Difference between revisions of "Signal Representation/Discrete Fourier Transform (DFT)"

From LNTwww
 
(65 intermediate revisions by 5 users not shown)
Line 1: Line 1:
 
   
 
   
 
{{Header
 
{{Header
|Untermenü=Zeit- und frequenzdiskrete Signaldarstellung
+
|Untermenü=Time and Frequency-Discrete Signal Representation
|Vorherige Seite=Zeitdiskrete Signaldarstellung
+
|Vorherige Seite=Discrete-Time Signal Representation
|Nächste Seite=Fehlermöglichkeiten bei Anwendung der DFT
+
|Nächste Seite=Possible Errors When Using DFT
 
}}
 
}}
  
==Argumente für die diskrete Realisierung der FT==
+
==Arguments for the discrete implementation of the Fourier transform==
 +
<br>
 +
The&nbsp; &raquo;'''Fourier transform'''&laquo;&nbsp; according to the previous description in chapter&nbsp; [[Signal_Representation/The_Fourier_Transform_and_its_Inverse#The_first_Fourier_integral|&raquo;Aperiodic Signals &ndash; Pulses&laquo;]]&nbsp; has an infinitely high selectivity due to the unlimited extension of the integration interval and is therefore an ideal theoretical tool of spectral analysis.
  
Die '''Fouriertransformation''' gemäß der bisherigen Beschreibung im Kapitel [[Signaldarstellung/Fouriertransformation_und_-rücktransformation|Aperiodische Signale - Impulse]] weist aufgrund der unbegrenzten Ausdehnung des Integrationsintervalls eine unendlich hohe Selektivität auf und ist deshalb ein ideales theoretisches Hilfsmittel der Spektralanalyse.
+
If the spectral components&nbsp; $X(f)$&nbsp; of a time function&nbsp; $x(t)$&nbsp; are to be determined numerically,&nbsp; the general transformation equations
Sollen die Spektralanteile $X(f)$ einer Zeitfunktion $x(t)$ numerisch ermittelt werden, so sind die allgemeinen Transformationsgleichungen
 
 
   
 
   
$$\begin{align*}X(f) & =  \int_{-\infty
+
:$$\begin{align*}X(f) & =  \int_{-\infty
  }^{+\infty}x(t) \cdot {\rm e}^{-{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi f t}\hspace{0.1cm} {\rm d}t\hspace{0.5cm} \Rightarrow\hspace{0.5cm} {\boldsymbol {\rm Hintransformation}}
+
  }^{+\infty}x(t) \cdot {\rm e}^{-{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi f t}\hspace{0.1cm} {\rm d}t\hspace{0.5cm} \Rightarrow\hspace{0.5cm} \text{Transform}\hspace{0.7cm} \Rightarrow\hspace{0.5cm} \text{first  Fourier integral}
 
  \hspace{0.05cm},\\
 
  \hspace{0.05cm},\\
 
x(t) & =  \int_{-\infty
 
x(t) & =  \int_{-\infty
 
  }^{+\infty}\hspace{-0.15cm}X(f) \cdot {\rm e}^{\hspace{0.05cm}+{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi f t}\hspace{0.1cm} {\rm d}f\hspace{0.35cm} \Rightarrow\hspace{0.5cm}
 
  }^{+\infty}\hspace{-0.15cm}X(f) \cdot {\rm e}^{\hspace{0.05cm}+{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi f t}\hspace{0.1cm} {\rm d}f\hspace{0.35cm} \Rightarrow\hspace{0.5cm}
{\boldsymbol {\rm R\ddot{u}cktransformation}}
+
\text{Inverse Transform}\hspace{0.4cm} \Rightarrow\hspace{0.5cm} \text{second Fourier integral}
 
  \hspace{0.05cm}\end{align*}$$
 
  \hspace{0.05cm}\end{align*}$$
  
aus zwei Gründen ungeeignet:
+
are unsuitable for two reasons:
*Die Gleichungen gelten ausschließlich für zeitkontinuierliche Signale. Mit Digitalrechnern oder Signalprozessoren können jedoch nur zeitdiskrete Signale verarbeitet werden.
+
#The equations apply exclusively to continuous-time signals.&nbsp; With digital computers or signal processors,&nbsp; however,&nbsp; one can only process discrete-time signals.
*Für eine numerische Auswertung der beiden Fourierintegrale ist es erforderlich, das jeweilige Integrationsintervall auf einen endlichen Wert zu begrenzen.
+
#For a numerical evaluation of the two Fourier integrals it is necessary to limit the respective integration interval to a finite value.
  
  
Daraus ergibt sich folgende Konsequenz: Ein kontinuierliches Signal muss vor der numerischen Bestimmung seiner Spektraleigenschaften zwei Prozesse durchlaufen, nämlich
+
{{BlaueBox|TEXT=
*den der '''Abtastung''' zur Diskretisierung, und
+
$\text{This leads to the following consequence:}$&nbsp;
*den der '''Fensterung''' zur Begrenzung des Integrationsintervalls.
 
  
 +
A&nbsp; &raquo;'''continuous-valued signal'''&laquo;&nbsp; must undergo two processes before the numerical determination of its spectral properties,&nbsp; viz.
 +
#that of&nbsp; &raquo;'''sampling'''&laquo;&nbsp; for discretization,&nbsp; and<br>
 +
#that of&nbsp; &raquo;'''windowing'''&laquo;&nbsp; to limit the integration interval.}}
  
Im Folgenden wird ausgehend von einer aperiodischen Zeitfunktion $x(t)$ und dem dazugehörigen Fourierspektrum $X(f)$ eine für die Rechnerverarbeitung geeignete zeit– und frequenzdiskrete Beschreibung schrittweise entwickelt.
 
  
 +
In the following,&nbsp; starting from an aperiodic time function&nbsp; $x(t)$&nbsp; and the corresponding Fourier spectrum&nbsp; $X(f)$&nbsp; a time and frequency-discrete description suitable for computer processing is developed step by step.
  
==Zeitdiskretisierung - Periodifizierung im Frequenzbereich==
 
  
Die folgenden Grafiken zeigen einheitlich links den Zeitbereich und rechts den Frequenzbereich. Ohne Einschränkung der Allgemeingültigkeit sind $x(t)$ und $X(f)$ jeweils reell und gaußförmig.
 
  
[[File:P_ID1132__Sig_T_5_1_S2_neu.png|Diskretisierung im Zeitbereich – Periodifizierung im Frequenzbereich]]
+
==Time discretization &ndash; Periodification in the frequency domain==
 +
<br>
 +
The following graphs show uniformly the time domain on the left and the frequency domain on the right.&nbsp;
 +
*Without limiting generality,&nbsp; $x(t)$&nbsp; and&nbsp; $X(f)$&nbsp; are each real and Gaussian.
  
Entsprechend dem Kapitel [[Signaldarstellung/Zeitdiskrete_Signaldarstellung|Zeitdiskrete Signaldarstellung]] kann man die Abtastung des Zeitsignals $x(t)$ durch die Multiplikation mit einem Diracpuls $p_{\delta}(t)$ beschreiben. Es ergibt sich das im Abstand $T_{\rm A}$ abgetastete Zeitsignal
+
*According to the chapter&nbsp; [[Signal_Representation/Time_Discrete_Signal_Representation|&raquo;Discrete-Time Signal Representation&laquo;]]&nbsp; one can describe the sampling of the time signal&nbsp; $x(t)$&nbsp; by multiplying it by a Dirac delta train &nbsp; &rArr; &nbsp; <br>&raquo;Dirac comb&nbsp; in the time domain&laquo;
 +
[[File:P_ID1132__Sig_T_5_1_S2_neu.png|right|frame| Time discretization &nbsp; &rArr; &nbsp; Periodification in the frequency domain.&nbsp; <u>Notation:</u><br> &nbsp; &nbsp; ${\rm A}\{x(t)\}$:&nbsp; Signal&nbsp; $x(t)$&nbsp; after&nbsp; &raquo;sampling&laquo;&nbsp; $($German:&nbsp; "Abtastung"$)$ &nbsp; &rArr; &nbsp; $\rm A\{\text{...}\}$ <br> &nbsp; &nbsp; ${\rm P}\{X(f)\}$:&nbsp; Spectrum&nbsp; $X(f)$&nbsp; after&nbsp; &raquo;periodification&laquo;&nbsp;  &nbsp; &rArr; &nbsp; $\rm P\{\text{...}\}$]]
 +
 
 +
:$$p_{\delta}(t) =  \sum_{\nu = - \infty }^{+\infty} T_{\rm A} \cdot
 +
\delta (t- \nu \cdot T_{\rm A}
 +
)\hspace{0.05cm}.$$
 +
 
 +
The result is the time signal sampled at a distance&nbsp; $T_{\rm A}$:&nbsp;
 
   
 
   
$${\rm A}\{x(t)\} =  \sum_{\nu = - \infty }^{+\infty} T_{\rm A} \cdot x(\nu \cdot T_{\rm A})\cdot
+
:$${\rm A}\{x(t)\} =  \sum_{\nu = - \infty }^{+\infty} T_{\rm A} \cdot x(\nu \cdot T_{\rm A})\cdot
 
  \delta (t- \nu \cdot T_{\rm A}
 
  \delta (t- \nu \cdot T_{\rm A}
 
  )\hspace{0.05cm}.$$
 
  )\hspace{0.05cm}.$$
  
Das abgetastete Signal $\text{A}\{ x(t)\}$ transformieren wir nun in den Frequenzbereich. Der Multiplikation des Diracpulses $p_{\delta}(t)$ mit $x(t)$ entspricht im Frequenzbereich die Faltung von $P_{\delta}(f)$ mit $X(f)$. Es ergibt sich das periodifizierte Spektrum $\text{P}\{ X(f)\}$, wobei $f_{\rm P}$ die Frequenzperiode der Funktion $\text{P}\{ X(f)\}$ angibt:
+
We transform this sampled signal&nbsp; $\text{A}\{ x(t)\}$&nbsp; into the frequency domain:
 +
*The multiplication of the Dirac comb&nbsp; $p_{\delta}(t)$&nbsp; with&nbsp; $x(t)$&nbsp; corresponds in the frequency domain to the convolution of&nbsp; $P_{\delta}(f)$&nbsp; with&nbsp; $X(f)$.
 +
 
 +
*The result is the periodified spectrum&nbsp; $\text{P}\{ X(f)\}$,&nbsp; where&nbsp; $f_{\rm
 +
P}$&nbsp; indicates the frequency period of the function&nbsp; $\text{P}\{ X(f)\}$&nbsp;:
 
   
 
   
$${\rm A}\{x(t)\} \hspace{0.2cm}\circ\!\!-\!\!\!-\!\!\!-\!\!\bullet\, \hspace{0.2cm} {\rm P}\{X(f)\} =  \sum_{\mu = - \infty }^{+\infty}
+
:$${\rm A}\{x(t)\} \hspace{0.2cm}\circ\!\!-\!\!\!-\!\!\!-\!\!\bullet\, \hspace{0.2cm} {\rm P}\{X(f)\} =  \sum_{\mu = - \infty }^{+\infty}
  X (f- \mu \cdot f_{\rm P} )\hspace{0.5cm} {\rm mit }\hspace{0.5cm}f_{\rm
+
  X (f- \mu \cdot f_{\rm P} ).$$
P}= {1}/{T_{\rm A}}\hspace{0.05cm}.$$
+
 
 +
This relation was already derived in the chapter&nbsp; [[Signal_Representation/Time_Discrete_Signal_Representation|"Discrete-Time Signal Representation"]]&nbsp;  but with slightly different nomenclature:
 +
*We now denote the sampled signal by&nbsp; $\text{A}\{ x(t)\}$&nbsp; instead of&nbsp; $x_{\rm A}(t)$.
  
Dieser Zusammenhang wurde ebenfalls bereits im Kapitel [[Signaldarstellung/Zeitdiskrete_Signaldarstellung|Zeitdiskrete Signaldarstellung]]  hergeleitet, jedoch mit etwas anderer Nomenklatur. Diese Nomenklaturänderung wird auf den nachfolgenden Seiten begründet:
+
* The&nbsp; &raquo;frequency period&laquo;&nbsp; is now denoted by&nbsp; $f_{\rm P} = 1/T_{\rm A}$&nbsp; instead of&nbsp; $f_{\rm A} = 1/T_{\rm A}$.
*Das abgetastete Signal $\text{A}\{ x(t)\}$ anstelle von $x_{\rm A}(t)$.
 
* Die '''Frequenzperiode''' $f_{\rm P}$ = $1/T_{\rm A}$ anstelle von $f_{\rm A} = 1/T_{\rm A}$.
 
  
  
Die Grafik zeigt den hier beschriebenen Funktionalzusammenhang. Hierzu ist anzumerken:
+
These nomenclature changes are justified in the following sections.
*Die Frequenzperiode $f_{\rm P}$ wurde hier aus Darstellungsgründen bewusst klein gewählt, so dass die Überlappung der zu summierenden Spektren deutlich zu erkennen ist.
 
*In der Praxis sollte $f_{\rm P}$ aufgrund des Abtasttheorems mindestens doppelt so groß sein wie die größte im Signal $x(t)$ enthaltene Frequenz.
 
*Ist dies nicht erfüllt, so muss mit '''Aliasing''' gerechnet werden – siehe Kapitel [[Signaldarstellung/Fehlermöglichkeiten_bei_Anwendung_der_DFT|Fehlermöglichkeiten bei Anwendung der DFT]].
 
  
 +
The graph above shows the functional relationship described here.&nbsp; It should be noted:
 +
#The frequency period&nbsp; $f_{\rm P}$&nbsp; has been deliberately chosen to be small here so that the overlap of the spectra to be summed can be clearly seen.
 +
#In practice&nbsp; $f_{\rm P}$&nbsp; should be at least twice as large as the largest frequency contained in the signal&nbsp; $x(t)$&nbsp; due to the sampling theorem.
 +
#If this is not fulfilled,&nbsp; then&nbsp; &raquo;'''aliasing'''&laquo;&nbsp; must be expected - see chapter&nbsp; [[Signal_Representation/Possible_Errors_When_Using_DFT|&raquo;Possible Errors when using DFT&laquo;]].
  
==Frequenzdiskretisierung - Periodifizierung im Zeitbereich==
 
  
Die Diskretisierung von $X(f)$ lässt sich ebenfalls durch eine Multiplikation mit einem Diracpuls beschreiben. Es ergibt sich das im Abstand $f_{\rm A}$ abgetastete Spektrum:
+
==Frequency discretization &ndash; Periodification in the time domain ==
+
<br>
$${\rm A}\{X(f)\} =  X(f) \cdot  \sum_{\mu = - \infty }^{+\infty}
+
The discretization of&nbsp; $X(f)$&nbsp; can also be described by a multiplication with a Dirac comb in the frequency domain.&nbsp; The result is the sampled spectrum with distance&nbsp; $f_{\rm A}$:  
 +
:$${\rm A}\{X(f)\} =  X(f) \cdot  \sum_{\mu = - \infty }^{+\infty}
 
  f_{\rm A} \cdot \delta (f- \mu \cdot f_{\rm A } ) =  \sum_{\mu = - \infty }^{+\infty}
 
  f_{\rm A} \cdot \delta (f- \mu \cdot f_{\rm A } ) =  \sum_{\mu = - \infty }^{+\infty}
 
  f_{\rm A} \cdot X(\mu \cdot f_{\rm A } ) \cdot\delta (f- \mu \cdot f_{\rm A } )\hspace{0.05cm}.$$
 
  f_{\rm A} \cdot X(\mu \cdot f_{\rm A } ) \cdot\delta (f- \mu \cdot f_{\rm A } )\hspace{0.05cm}.$$
  
Transformiert man den hier verwendeten Frequenz–Diracpuls (mit Impulsgewichten $f_{\rm A}$) in den Zeitbereich, so erhält man mit $T_{\rm P} = 1/f_{\rm A}$:
+
*If one transforms the Dirac comb $($with impulse weights&nbsp; $f_{\rm A})$&nbsp; into the time domain,&nbsp; one obtains with&nbsp; $T_{\rm P} = 1/f_{\rm A}$:
 
   
 
   
$$\sum_{\mu = - \infty }^{+\infty}
+
:$$\sum_{\mu = - \infty }^{+\infty}
 
  f_{\rm A} \cdot \delta (f- \mu \cdot f_{\rm A } ) \hspace{0.2cm}\bullet\!\!-\!\!\!-\!\!\!-\!\!\circ\, \hspace{0.2cm}
 
  f_{\rm A} \cdot \delta (f- \mu \cdot f_{\rm A } ) \hspace{0.2cm}\bullet\!\!-\!\!\!-\!\!\!-\!\!\circ\, \hspace{0.2cm}
 
  \sum_{\nu = - \infty }^{+\infty}
 
  \sum_{\nu = - \infty }^{+\infty}
 
   \delta (t- \nu \cdot T_{\rm P } ) \hspace{0.05cm}.$$
 
   \delta (t- \nu \cdot T_{\rm P } ) \hspace{0.05cm}.$$
  
Die Multiplikation mit $X(f)$ entspricht im Zeitbereich der Faltung mit $x(t)$. Man erhält das im Abstand $T_{\rm P}$ periodifizierte Signal $\text{P}\{ x(t)\}$:
+
*The multiplication with&nbsp; $X(f)$&nbsp; corresponds in the time domain to the convolution with&nbsp; $x(t)$.&nbsp; One obtains the signal&nbsp; $\text{P}\{ x(t)\}$&nbsp; periodified with distance&nbsp; $T_{\rm P}$:
 
   
 
   
$${\rm A}\{X(f)\} \hspace{0.2cm}\bullet\!\!-\!\!\!-\!\!\!-\!\!\circ\, \hspace{0.2cm}
+
:$${\rm A}\{X(f)\} \hspace{0.2cm}\bullet\!\!-\!\!\!-\!\!\!-\!\!\circ\, \hspace{0.2cm}
 
  {\rm P}\{x(t)\} = x(t) \star \sum_{\nu = - \infty }^{+\infty}
 
  {\rm P}\{x(t)\} = x(t) \star \sum_{\nu = - \infty }^{+\infty}
 
   \delta (t- \nu \cdot T_{\rm P } )= \sum_{\nu = - \infty }^{+\infty}
 
   \delta (t- \nu \cdot T_{\rm P } )= \sum_{\nu = - \infty }^{+\infty}
 
   x (t- \nu \cdot T_{\rm P } ) \hspace{0.05cm}.$$
 
   x (t- \nu \cdot T_{\rm P } ) \hspace{0.05cm}.$$
  
Dieser Zusammenhang ist in der Grafik veranschaulicht. Aufgrund der groben Frequenzrasterung ergibt sich in diesem Beispiel für die Zeitperiode $T_{\rm P}$ ein relativ kleiner Wert, so dass sich das periodifizierte Zeitsignal $\text{P}\{ x(t)\}$ aufgrund von Überlappungen deutlich von $x(t)$ unterscheidet.
 
  
[[File:P_ID1134__Sig_T_5_1_S3_neu.png|Diskretisierung im Frequenzbereich – Periodifizierung im Zeitbereich]]
+
{{GraueBox|TEXT=
 +
[[File:P_ID1134__Sig_T_5_1_S3_neu.png|right|frame|Frequency discretization &ndash; Periodization in the time domain]]
 +
$\text{Example 1:}$&nbsp;
 +
This correlation is illustrated in the graph:
 +
*Due to the coarse frequency rastering,&nbsp; this example results in a relatively small value for the time period&nbsp; $T_{\rm P}$&nbsp;.
 +
 
 +
 
 +
* Therefore due to overlaps,&nbsp; the&nbsp; $($blue$)$&nbsp; periodified time signal&nbsp; $\text{P}\{ x(t)\}$&nbsp; differs significantly from&nbsp; $x(t)$.
 +
 
  
 +
*If one wants to achieve&nbsp; $\text{P}\{ x(t)\} \approx x(t)$&nbsp; then&nbsp; $T_{\rm P}$&nbsp; must be chosen much larger than in this example.}}
  
==Finite Signaldarstellung==
+
==Finite signal representation==
 +
<br>
 +
[[File:P_ID1135__Sig_T_5_1_S4_neu.png|right|frame|Finite signals of the Discrete Fourier Transform&nbsp; $\rm (DFT)$]]
 +
One arrives at the so-called&nbsp; &raquo;finite signal representation&laquo;,&nbsp; if both
 +
*the time function&nbsp; $x(t)$,&nbsp;
 +
 +
*the spectral function&nbsp; $X(f)$
  
Zur '''finiten Signaldarstellung''' kommt man, wenn sowohl die Zeitfunktion $x(t)$ wie auch die Spektralfunktion $X(f)$ ausschließlich durch ihre Abtastwerte angegeben werden.
 
  
[[File:P_ID1135__Sig_T_5_1_S4_neu.png|Finite Signale der DFT]]
+
are specified exclusively by their sample values.&nbsp; This graph is to be interpreted as follows:
 +
*In the left graph,&nbsp; the function&nbsp; $\text{A}\{ \text{P}\{ x(t)\}\}$&nbsp; is drawn in blue.&nbsp; This results from sampling the periodified time function&nbsp; $\text{P}\{ x(t)\}$&nbsp; with equidistant Dirac deltas with distance&nbsp; $T_{\rm A} = 1/f_{\rm P}$.
  
Diese Grafik ist wie folgt zu interpretieren:
+
*In the right graph,&nbsp; the function&nbsp; $\text{P}\{ \text{A}\{ X(f)\}\}$&nbsp; is drawn in green.&nbsp; This results from the periodification&nbsp; $($with&nbsp; $f_{\rm P})$&nbsp; of the sampled spectral function&nbsp; $\{ \text{A}\{ X(f)\}\}$.
*Im linken Bild blau eingezeichnet ist die Funktion $\text{A}\{ \text{P}\{ x(t)\}\}$. Diese ergibt sich durch Abtastung der periodifizierten Zeitfunktion $\text{P}\{ x(t)\}$ mit äquidistanten Diracimpulsen im Abstand $T_{\rm A} = 1/f_{\rm P}$.
+
*Im rechten Bild grün eingezeichnet ist die Funktion $\text{P}\{ \text{A}\{ X(f)\}\}$. Diese ergibt sich durch Periodifizierung (mit $f_{\rm P}$) der abgetasteten Spektralfunktion $\{ \text{A}\{ X(f)\}\}$.
+
*There is a Fourier correspondence between the blue finite signal&nbsp; $($in the left sketch$)$&nbsp; and the green finite signal&nbsp; $($in the right sketch$)$,&nbsp; as follows:
*Zwischen dem blauen finiten Signal und dem grünen finiten Signal besteht eine Fourierkorrespondenz:
 
 
   
 
   
 
:$${\rm A}\{{\rm P}\{x(t)\}\} \hspace{0.2cm}\circ\!\!-\!\!\!-\!\!\!-\!\!\bullet\, \hspace{0.2cm} {\rm P}\{{\rm A}\{X(f)\}\} \hspace{0.05cm}.$$
 
:$${\rm A}\{{\rm P}\{x(t)\}\} \hspace{0.2cm}\circ\!\!-\!\!\!-\!\!\!-\!\!\bullet\, \hspace{0.2cm} {\rm P}\{{\rm A}\{X(f)\}\} \hspace{0.05cm}.$$
  
*Die Diraclinien der periodischen Fortsetzung $\text{P}\{ \text{A}\{ X(f)\}\}$ der abgetasteten Spektralfunktion fallen allerdings nur dann in das gleiche Frequenzraster wie diejenigen von $\text{A}\{ X(f)\}$, wenn die Frequenzperiode $f_{\rm P}$ ein ganzzahliges Vielfaches ($N$) des Frequenzabtastabstandes $f_{\rm A}$ ist.
+
*The Dirac delta lines of the periodic continuation&nbsp; $\text{P}\{ \text{A}\{ X(f)\}\}$&nbsp; of the sampled spectral function,&nbsp; however,&nbsp; only fall into the same frequency grid as those of&nbsp; $\text{A}\{ X(f)\}$&nbsp; if the frequency period&nbsp; $f_{\rm P}$&nbsp; is an integer multiple&nbsp; $(N)$&nbsp; of the frequency sampling interval&nbsp; $f_{\rm A}$.
*Deshalb muss bei Anwendung der finiten Signaldarstellung stets gelten, wobei für die natürliche Zahl $N$ in der Praxis meist eine Zweierpotenz verwendet wird (der obigen Grafik liegt der Wert $N = 8$ zugrunde):
+
 
 +
*Therefore,&nbsp; when using the finite signal representation,&nbsp; the following condition must always be fulfilled,&nbsp; where in practice the natural number&nbsp; $N$&nbsp; is usually a power of two&nbsp; $($the above graph is based on the value&nbsp; $N = 8)$:
 
   
 
   
 
:$$f_{\rm P} = N \cdot f_{\rm A} \hspace{0.5cm} \Rightarrow\hspace{0.5cm} {1}/{T_{\rm A}}= N \cdot f_{\rm A} \hspace{0.5cm} \Rightarrow\hspace{0.5cm}
 
:$$f_{\rm P} = N \cdot f_{\rm A} \hspace{0.5cm} \Rightarrow\hspace{0.5cm} {1}/{T_{\rm A}}= N \cdot f_{\rm A} \hspace{0.5cm} \Rightarrow\hspace{0.5cm}
 
  N \cdot f_{\rm A}\cdot T_{\rm A} = 1\hspace{0.05cm}.$$
 
  N \cdot f_{\rm A}\cdot T_{\rm A} = 1\hspace{0.05cm}.$$
  
*Bei Einhaltung der Bedingung $N \cdot f_{\rm A} \cdot T_{\rm A} = 1$ ist die Reihenfolge von Periodifizierung und Abtastung vertauschbar. Somit gilt:
+
*If the condition&nbsp; $N \cdot f_{\rm A} \cdot T_{\rm A} = 1$&nbsp; is fulfilled then the order of periodization and sampling is interchangeable.&nbsp; Thus:
 
   
 
   
 
:$${\rm A}\{{\rm P}\{x(t)\}\} = {\rm P}\{{\rm A}\{x(t)\}\}\hspace{0.2cm}\circ\!\!-\!\!\!-\!\!\!-\!\!\bullet\, \hspace{0.2cm}
 
:$${\rm A}\{{\rm P}\{x(t)\}\} = {\rm P}\{{\rm A}\{x(t)\}\}\hspace{0.2cm}\circ\!\!-\!\!\!-\!\!\!-\!\!\bullet\, \hspace{0.2cm}
 
  {\rm P}\{{\rm A}\{X(f)\}\} = {\rm A}\{{\rm P}\{X(f)\}\}\hspace{0.05cm}.$$
 
  {\rm P}\{{\rm A}\{X(f)\}\} = {\rm A}\{{\rm P}\{X(f)\}\}\hspace{0.05cm}.$$
  
Die Zeitfunktion $\text{P}\{ \text{A}\{ x(t)\}\}$} besitzt die Periode $T_{\rm P} = N \cdot T_{\rm A}$ und die Periode im Frequenzbereich ist $f_{\rm P} = N \cdot f_{\rm A}$. Zur Beschreibung des diskretisierten Zeit– und Frequenzverlaufs reichen somit jeweils $N$ '''komplexe Zahlenwerte''' in Form von Impulsgewichten aus.
+
{{BlaueBox|TEXT=
 +
$\text{Conclusions:}$&nbsp;
 +
#The time function&nbsp; $\text{P}\{ \text{A}\{ x(t)\}\}$&nbsp; has the period&nbsp; $T_{\rm P} = N \cdot T_{\rm A}$.
 +
#The period in the frequency domain is&nbsp; $f_{\rm P} = N \cdot f_{\rm A}$.  
 +
#For the description of the discretized time and frequency response in each case&nbsp; $N$&nbsp; '''complex numerical values'''&nbsp; in the form of impulse weights are sufficient.}}
  
{{Beispiel}}
 
Es liegt ein impulsartiges Signal $x(t)$ in abgetasteter Form vor, wobei  der Abstand zweier Abtastwerte $T_{\rm A} = 1\,\text{μs}$ beträgt.
 
Nach einer diskreten Fouriertransformation mit $N$ = 512 liegt das Spektrum $X(f)$ in Form von Abtastwerten im Abstand $f_{\rm A} = (N \cdot T_{\rm A})^{–1} \approx 1.953\,\text{kHz} $ vor. Vergrößert man $N$ auf 2048, so ergibt sich ein feineres Frequenzraster mit $f_{\rm A} \approx 488\,\text{Hz}$.
 
  
{{end}}
+
{{GraueBox|TEXT=
 +
$\text{Example 2:}$&nbsp;
 +
A time-limited&nbsp; $($pulse-like$)$&nbsp; signal&nbsp; $x(t)$&nbsp; is present in sampled form,&nbsp; where the distance between two samples is&nbsp; $T_{\rm A} = 1\, {\rm &micro; s}$:
 +
*After a discrete Fourier transform with&nbsp; $N = 512$&nbsp; the spectrum&nbsp; $X(f)$&nbsp; is available in form of frequency-samples at spacing&nbsp; $f_{\rm A} = (N \cdot T_{\rm A})^{-1} \approx 1.953\,\text{kHz} $.
  
 +
*Increasing the DFT parameter to&nbsp; $N= 2048$ results in a&nbsp; $($four times$)$&nbsp; finer frequency grid with&nbsp; $f_{\rm A} \approx 488\,\text{Hz}$.}}
  
==Von der kontinuierlichen zur diskreten Fouriertransformation==
 
  
Aus dem herkömmlichen [[Signaldarstellung/Fouriertransformation_und_-rücktransformation#Das_erste_Fourierintegral|ersten Fourierintegral]]
+
==From the continuous to the discrete Fourier transform==
 +
<br>
 +
From the conventional&nbsp; [[Signal_Representation/Fourier_Transform_and_its_Inverse#The_first_Fourier_integral|&raquo;first Fourier integral&laquo;]]
 
   
 
   
$$X(f) =\int_{-\infty
+
:$$X(f) =\int_{-\infty
  }^{+\infty}x(t) \cdot {\rm e}^{-{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi f  \hspace{0.05cm}t}\hspace{0.1cm} {\rm d}t$$
+
  }^{+\infty}x(t) \cdot {\rm e}^{-{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi \hspace{0.05cm}\cdot \hspace{0.05cm} \hspace{0.05cm}\cdot \hspace{0.05cm}t}\hspace{0.1cm} {\rm d}t$$
  
entsteht durch Diskretisierung ($\text{d}t \to T_{\rm A}$, $t \to \nu \cdot T_{\rm A}$, $f \to \mu \cdot f_{\rm A}$, $T_{\rm A} \cdot f_{\rm A} = 1/N$) die Summe
+
arises from discretization&nbsp; $(\text{d}t \to T_{\rm A}$,&nbsp; $t \to \nu \cdot T_{\rm A}$,&nbsp; $f \to \mu \cdot f_{\rm A}$,&nbsp; $T_{\rm A} \cdot f_{\rm A} = 1/N)$&nbsp; the sampled and periodified spectral function
 
   
 
   
$${\rm P}\{X(\mu \cdot f_{\rm A})\} = T_{\rm A} \cdot \sum_{\nu = 0 }^{N-1}
+
:$${\rm P}\{X(\mu \cdot f_{\rm A})\} = T_{\rm A} \cdot \sum_{\nu = 0 }^{N-1}
 
   {\rm P}\{x(\nu \cdot T_{\rm A})\}\cdot  {\rm e}^{-{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi \hspace{0.05cm} \cdot \hspace{0.05cm}\nu \hspace{0.05cm}
 
   {\rm P}\{x(\nu \cdot T_{\rm A})\}\cdot  {\rm e}^{-{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi \hspace{0.05cm} \cdot \hspace{0.05cm}\nu \hspace{0.05cm}
 
  \cdot \hspace{0.05cm}\mu /N} \hspace{0.05cm}.$$
 
  \cdot \hspace{0.05cm}\mu /N} \hspace{0.05cm}.$$
  
Es ist berücksichtigt, dass aufgrund der Diskretisierung jeweils die periodifizierten Funktionen einzusetzen sind. Aus Gründen einer vereinfachten Schreibweise nehmen wir nun die folgenden Substitutionen vor:
+
It is taken into account that due to the discretization,&nbsp; the periodified functions are to be used in each case.  
*Die $N$ '''Zeitbereichskoeffizienten''' seien mit der Laufvariablen $\nu$ = 0, ... , $N - 1$:
+
 
 +
For reasons of simplified notation, we now make the following substitutions:
 +
*The&nbsp; $N$&nbsp; &raquo;time-domain coefficients&laquo;&nbsp; are with the variable&nbsp; $\nu = 0$, ... , $N - 1$:
 
:$$d(\nu) =
 
:$$d(\nu) =
 
   {\rm P}\left\{x(t)\right\}{\big|}_{t \hspace{0.05cm}= \hspace{0.05cm}\nu \hspace{0.05cm}\cdot \hspace{0.05cm}T_{\rm A}}\hspace{0.05cm}.$$
 
   {\rm P}\left\{x(t)\right\}{\big|}_{t \hspace{0.05cm}= \hspace{0.05cm}\nu \hspace{0.05cm}\cdot \hspace{0.05cm}T_{\rm A}}\hspace{0.05cm}.$$
*Die $N$ '''Frequenzbereichskoeffizienten''' seien mit der Laufvariablen $\mu$ = 0, ... , $N$ – 1:
+
*Let&nbsp; $N$&nbsp; &raquo;frequency domain coefficients&laquo;&nbsp; be associated with the variable&nbsp; $\mu = 0,$ ... , $N-1$:
 
:$$D(\mu) = f_{\rm A} \cdot
 
:$$D(\mu) = f_{\rm A} \cdot
 
   {\rm P}\left\{X(f)\right\}{\big|}_{f \hspace{0.05cm}= \hspace{0.05cm}\mu \hspace{0.05cm}\cdot \hspace{0.05cm}f_{\rm A}}\hspace{0.05cm}.$$
 
   {\rm P}\left\{X(f)\right\}{\big|}_{f \hspace{0.05cm}= \hspace{0.05cm}\mu \hspace{0.05cm}\cdot \hspace{0.05cm}f_{\rm A}}\hspace{0.05cm}.$$
*Abkürzend wird für den von $N$ abhängigen  '''komplexen Drehfaktor'''  geschrieben:
+
*Abbreviation is written for the from&nbsp; $N$&nbsp; dependent&nbsp; &raquo;complex rotation factor&laquo;&nbsp;:
:$$w = {\rm e}^{-{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi /N}
+
:$$w = {\rm e}^{-{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi /N}
  = \cos \left( {2 \pi}/{N}\right)-{\rm j} \cdot \sin \left( {2 \pi}/{N}\right)
+
  = \cos \left( {2 \pi}/{N}\right)-{\rm j} \cdot \sin \left( {2 \pi}/{N}\right)
 
  \hspace{0.05cm}.$$  
 
  \hspace{0.05cm}.$$  
  
 +
{{BlaueBox|TEXT=
 +
$\text{Definition:}$&nbsp;
 +
The term&nbsp; &raquo;Discrete Fourier Transform&laquo;&nbsp; $\rm  (DFT)$&nbsp; means the calculation of the&nbsp; $N$&nbsp; spectral coefficients&nbsp; $D(\mu)$&nbsp; from the&nbsp; $N$&nbsp; signal coefficients&nbsp; $d(\nu)$:
 +
[[File:P_ID2730__Sig_T_5_1_S5_neu.png|right|frame|On defining the&nbsp; &raquo;Discrete Fourier Transform&laquo;&nbsp; $\rm (DFT)$&nbsp; with&nbsp; $N=8$]]
 +
:$$D(\mu) = \frac{1}{N} \cdot \sum_{\nu = 0 }^{N-1}
 +
  d(\nu)\cdot  {w}^{\hspace{0.05cm}\nu \hspace{0.07cm} \cdot \hspace{0.05cm}\mu} \hspace{0.05cm}. $$
 +
:
 +
In the diagram you can see:
 +
*The&nbsp; $N = 8$&nbsp; signal coefficients&nbsp; $d(\nu)$&nbsp; by the blue filling,
  
{{Definition}}
+
*the&nbsp; $N = 8$&nbsp; spectral coefficients&nbsp; $D(\mu)$&nbsp; at the green filling.}}
Die Gleichung der '''Diskreten Fouriertransformation''' (kurz DFT) lautet:
 
 
$$D(\mu) = \frac{1}{N} \cdot \sum_{\nu = 0 }^{N-1}
 
  d(\nu)\cdot  {w}^{\hspace{0.05cm}\nu \hspace{0.03cm} \cdot \hspace{0.05cm}\mu} \hspace{0.05cm}. $$
 
 
 
{{end}}
 
 
 
Oder in Worten: Unter dem Begriff ''Diskrete Fouriertransformation'' versteht man die Berechnung der $N$ Spektralkoeffizienten $D(\mu)$ aus den $N$ Signalkoeffizienten $d(\nu)$.
 
  
[[File:P_ID2730__Sig_T_5_1_S5_neu.png|Zur Definition der DFT]]
 
  
In der Grafik erkennt man die $N$ = 8 Signalkoeffizienten $d(\nu)$ an der blauen Füllung und die $N$ = 8 Spektralkoeffizienten $D(\mu)$ an der grünen Füllung.
+
==Inverse discrete Fourier transform==
 +
<br>
 +
The&nbsp; &raquo;inverse discrete Fourier transform&laquo;&nbsp;  describes the&nbsp;  [[Signal_Representation/Fourier_Transform_and_its_Inverse#The_second_Fourier_integral|&raquo;second Fourier integral&laquo;]]:
 
   
 
   
 
+
:$$\begin{align*}x(t) & =  \int_{-\infty
==Inverse Diskrete Fouriertransformation==
+
  }^{+\infty}X(f) \cdot {\rm e}^{\hspace{0.05cm}{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi \hspace{0.05cm}\cdot \hspace{0.05cm} f \hspace{0.05cm}\cdot \hspace{0.05cm}
 
 
Die ''Inverse Diskrete Fouriertransformation'' (IDFT) beschreibt das zweite Fourierintegral
 
 
$$\begin{align*}x(t) & =  \int_{-\infty
 
  }^{+\infty}X(f) \cdot {\rm e}^{\hspace{0.05cm}{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi f \hspace{0.05cm}
 
 
  t}\hspace{0.1cm} {\rm d}f\end{align*}$$
 
  t}\hspace{0.1cm} {\rm d}f\end{align*}$$
  
in diskretisierter Form. Man erhält mit dem Übergang $\text{d}f \to f_A$:
+
in discretized form: &nbsp;
+
:$$d(\nu) =
$$d(\nu) =
 
 
   {\rm P}\left\{x(t)\right\}{\big|}_{t \hspace{0.05cm}= \hspace{0.05cm}\nu \hspace{0.05cm}\cdot \hspace{0.05cm}T_{\rm
 
   {\rm P}\left\{x(t)\right\}{\big|}_{t \hspace{0.05cm}= \hspace{0.05cm}\nu \hspace{0.05cm}\cdot \hspace{0.05cm}T_{\rm
   A}}\hspace{0.05cm},\hspace{1.55cm} \nu = 0, ... \hspace{0.05cm},
+
   A}}\hspace{0.01cm}.$$
  N-1\hspace{0.05cm},$$
+
 
 +
{{BlaueBox|TEXT= $\text{Definition:}$&nbsp; The term&nbsp; &raquo;Inverse Discrete Fourier Transform&raquo; &nbsp; $\rm (IDFT)$&nbsp;&nbsp; means the calculation of the signal coefficients&nbsp; $d(\nu)$&nbsp; from the spectral coefficients&nbsp; $D(\mu)$:
 +
[[File:P_ID2731__Sig_T_5_1_S6_neu.png|right|frame|On defining the&nbsp; &raquo;Inverse Discrete Fourier Transform&laquo;&nbsp; $\rm (IDFT)$&nbsp; with&nbsp; $N=8$]]
 +
:$$d(\nu) = \sum_{\mu = 0 }^{N-1}
 +
D(\mu) \cdot {w}^{-\nu \hspace{0.07cm} \cdot \hspace{0.05cm}\mu} \hspace{0.05cm}.$$
  
{{Definition}}
+
*With indices&nbsp; $\nu = 0, \hspace{0.05cm}\text{...} \hspace{0.05cm}, N-1, $&nbsp; and &nbsp; $\mu = 0, \hspace{0.05cm}\text{...} \hspace{0.05cm}, N-1$&nbsp; holds:
Die Gleichung der '''Inversen Diskreten Fouriertransformation''' (kurz IDFT) lautet:
+
:$$d(\nu) =
 +
  {\rm P}\left\{x(t)\right\}{\big \vert}_{t \hspace{0.05cm}= \hspace{0.05cm}\nu \hspace{0.05cm}\cdot \hspace{0.05cm}T_{\rm
 +
  A} }\hspace{0.01cm},$$
 
   
 
   
$$d(\nu) = \sum_{\mu = 0 }^{N-1}
+
:$$D(\mu) = f_{\rm A} \cdot
D(\mu) \cdot  {w}^{-\nu \hspace{0.03cm} \cdot \hspace{0.05cm}\mu} \hspace{0.05cm}.$$
+
  {\rm P}\left\{X(f)\right\}{\big \vert}_{f \hspace{0.05cm}= \hspace{0.05cm}\mu \hspace{0.05cm}\cdot \hspace{0.05cm}f_{\rm A} }
{{end}}
+
  \hspace{0.01cm},$$
  
 +
:$$w  = {\rm e}^{- {\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi /N}
 +
\hspace{0.01cm}.$$
  
[[File:P_ID2731__Sig_T_5_1_S6_neu.png|Zur Definition der IDFT]]
+
*A comparison between&nbsp; [[Signal_Representation/Discrete_Fourier_Transform_(DFT)#From_the_continuous_to_the_discrete_Fourier_transform|&raquo;DFT&laquo;]]&nbsp; and&nbsp; &raquo;IDFT&laquo;&nbsp; shows that exactly the same algorithm can be used.&nbsp;
  
In anderen Worten: Unter dem Begriff ''Inverse Diskrete Fouriertransformation'' versteht man die Berechnung der Signalkoeffizienten $d(ν)$ aus den Spektralkoeffizienten $D(\mu)$.
+
*The only differences between IDFT and DFT are:
 +
#The exponent of the rotation factor is to be applied with different sign.
 +
#In the IDFT,&nbsp; the division by&nbsp; $N$&nbsp; is omitted.}}
  
Es gelten auch hier die Definitionen:
 
 
$$d(\nu) =
 
  {\rm P}\left\{x(t)\right\}{\big|}_{t \hspace{0.05cm}= \hspace{0.05cm}\nu \hspace{0.05cm}\cdot \hspace{0.05cm}T_{\rm
 
  A}}\hspace{0.05cm},\hspace{1.55cm} \nu = 0, ... \hspace{0.05cm},
 
  N-1\hspace{0.05cm},$$
 
 
$$D(\mu) = f_{\rm A} \cdot
 
  {\rm P}\left\{X(f)\right\}{\big|}_{f \hspace{0.05cm}= \hspace{0.05cm}\mu \hspace{0.05cm}\cdot \hspace{0.05cm}f_{\rm A}}
 
  \hspace{0.05cm},\hspace{0.3cm} \mu = 0, ... \hspace{0.05cm},
 
  N-1\hspace{0.05cm},$$
 
  
$$w  = {\rm e}^{-{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi /N}
+
==Interpretation of DFT and IDFT==
\hspace{0.05cm}.$$
+
<br>
+
The graph shows the discrete coefficients in the time and frequency domain together with the periodified continuous-time functions.
Ein Vergleich zwischen der DFT und IDFT zeigt, dass genau der gleiche Algorithmus verwendet werden kann. Die einzigen Unterschiede der IDFT gegenüber der DFT sind:
 
*Der Exponent des Drehfaktors ist mit unterschiedlichem Vorzeichen anzusetzen.
 
*Bei der IDFT entfällt die Division durch $N$.
 
  
 +
[[File:P_ID1136__Sig_T_5_1_S7_neu.png|right|frame|Time and frequency coefficients of the DFT]]
  
==Interpretation von DFT und IDFT==
+
When using DFT or IDFT,&nbsp; please note:
 +
#According to the above definitions, the DFT coefficients&nbsp; $d(ν)$&nbsp; and&nbsp; $D(\mu)$&nbsp; always have the unit of the time function.
 +
#Dividing&nbsp; $D(\mu)$&nbsp; by&nbsp; $f_{\rm A}$,&nbsp; one obtains the spectral value&nbsp; $X(\mu \cdot f_{\rm A})$.
 +
#The spectral coefficients&nbsp; $D(\mu)$&nbsp; must always be complex in order to be able to consider odd time functions.
 +
#One also uses complex time coefficients&nbsp; $d(\nu)$ &nbsp; &rArr; &nbsp; DFT and IDFT are also applicable to band-pass signals.
 +
#The basic interval for&nbsp; $\nu$&nbsp; and&nbsp; $\mu$&nbsp; is usually defined as the range from&nbsp; $0$&nbsp; to&nbsp; $N - 1$&nbsp; $($filled circles in the graph$)$.
 +
<br clear=all>
 +
With the complex-valued number sequences&nbsp;
 +
:$$\langle \hspace{0.03cm}d(\nu)\hspace{0.03cm}\rangle = \langle \hspace{0.03cm}d(0), \hspace{0.05cm}\text{...} \hspace{0.05cm} , d(N-1) \hspace{0.03cm}\rangle,$$
 +
:$$\langle \hspace{0.03cm}D(\mu)\hspace{0.03cm}\rangle = \langle \hspace{0.03cm}D(0), \hspace{0.05cm}\text{...} \hspace{0.05cm} , D(N-1) \hspace{0.03cm}\rangle,$$
  
Die folgende Grafik zeigt nochmals die diskreten Koeffizienten im Zeit– und Frequenzbereich zusammen mit den periodifizierten zeitkontinuierlichen Funktionen.
+
DFT and IDFT are symbolized similarly to the conventional Fourier transform:
 +
:$$\langle \hspace{0.03cm} D(\mu)\hspace{0.03cm}\rangle \hspace{0.2cm}\bullet\!\!-\!\!\!-(N)\!-\!\!\!-\!\!\hspace{0.05cm}\circ\, \hspace{0.2cm} \langle \hspace{0.03cm} d(\nu) \hspace{0.03cm}\rangle \hspace{0.05cm}.$$
 +
#If the function&nbsp; $x(t)$&nbsp; is already limited to the range&nbsp; $0 \le t \lt N \cdot T_{\rm A}$&nbsp; then the IDFT output directly give the samples of the time function:  &nbsp;  $d(\nu) = x(\nu \cdot T_{\rm A}).$
 +
#If&nbsp; $x(t)$&nbsp; is shifted with respect to the basic interval,&nbsp; one has to choose the association shown in&nbsp; $\text{Example 3}$&nbsp; between&nbsp; $x(t)$&nbsp; and the coefficients&nbsp; $d(\nu)$.
  
[[File:P_ID1136__Sig_T_5_1_S7_neu.png|Zeit- und Frequenzbereichskoeffizienten der DFT]]
 
  
Bei Anwendung von DFT bzw. IDFT ist zu beachten:
+
{{GraueBox|TEXT=
*Nach obigen Definitionen besitzen die DFT–Koeffizienten $d(ν)$ und $D(\mu)$ stets die Einheit der Zeitfunktion. Dividiert man $D(\mu)$ durch $f_A$, so erhält man den Spektralwert $X(\mu \cdot f_A)$.
+
$\text{Example 3:}$&nbsp;
*Die Spektralkoeffizienten $D(\mu)$ müssen stets komplex angesetzt werden, um auch ungerade Zeitfunktionen berücksichtigen zu können.
+
The upper graph shows the asymmetric triangular pulse&nbsp; $x(t)$ whose absolute width is smaller than&nbsp; $T_{\rm P} = N \cdot T_{\rm A}$.  
*Aus Symmetriegründen verwendet man meist komplexe Zeitkoeffizienten $d(ν)$, um auch Bandpass–Signale im äquivalenten Tiefpassbereich transformieren zu können.
 
*Als Grundintervall für $ν$ $\mu$ definiert man meist – wie in obiger Grafik – den Bereich von 0 bis $N$ – 1. Mit den komplexwertigen Zahlenfolgen
 
$$\langle d(\nu)\rangle  = \langle d(0), ... , d(N-1) \rangle  \hspace{0.2cm}{\rm sowie}\hspace{0.2cm} \langle D(\mu)\rangle  =   \langle D(0), ... , D(N-1) \rangle$$
 
werden DFT und IDFT ähnlich wie die herkömmliche Fouriertransformation symbolisiert:
 
$$\langle D(\mu)\rangle \hspace{0.2cm}\bullet\!\!-\!\!\!-(N)\!-\!\!\!-\!\!\hspace{0.05cm}\circ\, \hspace{0.2cm} \langle d(\nu)\rangle  \hspace{0.05cm}.$$  
 
*Ist die Zeitfunktion $x(t)$ bereits auf den Bereich 0 ≤ $t$ < $N \cdot T_A$ begrenzt, dann geben die Zeitkoeffizienten direkt die Abtastwerte der Zeitfunktion an  ⇒  $d(ν) = x(ν \cdot T)$.
 
*Ist das Zeitsignal $x(t)$ gegenüber dem Grundintervall verschoben, so muss man die auf der nächsten Seite gezeigte Zuordnung zwischen $x(t)$ und den Koeffizienten $d(ν)$ wählen.
 
  
 +
[[File:EN_Sig_T_5_1_S7b_neu.png|right|frame|On assigning of the DFT coefficients with&nbsp; $N=8$]]
  
{{Beispiel}}
 
  
Die obere Grafik zeigt den unsymmetrischen Dreieckimpuls $x(t)$, dessen absolute Breite kleiner ist als $T_P = N \cdot T_A$.  
+
The sketch below shows the assigned DFT coefficients $($valid for&nbsp; $N = 8)$.
  
[[File:P_ID1139__Sig_T_5_1_S7b_neu.png|Zur Belegung der DFT-Koeffizienten]]
+
*For&nbsp; $\nu = 0,\hspace{0.05cm}\text{...} \hspace{0.05cm} , N/2 = 4$&nbsp; holds &nbsp; $d(\nu) = x(\nu \cdot T_{\rm A})$ :
  
Die untere Skizze zeigt die zugeordneten DFT–Koeffizienten für das Beispiel $N$ = 8.
+
:$$d(0) = x (0)\hspace{0.05cm}, \hspace{0.15cm}
 +
d(1) = x (T_{\rm A})\hspace{0.05cm}, \hspace{0.15cm}
 +
d(2) = x (2T_{\rm A})\hspace{0.05cm}, $$
 +
:$$d(3) = x (3T_{\rm A})\hspace{0.05cm}, \hspace{0.15cm}
 +
d(4) = x (4T_{\rm A})\hspace{0.05cm}.$$
 +
*The coefficients&nbsp; $d(5)$,&nbsp; $d(6)$&nbsp; and&nbsp; d$(7)$&nbsp; are to be set as follows:
  
Für die Zeitindizes $ν = 0$, ... , $N/2 = 4$ gilt $d(ν) = x(ν \cdot T_A)$.
+
:$$d(5) = x (-3T_{\rm A})\hspace{0.05cm}, \hspace{0.35cm}
+
d(6) = x (-2T_{\rm A})\hspace{0.05cm}, \hspace{0.35cm}
Dagegen sind die Koeffizienten $d$(5), $d$6) und $d$(7) wie folgt zu setzen:
+
d(7) = x (-T_{\rm A})\hspace{0.05cm}$$  
+
:$$ \Rightarrow \hspace{0.2cm}d(\nu) = x \big ((\nu\hspace{-0.05cm} - \hspace{-0.05cm} N ) \cdot T_{\rm  A}\big ).  $$
{{end}}
 
  
==Aufgaben zum Kapitel==
+
}}
  
[[Aufgaben:5.2 Inverse DFT|A5.2 Inverse DFT]]
+
==Exercises for the chapter==
 +
<br>
 +
[[Aufgaben:Exercise 5.2: Inverse Discrete Fourier Transform|Exercise 5.2: Inverse Discrete Fourier Transform]]
  
[[Aufgaben:5.2Z DFT eines Dreieckimpulses|Z5.2 DFT eines Dreieckimpulses]]
+
[[Aufgaben:Exercise 5.2Z: DFT of a Triangular Pulse|Exercise 5.2Z: DFT of a Triangular Pulse]]
  
 
      
 
      
 
{{Display}}
 
{{Display}}

Latest revision as of 16:57, 28 June 2023

Arguments for the discrete implementation of the Fourier transform


The  »Fourier transform«  according to the previous description in chapter  »Aperiodic Signals – Pulses«  has an infinitely high selectivity due to the unlimited extension of the integration interval and is therefore an ideal theoretical tool of spectral analysis.

If the spectral components  $X(f)$  of a time function  $x(t)$  are to be determined numerically,  the general transformation equations

$$\begin{align*}X(f) & = \int_{-\infty }^{+\infty}x(t) \cdot {\rm e}^{-{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi f t}\hspace{0.1cm} {\rm d}t\hspace{0.5cm} \Rightarrow\hspace{0.5cm} \text{Transform}\hspace{0.7cm} \Rightarrow\hspace{0.5cm} \text{first Fourier integral} \hspace{0.05cm},\\ x(t) & = \int_{-\infty }^{+\infty}\hspace{-0.15cm}X(f) \cdot {\rm e}^{\hspace{0.05cm}+{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi f t}\hspace{0.1cm} {\rm d}f\hspace{0.35cm} \Rightarrow\hspace{0.5cm} \text{Inverse Transform}\hspace{0.4cm} \Rightarrow\hspace{0.5cm} \text{second Fourier integral} \hspace{0.05cm}\end{align*}$$

are unsuitable for two reasons:

  1. The equations apply exclusively to continuous-time signals.  With digital computers or signal processors,  however,  one can only process discrete-time signals.
  2. For a numerical evaluation of the two Fourier integrals it is necessary to limit the respective integration interval to a finite value.


$\text{This leads to the following consequence:}$ 

A  »continuous-valued signal«  must undergo two processes before the numerical determination of its spectral properties,  viz.

  1. that of  »sampling«  for discretization,  and
  2. that of  »windowing«  to limit the integration interval.


In the following,  starting from an aperiodic time function  $x(t)$  and the corresponding Fourier spectrum  $X(f)$  a time and frequency-discrete description suitable for computer processing is developed step by step.


Time discretization – Periodification in the frequency domain


The following graphs show uniformly the time domain on the left and the frequency domain on the right. 

  • Without limiting generality,  $x(t)$  and  $X(f)$  are each real and Gaussian.
  • According to the chapter  »Discrete-Time Signal Representation«  one can describe the sampling of the time signal  $x(t)$  by multiplying it by a Dirac delta train   ⇒  
    »Dirac comb  in the time domain«
Time discretization   ⇒   Periodification in the frequency domain.  Notation:
    ${\rm A}\{x(t)\}$:  Signal  $x(t)$  after  »sampling«  $($German:  "Abtastung"$)$   ⇒   $\rm A\{\text{...}\}$
    ${\rm P}\{X(f)\}$:  Spectrum  $X(f)$  after  »periodification«    ⇒   $\rm P\{\text{...}\}$
$$p_{\delta}(t) = \sum_{\nu = - \infty }^{+\infty} T_{\rm A} \cdot \delta (t- \nu \cdot T_{\rm A} )\hspace{0.05cm}.$$

The result is the time signal sampled at a distance  $T_{\rm A}$: 

$${\rm A}\{x(t)\} = \sum_{\nu = - \infty }^{+\infty} T_{\rm A} \cdot x(\nu \cdot T_{\rm A})\cdot \delta (t- \nu \cdot T_{\rm A} )\hspace{0.05cm}.$$

We transform this sampled signal  $\text{A}\{ x(t)\}$  into the frequency domain:

  • The multiplication of the Dirac comb  $p_{\delta}(t)$  with  $x(t)$  corresponds in the frequency domain to the convolution of  $P_{\delta}(f)$  with  $X(f)$.
  • The result is the periodified spectrum  $\text{P}\{ X(f)\}$,  where  $f_{\rm P}$  indicates the frequency period of the function  $\text{P}\{ X(f)\}$ :
$${\rm A}\{x(t)\} \hspace{0.2cm}\circ\!\!-\!\!\!-\!\!\!-\!\!\bullet\, \hspace{0.2cm} {\rm P}\{X(f)\} = \sum_{\mu = - \infty }^{+\infty} X (f- \mu \cdot f_{\rm P} ).$$

This relation was already derived in the chapter  "Discrete-Time Signal Representation"  but with slightly different nomenclature:

  • We now denote the sampled signal by  $\text{A}\{ x(t)\}$  instead of  $x_{\rm A}(t)$.
  • The  »frequency period«  is now denoted by  $f_{\rm P} = 1/T_{\rm A}$  instead of  $f_{\rm A} = 1/T_{\rm A}$.


These nomenclature changes are justified in the following sections.

The graph above shows the functional relationship described here.  It should be noted:

  1. The frequency period  $f_{\rm P}$  has been deliberately chosen to be small here so that the overlap of the spectra to be summed can be clearly seen.
  2. In practice  $f_{\rm P}$  should be at least twice as large as the largest frequency contained in the signal  $x(t)$  due to the sampling theorem.
  3. If this is not fulfilled,  then  »aliasing«  must be expected - see chapter  »Possible Errors when using DFT«.


Frequency discretization – Periodification in the time domain


The discretization of  $X(f)$  can also be described by a multiplication with a Dirac comb in the frequency domain.  The result is the sampled spectrum with distance  $f_{\rm A}$:

$${\rm A}\{X(f)\} = X(f) \cdot \sum_{\mu = - \infty }^{+\infty} f_{\rm A} \cdot \delta (f- \mu \cdot f_{\rm A } ) = \sum_{\mu = - \infty }^{+\infty} f_{\rm A} \cdot X(\mu \cdot f_{\rm A } ) \cdot\delta (f- \mu \cdot f_{\rm A } )\hspace{0.05cm}.$$
  • If one transforms the Dirac comb $($with impulse weights  $f_{\rm A})$  into the time domain,  one obtains with  $T_{\rm P} = 1/f_{\rm A}$:
$$\sum_{\mu = - \infty }^{+\infty} f_{\rm A} \cdot \delta (f- \mu \cdot f_{\rm A } ) \hspace{0.2cm}\bullet\!\!-\!\!\!-\!\!\!-\!\!\circ\, \hspace{0.2cm} \sum_{\nu = - \infty }^{+\infty} \delta (t- \nu \cdot T_{\rm P } ) \hspace{0.05cm}.$$
  • The multiplication with  $X(f)$  corresponds in the time domain to the convolution with  $x(t)$.  One obtains the signal  $\text{P}\{ x(t)\}$  periodified with distance  $T_{\rm P}$:
$${\rm A}\{X(f)\} \hspace{0.2cm}\bullet\!\!-\!\!\!-\!\!\!-\!\!\circ\, \hspace{0.2cm} {\rm P}\{x(t)\} = x(t) \star \sum_{\nu = - \infty }^{+\infty} \delta (t- \nu \cdot T_{\rm P } )= \sum_{\nu = - \infty }^{+\infty} x (t- \nu \cdot T_{\rm P } ) \hspace{0.05cm}.$$


Frequency discretization – Periodization in the time domain

$\text{Example 1:}$  This correlation is illustrated in the graph:

  • Due to the coarse frequency rastering,  this example results in a relatively small value for the time period  $T_{\rm P}$ .


  • Therefore due to overlaps,  the  $($blue$)$  periodified time signal  $\text{P}\{ x(t)\}$  differs significantly from  $x(t)$.


  • If one wants to achieve  $\text{P}\{ x(t)\} \approx x(t)$  then  $T_{\rm P}$  must be chosen much larger than in this example.

Finite signal representation


Finite signals of the Discrete Fourier Transform  $\rm (DFT)$

One arrives at the so-called  »finite signal representation«,  if both

  • the time function  $x(t)$, 
  • the spectral function  $X(f)$


are specified exclusively by their sample values.  This graph is to be interpreted as follows:

  • In the left graph,  the function  $\text{A}\{ \text{P}\{ x(t)\}\}$  is drawn in blue.  This results from sampling the periodified time function  $\text{P}\{ x(t)\}$  with equidistant Dirac deltas with distance  $T_{\rm A} = 1/f_{\rm P}$.
  • In the right graph,  the function  $\text{P}\{ \text{A}\{ X(f)\}\}$  is drawn in green.  This results from the periodification  $($with  $f_{\rm P})$  of the sampled spectral function  $\{ \text{A}\{ X(f)\}\}$.
  • There is a Fourier correspondence between the blue finite signal  $($in the left sketch$)$  and the green finite signal  $($in the right sketch$)$,  as follows:
$${\rm A}\{{\rm P}\{x(t)\}\} \hspace{0.2cm}\circ\!\!-\!\!\!-\!\!\!-\!\!\bullet\, \hspace{0.2cm} {\rm P}\{{\rm A}\{X(f)\}\} \hspace{0.05cm}.$$
  • The Dirac delta lines of the periodic continuation  $\text{P}\{ \text{A}\{ X(f)\}\}$  of the sampled spectral function,  however,  only fall into the same frequency grid as those of  $\text{A}\{ X(f)\}$  if the frequency period  $f_{\rm P}$  is an integer multiple  $(N)$  of the frequency sampling interval  $f_{\rm A}$.
  • Therefore,  when using the finite signal representation,  the following condition must always be fulfilled,  where in practice the natural number  $N$  is usually a power of two  $($the above graph is based on the value  $N = 8)$:
$$f_{\rm P} = N \cdot f_{\rm A} \hspace{0.5cm} \Rightarrow\hspace{0.5cm} {1}/{T_{\rm A}}= N \cdot f_{\rm A} \hspace{0.5cm} \Rightarrow\hspace{0.5cm} N \cdot f_{\rm A}\cdot T_{\rm A} = 1\hspace{0.05cm}.$$
  • If the condition  $N \cdot f_{\rm A} \cdot T_{\rm A} = 1$  is fulfilled then the order of periodization and sampling is interchangeable.  Thus:
$${\rm A}\{{\rm P}\{x(t)\}\} = {\rm P}\{{\rm A}\{x(t)\}\}\hspace{0.2cm}\circ\!\!-\!\!\!-\!\!\!-\!\!\bullet\, \hspace{0.2cm} {\rm P}\{{\rm A}\{X(f)\}\} = {\rm A}\{{\rm P}\{X(f)\}\}\hspace{0.05cm}.$$

$\text{Conclusions:}$ 

  1. The time function  $\text{P}\{ \text{A}\{ x(t)\}\}$  has the period  $T_{\rm P} = N \cdot T_{\rm A}$.
  2. The period in the frequency domain is  $f_{\rm P} = N \cdot f_{\rm A}$.
  3. For the description of the discretized time and frequency response in each case  $N$  complex numerical values  in the form of impulse weights are sufficient.


$\text{Example 2:}$  A time-limited  $($pulse-like$)$  signal  $x(t)$  is present in sampled form,  where the distance between two samples is  $T_{\rm A} = 1\, {\rm µ s}$:

  • After a discrete Fourier transform with  $N = 512$  the spectrum  $X(f)$  is available in form of frequency-samples at spacing  $f_{\rm A} = (N \cdot T_{\rm A})^{-1} \approx 1.953\,\text{kHz} $.
  • Increasing the DFT parameter to  $N= 2048$ results in a  $($four times$)$  finer frequency grid with  $f_{\rm A} \approx 488\,\text{Hz}$.


From the continuous to the discrete Fourier transform


From the conventional  »first Fourier integral«

$$X(f) =\int_{-\infty }^{+\infty}x(t) \cdot {\rm e}^{-{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi \hspace{0.05cm}\cdot \hspace{0.05cm} f \hspace{0.05cm}\cdot \hspace{0.05cm}t}\hspace{0.1cm} {\rm d}t$$

arises from discretization  $(\text{d}t \to T_{\rm A}$,  $t \to \nu \cdot T_{\rm A}$,  $f \to \mu \cdot f_{\rm A}$,  $T_{\rm A} \cdot f_{\rm A} = 1/N)$  the sampled and periodified spectral function

$${\rm P}\{X(\mu \cdot f_{\rm A})\} = T_{\rm A} \cdot \sum_{\nu = 0 }^{N-1} {\rm P}\{x(\nu \cdot T_{\rm A})\}\cdot {\rm e}^{-{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi \hspace{0.05cm} \cdot \hspace{0.05cm}\nu \hspace{0.05cm} \cdot \hspace{0.05cm}\mu /N} \hspace{0.05cm}.$$

It is taken into account that due to the discretization,  the periodified functions are to be used in each case.

For reasons of simplified notation, we now make the following substitutions:

  • The  $N$  »time-domain coefficients«  are with the variable  $\nu = 0$, ... , $N - 1$:
$$d(\nu) = {\rm P}\left\{x(t)\right\}{\big|}_{t \hspace{0.05cm}= \hspace{0.05cm}\nu \hspace{0.05cm}\cdot \hspace{0.05cm}T_{\rm A}}\hspace{0.05cm}.$$
  • Let  $N$  »frequency domain coefficients«  be associated with the variable  $\mu = 0,$ ... , $N-1$:
$$D(\mu) = f_{\rm A} \cdot {\rm P}\left\{X(f)\right\}{\big|}_{f \hspace{0.05cm}= \hspace{0.05cm}\mu \hspace{0.05cm}\cdot \hspace{0.05cm}f_{\rm A}}\hspace{0.05cm}.$$
  • Abbreviation is written for the from  $N$  dependent  »complex rotation factor« :
$$w = {\rm e}^{-{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi /N} = \cos \left( {2 \pi}/{N}\right)-{\rm j} \cdot \sin \left( {2 \pi}/{N}\right) \hspace{0.05cm}.$$

$\text{Definition:}$  The term  »Discrete Fourier Transform«  $\rm (DFT)$  means the calculation of the  $N$  spectral coefficients  $D(\mu)$  from the  $N$  signal coefficients  $d(\nu)$:

On defining the  »Discrete Fourier Transform«  $\rm (DFT)$  with  $N=8$
$$D(\mu) = \frac{1}{N} \cdot \sum_{\nu = 0 }^{N-1} d(\nu)\cdot {w}^{\hspace{0.05cm}\nu \hspace{0.07cm} \cdot \hspace{0.05cm}\mu} \hspace{0.05cm}. $$

In the diagram you can see:

  • The  $N = 8$  signal coefficients  $d(\nu)$  by the blue filling,
  • the  $N = 8$  spectral coefficients  $D(\mu)$  at the green filling.


Inverse discrete Fourier transform


The  »inverse discrete Fourier transform«  describes the  »second Fourier integral«:

$$\begin{align*}x(t) & = \int_{-\infty }^{+\infty}X(f) \cdot {\rm e}^{\hspace{0.05cm}{\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi \hspace{0.05cm}\cdot \hspace{0.05cm} f \hspace{0.05cm}\cdot \hspace{0.05cm} t}\hspace{0.1cm} {\rm d}f\end{align*}$$

in discretized form:  

$$d(\nu) = {\rm P}\left\{x(t)\right\}{\big|}_{t \hspace{0.05cm}= \hspace{0.05cm}\nu \hspace{0.05cm}\cdot \hspace{0.05cm}T_{\rm A}}\hspace{0.01cm}.$$

$\text{Definition:}$  The term  »Inverse Discrete Fourier Transform»   $\rm (IDFT)$   means the calculation of the signal coefficients  $d(\nu)$  from the spectral coefficients  $D(\mu)$:

On defining the  »Inverse Discrete Fourier Transform«  $\rm (IDFT)$  with  $N=8$
$$d(\nu) = \sum_{\mu = 0 }^{N-1} D(\mu) \cdot {w}^{-\nu \hspace{0.07cm} \cdot \hspace{0.05cm}\mu} \hspace{0.05cm}.$$
  • With indices  $\nu = 0, \hspace{0.05cm}\text{...} \hspace{0.05cm}, N-1, $  and   $\mu = 0, \hspace{0.05cm}\text{...} \hspace{0.05cm}, N-1$  holds:
$$d(\nu) = {\rm P}\left\{x(t)\right\}{\big \vert}_{t \hspace{0.05cm}= \hspace{0.05cm}\nu \hspace{0.05cm}\cdot \hspace{0.05cm}T_{\rm A} }\hspace{0.01cm},$$
$$D(\mu) = f_{\rm A} \cdot {\rm P}\left\{X(f)\right\}{\big \vert}_{f \hspace{0.05cm}= \hspace{0.05cm}\mu \hspace{0.05cm}\cdot \hspace{0.05cm}f_{\rm A} } \hspace{0.01cm},$$
$$w = {\rm e}^{- {\rm j} \hspace{0.05cm}\cdot \hspace{0.05cm} 2 \pi /N} \hspace{0.01cm}.$$
  • A comparison between  »DFT«  and  »IDFT«  shows that exactly the same algorithm can be used. 
  • The only differences between IDFT and DFT are:
  1. The exponent of the rotation factor is to be applied with different sign.
  2. In the IDFT,  the division by  $N$  is omitted.


Interpretation of DFT and IDFT


The graph shows the discrete coefficients in the time and frequency domain together with the periodified continuous-time functions.

Time and frequency coefficients of the DFT

When using DFT or IDFT,  please note:

  1. According to the above definitions, the DFT coefficients  $d(ν)$  and  $D(\mu)$  always have the unit of the time function.
  2. Dividing  $D(\mu)$  by  $f_{\rm A}$,  one obtains the spectral value  $X(\mu \cdot f_{\rm A})$.
  3. The spectral coefficients  $D(\mu)$  must always be complex in order to be able to consider odd time functions.
  4. One also uses complex time coefficients  $d(\nu)$   ⇒   DFT and IDFT are also applicable to band-pass signals.
  5. The basic interval for  $\nu$  and  $\mu$  is usually defined as the range from  $0$  to  $N - 1$  $($filled circles in the graph$)$.


With the complex-valued number sequences 

$$\langle \hspace{0.03cm}d(\nu)\hspace{0.03cm}\rangle = \langle \hspace{0.03cm}d(0), \hspace{0.05cm}\text{...} \hspace{0.05cm} , d(N-1) \hspace{0.03cm}\rangle,$$
$$\langle \hspace{0.03cm}D(\mu)\hspace{0.03cm}\rangle = \langle \hspace{0.03cm}D(0), \hspace{0.05cm}\text{...} \hspace{0.05cm} , D(N-1) \hspace{0.03cm}\rangle,$$

DFT and IDFT are symbolized similarly to the conventional Fourier transform:

$$\langle \hspace{0.03cm} D(\mu)\hspace{0.03cm}\rangle \hspace{0.2cm}\bullet\!\!-\!\!\!-(N)\!-\!\!\!-\!\!\hspace{0.05cm}\circ\, \hspace{0.2cm} \langle \hspace{0.03cm} d(\nu) \hspace{0.03cm}\rangle \hspace{0.05cm}.$$
  1. If the function  $x(t)$  is already limited to the range  $0 \le t \lt N \cdot T_{\rm A}$  then the IDFT output directly give the samples of the time function:   $d(\nu) = x(\nu \cdot T_{\rm A}).$
  2. If  $x(t)$  is shifted with respect to the basic interval,  one has to choose the association shown in  $\text{Example 3}$  between  $x(t)$  and the coefficients  $d(\nu)$.


$\text{Example 3:}$  The upper graph shows the asymmetric triangular pulse  $x(t)$ whose absolute width is smaller than  $T_{\rm P} = N \cdot T_{\rm A}$.

On assigning of the DFT coefficients with  $N=8$


The sketch below shows the assigned DFT coefficients $($valid for  $N = 8)$.

  • For  $\nu = 0,\hspace{0.05cm}\text{...} \hspace{0.05cm} , N/2 = 4$  holds   $d(\nu) = x(\nu \cdot T_{\rm A})$ :
$$d(0) = x (0)\hspace{0.05cm}, \hspace{0.15cm} d(1) = x (T_{\rm A})\hspace{0.05cm}, \hspace{0.15cm} d(2) = x (2T_{\rm A})\hspace{0.05cm}, $$
$$d(3) = x (3T_{\rm A})\hspace{0.05cm}, \hspace{0.15cm} d(4) = x (4T_{\rm A})\hspace{0.05cm}.$$
  • The coefficients  $d(5)$,  $d(6)$  and  d$(7)$  are to be set as follows:
$$d(5) = x (-3T_{\rm A})\hspace{0.05cm}, \hspace{0.35cm} d(6) = x (-2T_{\rm A})\hspace{0.05cm}, \hspace{0.35cm} d(7) = x (-T_{\rm A})\hspace{0.05cm}$$
$$ \Rightarrow \hspace{0.2cm}d(\nu) = x \big ((\nu\hspace{-0.05cm} - \hspace{-0.05cm} N ) \cdot T_{\rm A}\big ). $$

Exercises for the chapter


Exercise 5.2: Inverse Discrete Fourier Transform

Exercise 5.2Z: DFT of a Triangular Pulse