Aufgaben:Exercise 4.7: Spectra of ASK and BPSK: Difference between revisions

From LNTwww
No edit summary
Hwang (talk | contribs)
No edit summary
Line 1: Line 1:


{{quiz-Header|Buchseite=Modulationsverfahren/Lineare digitale Modulation
{{quiz-Header|Buchseite=Modulation_Methods/Linear_Digital_Modulation
}}
}}


[[File:P_ID1701__Mod_A_4_6.png|right|frame|Leistungsdichtespektren von  $q(t)$  und  $s(t)$   – gültig für ASK und BPSK]]
[[File:P_ID1701__Mod_A_4_6.png|right|frame|Power-spectral densities of  $q(t)$  and  $s(t)$   – valid for ASK and BPSK]]
Die Sendesignale von ASK  (''Amplitude Shift Keying'')  und BPSK  (''Binary Phase Shift Keying'')  können beide in der Form
The transmitted signals of ASK  (''Amplitude Shift Keying'')  and BPSK  (''Binary Phase Shift Keying'')  can both be expressed in the form
:$$s(t) = q(t) · z(t)$$  
:$$s(t) = q(t) · z(t)$$  
dargestellt werden, wobei  $z(t)$  eine harmonische Schwingung mit der Frequenz  $f_{\rm T}$  und der Amplitude  $1$  darstellt.  Die Trägerphase  $ϕ_{\rm T}$  ist für die hier betrachteten Leistungsdichtespektren nicht von Bedeutung.
where  $z(t)$  represents a harmonic oscillation with frequency  $f_{\rm T}$  and amplitude  $1$.  The carrier phase  $ϕ_{\rm T}$  is not important for the power-spectral densities considered here.


*Die Quelle ist jeweils redundanzfrei, was bedeutet, dass die beiden möglichen Symbole $±1$ gleichwahrscheinlich sind und die Symbole statistisch voneinander unabhängig.
*In each case, the source is redundancy-free, which means that the two possible symbols $±1$ are equally probable and the symbols are statistically independent of each other.
*Bei ASK sind unipolare Amplitudenkoeffizienten – das heißt:  $a_ν ∈ \{0, 1\}$  – des Quellensignals
*For ASK, unipolar amplitude coefficients - that is:  $a_ν ∈ \{0, 1\}$  – of the source signal are
:$$ q(t) = \sum_{\nu = - \infty}^{+\infty}a_\nu \cdot g_q (t - \nu \cdot T)$$
:$$ q(t) = \sum_{\nu = - \infty}^{+\infty}a_\nu \cdot g_q (t - \nu \cdot T)$$
:anzusetzen, während im Fall der BPSK  $a_ν ∈ \{-1, +1\}$  zu berücksichtigen ist.  
:while in the case of BPSK  $a_ν ∈ \{-1, +1\}$  has to be considered.




In der Grafik sind die Leistungsdichtespektren  ${\it Φ}_q(f)$  und  ${\it Φ}_s(f)$  von Quellensignal und Sendesignal angegeben, die sich bei einem NRZ–Rechteckimpuls  $g_q(t)$  mit der Amplitude  $s_0 = 2 \ \rm V$  und der Dauer  $T = 1 \ \rm µ s$  ergeben.  Damit lautet die Spektralfunktion:
In the diagram, the power-spectral densities  ${\it Φ}_q(f)$  and  ${\it Φ}_s(f)$  of source signal and transmitted signal are given, respectively, for an NRZ rectangular pulse  $g_q(t)$  with amplitude  $s_0 = 2 \ \rm V$  and duration  $T = 1 \ \rm µ s$.  Thus the spectral function is:
:$$G_q(f) = s_0 \cdot T \cdot {\rm si}(\pi f T)\hspace{0.05cm}.$$
:$$G_q(f) = s_0 \cdot T \cdot {\rm si}(\pi f T)\hspace{0.05cm}.$$
Zu bestimmen sind die Konstanten  $A$,  $B$,  $C$  und  $D$  für die Modulationsverfahren  $\rm ASK$  und  $\rm BPSK$.
The constants  $A$,  $B$,  $C$  and  $D$  for the  $\rm ASK$  and  $\rm BPSK$ modulation methods are to be determined.




Line 26: Line 26:




''Hinweise:''  
''Notes:''  
*Die Aufgabe gehört zum  Kapitel  [[Modulation_Methods/Lineare_digitale_Modulation|Lineare digitale Modulation]].
*The exercise belongs to the chapter  [[Modulation_Methods/Linear_Digital_Modulation|Linear Digital Modulation]].
*Bezug genommen wird aber auch auf das Kapitel   [[Digital_Signal_Transmission/Grundlagen_der_codierten_Übertragung|Grundlagen der codierten Übertragung]]   im Buch „Digitalsignalübertragung”.
*However, reference is also made to the chapter   [[Digital_Signal_Transmission/Grundlagen_der_codierten_Übertragung|Basics of Coded Transmission]]   in the book "Digital Signal Transmission".
*Die Leistungen sind in   $\rm V^2$  anzugeben;  sie beziehen sich somit auf den Bezugswiderstand  $R = 1 \ \rm \Omega$.
*The powers are to be specified in   $\rm V^2$;  they thus refer to the reference resistance  $R = 1 \ \rm \Omega$.
   
   






===Fragebogen===
===Questions===


<quiz display=simple>
<quiz display=simple>
{Welche Werte  ergeben sich bei ASK für die Parameter &nbsp;$A = {\it Φ}_q(f = 0)$&nbsp; und &nbsp;$B$&nbsp; $($Diracgewicht bei &nbsp;$f = 0)$?
{Which values result for ASK for the parameters &nbsp;$A = {\it Φ}_q(f = 0)$&nbsp; and &nbsp;$B$&nbsp; $($Dirac weight at &nbsp;$f = 0)$?
|type="{}"}
|type="{}"}
$A \ = \ $ { 1 3% } $\ \cdot 10^{-6} \  \rm V^2/Hz$
$A \ = \ $ { 1 3% } $\ \cdot 10^{-6} \  \rm V^2/Hz$
Line 43: Line 43:




{Bestimmen Sie für das ASK–Sendesignal die Parameter &nbsp;$C = {\it Φ}_s(f = f_{\rm T})$&nbsp; und &nbsp;$D$&nbsp;  $($Diracgewicht bei $f = f_{\rm T})$ .
{Determine the parameters &nbsp;$C = {\it Φ}_s(f = f_{\rm T})$&nbsp; and &nbsp;$D$&nbsp;  $($Dirac weight at $f = f_{\rm T})$ for the ASK transmitted signal.
|type="{}"}
|type="{}"}
$C \ = \ $ { 0.25 3% } $\ \cdot 10^{-6} \  \rm V^2/Hz$
$C \ = \ $ { 0.25 3% } $\ \cdot 10^{-6} \  \rm V^2/Hz$
$D \ = \ $ { 0.25 3% } $\ \rm V^2$  
$D \ = \ $ { 0.25 3% } $\ \rm V^2$  


{Welche Werte  ergeben sich bei BPSK für die Parameter &nbsp;$A$&nbsp; und &nbsp;$B$?
{What are the values of BPSK for parameters &nbsp;$A$&nbsp; and &nbsp;$B$?
|type="{}"}
|type="{}"}
$A \ = \ $ { 4 3% }  $\ \cdot 10^{-6} \  \rm V^2/Hz$
$A \ = \ $ { 4 3% }  $\ \cdot 10^{-6} \  \rm V^2/Hz$
$B \ = \ $ { 0. } $\ \rm V^2$  
$B \ = \ $ { 0. } $\ \rm V^2$  


{Welche Werte  ergeben sich bei BPSK für die Parameter &nbsp;$C$&nbsp; und &nbsp;$D$?
{What are the values of parameters &nbsp;$C$&nbsp; and &nbsp;$D$ for BPSK?
|type="{}"}
|type="{}"}
$C \ = \ $ { 1 3% } $\ \cdot 10^{-6} \  \rm V^2/Hz$  
$C \ = \ $ { 1 3% } $\ \cdot 10^{-6} \  \rm V^2/Hz$  
$D \ = \ $ { 0. }  $\ \rm V^2$
$D \ = \ $ { 0. }  $\ \rm V^2$


{Welche Aussagen treffen immer zu, also auch dann, wenn &nbsp;$g_q(t)$&nbsp; kein NRZ–Rechteckimpuls ist?
{Which statements are always true, i.e., even if &nbsp;$g_q(t)$&nbsp; is not an NRZ rectangular pulse?
|type="[]"}
|type="[]"}
+ Der kontinuierliche Anteil von &nbsp;$ {\it Φ}_q(f)$&nbsp; ist formgleich mit &nbsp;$|G_q(f)|^2$.
+ The continuous part of &nbsp;$ {\it Φ}_q(f)$&nbsp; is equal in form to &nbsp;$|G_q(f)|^2$.
- ${\it Φ}_q(f)$&nbsp; beinhaltet bei ASK eine einzige Diraclinie $($bei $f = 0)$.
- ${\it Φ}_q(f)$&nbsp; contains a single Dirac line at ASK $($at $f = 0)$.
- ${\it Φ}_q(f)$ beinhaltet bei BPSK eine einzige Diraclinie $($bei $f = 0)$.
- ${\it Φ}_q(f)$ contains a single Dirac line at BPSK $($at $f = 0)$.


</quiz>
</quiz>


===Musterlösung===
===Solution===
{{ML-Kopf}}
{{ML-Kopf}}
'''(1)'''&nbsp; Der Gleichanteil des unipolaren redundanzfreien Quellensignals beträgt&nbsp; $m_q = s_0/2$.&nbsp; Das Diracgewicht ist somit&nbsp; $B = m_q^2 = s_0^2/4\hspace{0.15cm}\underline{ = 1 \ \rm V^2}$.  
'''(1)'''&nbsp; The DC component of the unipolar redundancy-free source signal is&nbsp; $m_q = s_0/2$.&nbsp; Thus, the Dirac weight is&nbsp; $B = m_q^2 = s_0^2/4\hspace{0.15cm}\underline{ = 1 \ \rm V^2}$.  


*Ohne diesen Gleichanteil ergäbe sich das stochastische Rechtecksignal&nbsp; $q(t) - m_q ∈ \{+s_0/2, -s_0/2\}$.  
*Without this DC component, the stochastic square wave signal&nbsp; $q(t) - m_q ∈ \{+s_0/2, -s_0/2\}$ would be obtained.
*Dieses gleichsignalfreie Signal besitzt den kontinuierlichen LDS–Anteil&nbsp; $(s_0/2)^2 · T · {\rm si}^2(πfT)$.  
*This DC-free signal has the continuous PSD component&nbsp; $(s_0/2)^2 · T · {\rm si}^2(πfT)$.  
*Hieraus lässt sich der gesuchte Wert bei der Frequenz&nbsp; $f = 0$&nbsp; ermitteln:
*From this, the value we are looking for at frequency&nbsp; $f = 0$&nbsp; can be determined:
:$$A = \frac {s_0^2 \cdot T }{4} = \frac {(2\,{\rm V})^2 \cdot
:$$A = \frac {s_0^2 \cdot T }{4} = \frac {(2\,{\rm V})^2 \cdot
10^{-6} \,{\rm s}}{4}\hspace{0.15cm}\underline {= 10^{-6} \,{\rm V^{2}/Hz}}.$$
10^{-6} \,{\rm s}}{4}\hspace{0.15cm}\underline {= 10^{-6} \,{\rm V^{2}/Hz}}.$$
Line 78: Line 78:




'''(2)'''&nbsp; Das Spektrum&nbsp; $Z(f)$&nbsp; eines Cosinussignals&nbsp; $z(t)$&nbsp; besteht aus zwei Diracfunktionen bei&nbsp; $\pm f_{\rm T}$, jeweils mit dem Gewicht&nbsp; $1/2$.
'''(2)'''&nbsp; The spectrum&nbsp; $Z(f)$&nbsp; of a cosine signal&nbsp; $z(t)$&nbsp; consists of two Dirac functions at&nbsp; $\pm f_{\rm T}$, each with weight&nbsp; $1/2$.
*Das Leistungsdichtespektrum&nbsp; ${\it Φ}_z(f)$&nbsp; besteht ebenfalls aus den beiden Diracfunktionen, nun aber mit jeweiligem Gewicht&nbsp; $1/4$.  
*The power-spectral density&nbsp; ${\it Φ}_z(f)$&nbsp; also consists of the two Dirac functions, but now with respective weights&nbsp; $1/4$.  
*Die Faltung&nbsp; ${\it Φ}_q(f) ∗ {\it Φ}_z(f)$&nbsp; ergibt das Leistungsdichtespektrum&nbsp; ${\it Φ}_s(f)$&nbsp; des Sendesignals.&nbsp; Daraus folgt:  
*The convolution&nbsp; ${\it Φ}_q(f) ∗ {\it Φ}_z(f)$&nbsp; gives the power-spectral density&nbsp; ${\it Φ}_s(f)$&nbsp; of the transmitted signal.&nbsp; It follows that:
:$$C =  {A}/{4} \hspace{0.15cm}\underline { = 0.25 \cdot 10^{-6} \,{\rm
:$$C =  {A}/{4} \hspace{0.15cm}\underline { = 0.25 \cdot 10^{-6} \,{\rm
V^{2}/Hz}},\hspace{0.2cm}D = {B}/{4}\hspace{0.15cm}\underline { = 0.25 \,{\rm V^{2}}}.$$
V^{2}/Hz}},\hspace{0.2cm}D = {B}/{4}\hspace{0.15cm}\underline { = 0.25 \,{\rm V^{2}}}.$$


''Anmerkung:'' &nbsp; Die Leistung pro Bit ergibt sich als das Integral über&nbsp; ${\it Φ}_s(f)$:
''Note:'' &nbsp; The power per bit is obtained as the integral over&nbsp; ${\it Φ}_s(f)$:
:$$P_{\rm S}  = \int_{ - \infty }^\infty \hspace{-0.3cm}  {{\it \Phi}_s(f)}\hspace{0.1cm} {\rm d}f
:$$P_{\rm S}  = \int_{ - \infty }^\infty \hspace{-0.3cm}  {{\it \Phi}_s(f)}\hspace{0.1cm} {\rm d}f
  = 2 \cdot \int_{ 0 }^\infty \hspace{-0.3cm}  {\left [ C \cdot {\rm si}^2(\pi f T) + D \cdot \delta (f - f_{\rm T}]\right ]}\hspace{0.1cm} {\rm
  = 2 \cdot \int_{ 0 }^\infty \hspace{-0.3cm}  {\left [ C \cdot {\rm si}^2(\pi f T) + D \cdot \delta (f - f_{\rm T}]\right ]}\hspace{0.1cm} {\rm
Line 94: Line 94:




'''(3)'''&nbsp; Bei BPSK ist das Quellensignal&nbsp; $q(t)$&nbsp; bipolar anzusetzen.  
'''(3)'''&nbsp; For BPSK, the source signal&nbsp; $q(t)$&nbsp; is to be bipolar.
*Im Leistungsdichtespektrum fehlt deshalb die Diraclinie &nbsp;  ⇒  &nbsp; $\underline{B = 0}$.
*Therefore, the Dirac line is missing in the power-spectral density&nbsp;  ⇒  &nbsp; $\underline{B = 0}$.
Der kontinuierliche LDS–Anteil ist viermal so groß wie bei der ASK:
The continuous PSD component is four times larger than in the ASK:
:$$A =  {s_0^2 \cdot T }\hspace{0.15cm}\underline { = 4 \cdot 10^{-6} \,{\rm V^{2}/Hz}}.$$
:$$A =  {s_0^2 \cdot T }\hspace{0.15cm}\underline { = 4 \cdot 10^{-6} \,{\rm V^{2}/Hz}}.$$






'''(4)'''&nbsp; Für die LDS–Parameter des BPSK–Sendesignals gilt analog zur ASK:
'''(4)'''&nbsp; For the PSD parameters of the BPSK transmitted signal, the following applies analogously to the ASK:
:$$C = \frac {A}{4}\hspace{0.15cm}\underline { = 10^{-6} \,{\rm V^{2}/Hz}},\hspace{0.2cm}D =
:$$C = \frac {A}{4}\hspace{0.15cm}\underline { = 10^{-6} \,{\rm V^{2}/Hz}},\hspace{0.2cm}D =
\frac {B}{4} \hspace{0.15cm}\underline {= 0}.$$
\frac {B}{4} \hspace{0.15cm}\underline {= 0}.$$
Line 107: Line 107:




'''(5)'''&nbsp; Richtig ist nur die <u>erste Aussage</u>:
'''(5)'''&nbsp; Only the <u>first statement</u> is correct:
* Bei BPSK&nbsp; (bipolares Quellensignal)&nbsp; beinhaltet&nbsp; ${\it Φ}_q(f)$&nbsp; auch dann keine einzige Diraclinie, wenn&nbsp; $g_q(t)$&nbsp; von der Rechteckform abweicht&nbsp; (gleichwahrscheinliche Symbole vorausgesetzt).  
* For BPSK&nbsp; (bipolar source signal)&nbsp; &nbsp; ${\it Φ}_q(f)$&nbsp; does not contain a single Dirac line even if&nbsp; $g_q(t)$&nbsp; deviates from the rectangular form&nbsp; (assuming equally probable symbols).  
*Dagegen beinhaltet das unipolare ASK–Quellensignal unendlich viele Diraclinien bei allen Vielfachen von&nbsp; $1/T$.  
*In contrast, the unipolar ASK source signal contains infinitely many Dirac lines at all multiples of&nbsp; $1/T$.  




Weitere Informationen zu diesem Thema finden Sie auf der Seite&nbsp;  „AKF und LDS bei unipolaren Binärsignalen”&nbsp; im Buch „Digitalsignalübertragung”.
For more information on this topic, see the page&nbsp;  "ACF and PSD for unipolar binary signals" in the book "Digital Signal Transmission".


{{ML-Fuß}}
{{ML-Fuß}}

Revision as of 17:40, 21 March 2022

Power-spectral densities of  $q(t)$  and  $s(t)$  – valid for ASK and BPSK

The transmitted signals of ASK  (Amplitude Shift Keying)  and BPSK  (Binary Phase Shift Keying)  can both be expressed in the form

$$s(t) = q(t) · z(t)$$

where  $z(t)$  represents a harmonic oscillation with frequency  $f_{\rm T}$  and amplitude  $1$.  The carrier phase  $ϕ_{\rm T}$  is not important for the power-spectral densities considered here.

  • In each case, the source is redundancy-free, which means that the two possible symbols $±1$ are equally probable and the symbols are statistically independent of each other.
  • For ASK, unipolar amplitude coefficients - that is:  $a_ν ∈ \{0, 1\}$  – of the source signal are
$$ q(t) = \sum_{\nu = - \infty}^{+\infty}a_\nu \cdot g_q (t - \nu \cdot T)$$
while in the case of BPSK  $a_ν ∈ \{-1, +1\}$  has to be considered.


In the diagram, the power-spectral densities  ${\it Φ}_q(f)$  and  ${\it Φ}_s(f)$  of source signal and transmitted signal are given, respectively, for an NRZ rectangular pulse  $g_q(t)$  with amplitude  $s_0 = 2 \ \rm V$  and duration  $T = 1 \ \rm µ s$.  Thus the spectral function is:

$$G_q(f) = s_0 \cdot T \cdot {\rm si}(\pi f T)\hspace{0.05cm}.$$

The constants  $A$,  $B$,  $C$  and  $D$  for the  $\rm ASK$  and  $\rm BPSK$ modulation methods are to be determined.





Notes:

  • The exercise belongs to the chapter  Linear Digital Modulation.
  • However, reference is also made to the chapter  Basics of Coded Transmission  in the book "Digital Signal Transmission".
  • The powers are to be specified in  $\rm V^2$;  they thus refer to the reference resistance  $R = 1 \ \rm \Omega$.



Questions

1 Which values result for ASK for the parameters  $A = {\it Φ}_q(f = 0)$  and  $B$  $($Dirac weight at  $f = 0)$?

$A \ = \ $ $\ \cdot 10^{-6} \ \rm V^2/Hz$
$B \ = \ $ $\ \rm V^2$

2 Determine the parameters  $C = {\it Φ}_s(f = f_{\rm T})$  and  $D$  $($Dirac weight at $f = f_{\rm T})$ for the ASK transmitted signal.

$C \ = \ $ $\ \cdot 10^{-6} \ \rm V^2/Hz$
$D \ = \ $ $\ \rm V^2$

3 What are the values of BPSK for parameters  $A$  and  $B$?

$A \ = \ $ $\ \cdot 10^{-6} \ \rm V^2/Hz$
$B \ = \ $ $\ \rm V^2$

4 What are the values of parameters  $C$  and  $D$ for BPSK?

$C \ = \ $ $\ \cdot 10^{-6} \ \rm V^2/Hz$
$D \ = \ $ $\ \rm V^2$

5 Which statements are always true, i.e., even if  $g_q(t)$  is not an NRZ rectangular pulse?

The continuous part of  $ {\it Φ}_q(f)$  is equal in form to  $|G_q(f)|^2$.
${\it Φ}_q(f)$  contains a single Dirac line at ASK $($at $f = 0)$.
${\it Φ}_q(f)$ contains a single Dirac line at BPSK $($at $f = 0)$.


Solution

(1)  The DC component of the unipolar redundancy-free source signal is  $m_q = s_0/2$.  Thus, the Dirac weight is  $B = m_q^2 = s_0^2/4\hspace{0.15cm}\underline{ = 1 \ \rm V^2}$.

  • Without this DC component, the stochastic square wave signal  $q(t) - m_q ∈ \{+s_0/2, -s_0/2\}$ would be obtained.
  • This DC-free signal has the continuous PSD component  $(s_0/2)^2 · T · {\rm si}^2(πfT)$.
  • From this, the value we are looking for at frequency  $f = 0$  can be determined:
$$A = \frac {s_0^2 \cdot T }{4} = \frac {(2\,{\rm V})^2 \cdot

10^{-6} \,{\rm s}}{4}\hspace{0.15cm}\underline {= 10^{-6} \,{\rm V^{2}/Hz}}.$$


(2)  The spectrum  $Z(f)$  of a cosine signal  $z(t)$  consists of two Dirac functions at  $\pm f_{\rm T}$, each with weight  $1/2$.

  • The power-spectral density  ${\it Φ}_z(f)$  also consists of the two Dirac functions, but now with respective weights  $1/4$.
  • The convolution  ${\it Φ}_q(f) ∗ {\it Φ}_z(f)$  gives the power-spectral density  ${\it Φ}_s(f)$  of the transmitted signal.  It follows that:
$$C = {A}/{4} \hspace{0.15cm}\underline { = 0.25 \cdot 10^{-6} \,{\rm

V^{2}/Hz}},\hspace{0.2cm}D = {B}/{4}\hspace{0.15cm}\underline { = 0.25 \,{\rm V^{2}}}.$$

Note:   The power per bit is obtained as the integral over  ${\it Φ}_s(f)$:

$$P_{\rm S} = \int_{ - \infty }^\infty \hspace{-0.3cm} {{\it \Phi}_s(f)}\hspace{0.1cm} {\rm d}f
= 2 \cdot \int_{ 0 }^\infty \hspace{-0.3cm}  {\left [ C \cdot {\rm si}^2(\pi f T) + D \cdot \delta (f - f_{\rm T}]\right ]}\hspace{0.1cm} {\rm
d}f= 2 \cdot \left [ \frac{C}{T}  + D \right ] =
2 \cdot \left [ \frac{0.25 \cdot 10^{-6} \,{\rm

V^{2}/Hz}}{10^{-6} \,{\rm s}} + 0.25 \,{\rm V^{2}} \right ] \hspace{0.15cm}\underline {= 1 \,{\rm V^{2}}}.$$


(3)  For BPSK, the source signal  $q(t)$  is to be bipolar.

  • Therefore, the Dirac line is missing in the power-spectral density  ⇒   $\underline{B = 0}$.
  • The continuous PSD component is four times larger than in the ASK:
$$A = {s_0^2 \cdot T }\hspace{0.15cm}\underline { = 4 \cdot 10^{-6} \,{\rm V^{2}/Hz}}.$$


(4)  For the PSD parameters of the BPSK transmitted signal, the following applies analogously to the ASK:

$$C = \frac {A}{4}\hspace{0.15cm}\underline { = 10^{-6} \,{\rm V^{2}/Hz}},\hspace{0.2cm}D =

\frac {B}{4} \hspace{0.15cm}\underline {= 0}.$$


(5)  Only the first statement is correct:

  • For BPSK  (bipolar source signal)    ${\it Φ}_q(f)$  does not contain a single Dirac line even if  $g_q(t)$  deviates from the rectangular form  (assuming equally probable symbols).
  • In contrast, the unipolar ASK source signal contains infinitely many Dirac lines at all multiples of  $1/T$.


For more information on this topic, see the page  "ACF and PSD for unipolar binary signals" in the book "Digital Signal Transmission".