=\delta (mn)\]
Therefore, we find that
\[\int_{\frac{1}{2}}^{\frac{1}{2}} S(e^{i2\pi f})e^{i2\pi fn}df=\int_{\frac{1}{2}}^{\frac{1}{2}} \sum_{_{m}m}s(m)e^{(i2\pi fm)}e^{i2\pi fn}df\]
\[\int_{\frac{1}{2}}^{\frac{1}{2}} S(e^{i2\pi f})e^{i2\pi fn}df=\sum_{_{m}m}s(m)\int_{\frac{1}{2}}^{\frac{1}{2}} e^{((i2\pi f))(mn)}df\]
\[\int_{\frac{1}{2}}^{\frac{1}{2}} S(e^{i2\pi f})e^{i2\pi fn}df=s(n)\]
The Fourier transform pairs in discretetime are
\[S(e^{i2\pi f})=\sum_{n=\infty }^{\infty }s(n)e^{(i2\pi fn)}\]
\[s(n)=\int_{\frac{1}{2}}^{\frac{1}{2}} S(e^{i2\pi f})e^{i2\pi fn}df\]
The properties of the discretetime Fourier transform mirror those of the analog Fourier transform. The DTFT properties table below shows similarities and differences. One important common property is Parseval's Theorem.
Fig. 5.6.4 DTFT Properties
\[\sum_{n=\infty }^{\infty }\left ( \left  s(n) \right  \right )^{2}=\int_{\frac{1}{2}}^{\frac{1}{2}}\left  S(e^{i2\pi f}) \right ^{2}df\]
To show this important property, we simply substitute the Fourier transform expression into the frequencydomain expression for power.
\[\int_{\frac{1}{2}}^{\frac{1}{2}}\left  S(e^{i2\pi f}) \right ^{2}df=\int_{\frac{1}{2}}^{\frac{1}{2}}\sum_{nn}s(n)e^{(i2\pi fn)}\sum_{mm}\overline{s(n)}e^{i2\pi fm}df\]
\[\int_{\frac{1}{2}}^{\frac{1}{2}}\left  S(e^{i2\pi f}) \right ^{2}df=\sum_{n,m,n,m}s(n)\overline{s(n)}\int_{\frac{1}{2}}^{\frac{1}{2}}e^{i2\pi f(mn)}df\]
Using the orthogonality relation, the integral equals δ(mn), where δ(n) is the unit sample. Thus, the double sum collapses into a single sum because nonzero values occur only when n=m, giving Parseval's Theorem as a result. We term
\[\sum_{nn}s^{2}(n)\]
the energy in the discretetime signal in spite of the fact that discretetime signals don't consume (or produce for that matter) energy. This terminology is a carryover from the analog world.
Exercise \(\PageIndex{1}\)
Suppose we obtained our discretetime signal from values of the product s(t)p_{Ts}(t), where the duration of the component pulses in p_{Ts}(t) is Δ. How is the discretetime signal energy related to the total energy contained in s(t)? Assume the signal is bandlimited and that the sampling rate was chosen appropriate to the Sampling Theorem's conditions.
If the sampling frequency exceeds the Nyquist frequency, the spectrum of the samples equals the analog spectrum, but over the normalized analog frequency fT. Thus, the energy in the sampled signal equals the original signal's energy multiplied by
Contributor

ContribEEOpenStax