Difference between revisions of "Aufgaben:Exercise 3.10: Maximum Likelihood Tree Diagram"

From LNTwww
 
(2 intermediate revisions by the same user not shown)
Line 3: Line 3:
  
 
[[File:P_ID1465__Dig_A_3_10_95.png|right|frame|Signals and tree diagram]]
 
[[File:P_ID1465__Dig_A_3_10_95.png|right|frame|Signals and tree diagram]]
As in  [[Aufgaben:Exercise_3.09:_Correlation_Receiver_for_Unipolar_Signaling|"Exercise 3.9"]]  we consider the joint decision of three binary symbols (bits) by means of the correlation receiver.
+
As in  [[Aufgaben:Exercise_3.09:_Correlation_Receiver_for_Unipolar_Signaling|"Exercise 3.9"]]  we consider the joint decision of three binary symbols  ("bits")  by means of the correlation receiver.
 
*The possible transmitted signals  $s_0(t), \ \text{...} \ , \ s_7(t)$  are bipolar.
 
*The possible transmitted signals  $s_0(t), \ \text{...} \ , \ s_7(t)$  are bipolar.
 +
 
*In the graphic the functions  $s_0(t)$,  $s_1(t)$,  $s_2(t)$  and  $s_3(t)$  are shown.
 
*In the graphic the functions  $s_0(t)$,  $s_1(t)$,  $s_2(t)$  and  $s_3(t)$  are shown.
 +
 
*The blue curves are valid for rectangular NRZ transmission pulses.
 
*The blue curves are valid for rectangular NRZ transmission pulses.
  
  
Below is drawn the so-called tree diagram for this constellation under the condition that the signal  $s_3(t)$  was transmitted. Shown here in the range from  $0$  to  $3T$  are the functions
+
Below is drawn the so-called  "tree diagram"  for this constellation under the condition that the signal  $s_3(t)$  was sent.  Shown here in the range from  $0$  to  $3T$  are the functions
 
:$$i_i(t)  =  \int_{0}^{t} s_3(\tau) \cdot s_i(\tau) \,{\rm d}
 
:$$i_i(t)  =  \int_{0}^{t} s_3(\tau) \cdot s_i(\tau) \,{\rm d}
 
\tau \hspace{0.3cm}( i = 0, \ \text{...} \  , 7)\hspace{0.05cm}.$$
 
\tau \hspace{0.3cm}( i = 0, \ \text{...} \  , 7)\hspace{0.05cm}.$$
  
*The correlation receiver compares the final values  $I_i = i_i(3T)$  with each other and searches for the largest possible value  $I_j$.  
+
*The correlation receiver compares the final values  $I_i = i_i(3T)$  with each other and searches for the largest possible value  $I_j$.
 +
 
*The corresponding signal  $s_j(t)$  is then the one most likely to have been sent according to the maximum likelihood criterion.
 
*The corresponding signal  $s_j(t)$  is then the one most likely to have been sent according to the maximum likelihood criterion.
  
 +
*Note that the correlation receiver generally makes the decision based on the corrected quantities  
 +
:$$W_i = I_i \ - E_i/2.$$
 +
*But since for bipolar rectangles all transmitted signals  $(i = 0,  \ \text{...} \  , \ 7)$  have exactly the same energy
 +
:$$E_i  =  \int_{0}^{3T} s_i^2(t) \,{\rm d} t,$$
  
Note that the correlation receiver generally makes the decision based on the corrected quantities  $W_i = I_i \ - E_i/2$.  But since for bipolar rectangles all transmitted signals  $(i = 0,  \ \text{...} \  , \ 7)$  have exactly the same energy
+
:the integrals  $I_i$  provide exactly the same maximum likelihood information as the corrected quantities  $W_i$.
:$$E_i  =  \int_{0}^{3T} s_i^2(t) \,{\rm d} t$$
 
 
 
the integrals  $I_i$  provide exactly the same maximum likelihood information as the corrected quantities  $W_i$.
 
  
 
The red signal waveforms  $s_i(t)$  are obtained from the blue ones by convolution with the impulse response  $h_{\rm G}(t)$  of a Gaussian low-pass filter with cutoff frequency  $f_{\rm G} \cdot T = 0.35$.  
 
The red signal waveforms  $s_i(t)$  are obtained from the blue ones by convolution with the impulse response  $h_{\rm G}(t)$  of a Gaussian low-pass filter with cutoff frequency  $f_{\rm G} \cdot T = 0.35$.  
 
*Each individual rectangular pulse is broadened.  
 
*Each individual rectangular pulse is broadened.  
*The red signal waveforms lead to intersymbol interference in case of threshold decision.
 
 
 
  
 +
*The red signal waveforms lead to  "intersymbol interference"  in case of threshold decision.
  
  
 
+
Note:  The exercise belongs to the chapter  [[Digital_Signal_Transmission/Optimal_Receiver_Strategies|"Optimal Receiver Strategies"]].
''Note:''
 
*The exercise belongs to the chapter  [[Digital_Signal_Transmission/Optimal_Receiver_Strategies|"Optimal Receiver Strategies"]].
 
 
   
 
   
  
Line 39: Line 39:
 
===Questions===
 
===Questions===
 
<quiz display=simple>
 
<quiz display=simple>
{Give the following normalized final values &nbsp;$I_i/E_{\rm B}$&nbsp; for rectangular signals (without noise).
+
{Give the following normalized final values &nbsp;$I_i/E_{\rm B}$&nbsp; for rectangular signals&nbsp; $($without noise$)$.
 
|type="{}"}
 
|type="{}"}
 
$I_0/E_{\rm B} \ = \ $  { -1.03--0.97 }
 
$I_0/E_{\rm B} \ = \ $  { -1.03--0.97 }
Line 49: Line 49:
 
|type="[]"}
 
|type="[]"}
 
- The tree diagram can be further described by straight line segments.
 
- The tree diagram can be further described by straight line segments.
+ If &nbsp;$I_3$&nbsp; is the maximum $I_i$ value, the receiver decides correctly.
+
+ If &nbsp;$I_3$&nbsp; is the maximum&nbsp; $I_i$&nbsp; value,&nbsp; the receiver decides correctly.
- &nbsp;$I_0 = I_6$ is valid independent of the strength of the noise.
+
- &nbsp;$I_0 = I_6$&nbsp; is valid independent of the strength of the noise.
  
{Which statements are valid for the red signal waveforms (with intersymbol interference)?
+
{Which statements are valid for the red signal waveforms&nbsp; $($with intersymbol interference$)$?
 
|type="[]"}
 
|type="[]"}
 
- The tree diagram can be further described by straight line segments.
 
- The tree diagram can be further described by straight line segments.
Line 66: Line 66:
 
===Solution===
 
===Solution===
 
{{ML-Kopf}}
 
{{ML-Kopf}}
'''(1)'''&nbsp; The left graph shows the tree diagram (without noise) with all final values. Highlighted in green is the curve $i_0(t)/E_{\rm B}$ with the final result $I_0/E_{\rm B} = \ &ndash;1$, which first rises linearly to $+1$ &ndash; the first bit of $s_0(t)$ and $s_3(t)$ in each case coincide &ndash; and then falls off over two bit durations.
+
'''(1)'''&nbsp; The left graph shows the tree diagram&nbsp; (without noise)&nbsp; with all final values.&nbsp; Highlighted in green is the curve&nbsp; $i_0(t)/E_{\rm B}$&nbsp; with the final result&nbsp; $I_0/E_{\rm B} = \ -1$,&nbsp; which first rises linearly to&nbsp; $+1$&nbsp; $($the first bit of&nbsp; $s_0(t)$&nbsp; and&nbsp; $s_3(t)$&nbsp; in each case coincide$)$&nbsp; and then falls off over two bit durations.
[[File:P_ID1466__Dig_A_3_10.png|right|frame|Tree diagram of the correlation receiver]]
+
[[File:EN_Dig_A_3_10_ML.png|right|frame|Tree diagram of the correlation receiver]]
The correct results are thus:
+
 
 +
*The correct results are thus:
 
:$$I_0/E_{\rm B}\hspace{0.15cm}\underline { = -1},$$
 
:$$I_0/E_{\rm B}\hspace{0.15cm}\underline { = -1},$$
 
:$$I_2/E_{\rm B} \hspace{0.15cm}\underline {= +1}, $$
 
:$$I_2/E_{\rm B} \hspace{0.15cm}\underline {= +1}, $$
Line 74: Line 75:
 
:$$I_6/E_{\rm B}\hspace{0.15cm}\underline { = -1}
 
:$$I_6/E_{\rm B}\hspace{0.15cm}\underline { = -1}
 
  \hspace{0.05cm}.$$
 
  \hspace{0.05cm}.$$
<br clear=all>
 
'''(2)'''&nbsp; Only the <u>second solution</u> is correct:
 
*In the presence of (noise) disturbances, the functions $i_i(t)$ no longer increase or decrease linearly, but have a curve as shown in the right graph.
 
*As long as $I_3 > I_{\it i&ne;3}$, the correlation receiver decides correctly.
 
*In the presence of noise, $I_0 &ne; I_6$ always holds, in contrast to the noise-free tree diagram.
 
  
  
'''(3)'''&nbsp; Only the <u>second statement</u> is true:  
+
'''(2)'''&nbsp; Only the&nbsp; <u>second solution</u>&nbsp; is correct:
*Since now the possible transmitted signals $s_i(t)$ can no longer be composed of isolated  horizontal sections, also the tree diagram without noise does not consist of straight line segments.
+
*In the presence of noises,&nbsp; the functions&nbsp; $i_i(t)$&nbsp; no longer increase or decrease linearly,&nbsp; but have a curve as shown in the right graph.
*Since the energies $E_i$ are different &ndash; this can be seen, for example, by comparing the (red) signals $s_0(t)$ and $s_2(t)$ &ndash; it is essential to use the corrected quantities $W_i$ for the decision.
+
*As long as&nbsp; $I_3 > I_{\it i&ne;3}$,&nbsp; the correlation receiver decides correctly.
*The use of the pure correlation values $I_i$ can already lead to wrong decisions without noise disturbances.
+
*In the presence of noise,&nbsp; $I_0 &ne; I_6$&nbsp; always holds,&nbsp; in contrast to the noise-free tree diagram.
 +
 
 +
 
 +
'''(3)'''&nbsp; Only the&nbsp; <u>second statement</u>&nbsp; is true:  
 +
*Since now the possible transmitted signals&nbsp; $s_i(t)$&nbsp; can no longer be composed of isolated  horizontal sections,&nbsp; also the tree diagram without noise does not consist of straight line segments.
 +
*Since the energies&nbsp; $E_i$&nbsp; are different &ndash; this can be seen e.g. by comparing the&nbsp; (red)&nbsp; signals&nbsp; $s_0(t)$&nbsp; and&nbsp; $s_2(t)$&nbsp; &ndash; it is essential to use the corrected quantities&nbsp; $W_i$&nbsp; for the decision.
 +
*The use of the uncorrected correlation values&nbsp; $I_i$&nbsp; can already lead to wrong decisions without noise disturbances.
  
  
'''(4)'''&nbsp; <u>Answer 1</u> is correct:
+
'''(4)'''&nbsp; <u>Answer 1</u>&nbsp; is correct:
*In the case <u>without intersymbol interference</u> (blue rectangular signals), all signals are limited to the range $0 \ ... \ 3T$.  
+
*In the case&nbsp; <u>without intersymbol interference</u>&nbsp; (blue rectangular signals),&nbsp; all signals are limited to the range&nbsp; $0 \ ... \ 3T$.  
*Outside this range the received signal $r(t)$ is pure noise.
+
*Outside this range the received signal&nbsp; $r(t)$&nbsp; is pure noise.&nbsp; Therefore in this case also the integration over the range&nbsp; $0 \ \text{...} \ 3T$.  
*Therefore in this case also the integration over the range $0 \ \text{...} \ 3T$.  
+
*In contrast,&nbsp; when intersymbol interference&nbsp; (red signals)&nbsp; is taken into account,&nbsp; the integrands&nbsp; $s_3(t) \cdot s_i(t)$&nbsp; also differ outside this range.
*In contrast, when intersymbol interference (red signals) is taken into account, the integrands $s_3(t) \cdot s_i(t)$ also differ outside this range.
+
*Therefore,&nbsp; if&nbsp; $t_1 = \ -T$&nbsp; and&nbsp; $t_2 = +4T$&nbsp; are chosen,&nbsp; the error probability of the correlation receiver is further reduced compared to the integration range&nbsp; $0 \ \text{...} \ 3T$.
*Therefore, if $t_1 = \ &ndash;T$ and $t_2 = +4T$ are chosen, the error probability of the correlation receiver is further reduced compared to the integration range $0 \ \text{...} \ 3T$ is further reduced.
 
 
{{ML-Fuß}}
 
{{ML-Fuß}}
  

Latest revision as of 17:24, 30 June 2022

Signals and tree diagram

As in  "Exercise 3.9"  we consider the joint decision of three binary symbols  ("bits")  by means of the correlation receiver.

  • The possible transmitted signals  $s_0(t), \ \text{...} \ , \ s_7(t)$  are bipolar.
  • In the graphic the functions  $s_0(t)$,  $s_1(t)$,  $s_2(t)$  and  $s_3(t)$  are shown.
  • The blue curves are valid for rectangular NRZ transmission pulses.


Below is drawn the so-called  "tree diagram"  for this constellation under the condition that the signal  $s_3(t)$  was sent.  Shown here in the range from  $0$  to  $3T$  are the functions

$$i_i(t) = \int_{0}^{t} s_3(\tau) \cdot s_i(\tau) \,{\rm d} \tau \hspace{0.3cm}( i = 0, \ \text{...} \ , 7)\hspace{0.05cm}.$$
  • The correlation receiver compares the final values  $I_i = i_i(3T)$  with each other and searches for the largest possible value  $I_j$.
  • The corresponding signal  $s_j(t)$  is then the one most likely to have been sent according to the maximum likelihood criterion.
  • Note that the correlation receiver generally makes the decision based on the corrected quantities  
$$W_i = I_i \ - E_i/2.$$
  • But since for bipolar rectangles all transmitted signals  $(i = 0, \ \text{...} \ , \ 7)$  have exactly the same energy
$$E_i = \int_{0}^{3T} s_i^2(t) \,{\rm d} t,$$
the integrals  $I_i$  provide exactly the same maximum likelihood information as the corrected quantities  $W_i$.

The red signal waveforms  $s_i(t)$  are obtained from the blue ones by convolution with the impulse response  $h_{\rm G}(t)$  of a Gaussian low-pass filter with cutoff frequency  $f_{\rm G} \cdot T = 0.35$.

  • Each individual rectangular pulse is broadened.
  • The red signal waveforms lead to  "intersymbol interference"  in case of threshold decision.


Note:  The exercise belongs to the chapter  "Optimal Receiver Strategies".



Questions

1

Give the following normalized final values  $I_i/E_{\rm B}$  for rectangular signals  $($without noise$)$.

$I_0/E_{\rm B} \ = \ $

$I_2/E_{\rm B} \ = \ $

$I_4/E_{\rm B} \ = \ $

$I_6/E_{\rm B} \ = \ $

2

Which statements are valid when considering a noise term?

The tree diagram can be further described by straight line segments.
If  $I_3$  is the maximum  $I_i$  value,  the receiver decides correctly.
 $I_0 = I_6$  is valid independent of the strength of the noise.

3

Which statements are valid for the red signal waveforms  $($with intersymbol interference$)$?

The tree diagram can be further described by straight line segments.
The signal energies  $E_i(i = 0, \ \text{...} \ , 7$)  are different.
Both the decision variables  $I_i$  and  $W_i$  are suitable.

4

How should the intergration range  $(t_1 \ \text{...} \ t_2)$  be chosen?

Without intersymbol interference (blue),  $t_1 = 0$  and  $t_2 = 3T$  are best possible.
With intersymbol interference (red),  $t_1 = 0$  and  $t_2 = 3T$  are best possible.


Solution

(1)  The left graph shows the tree diagram  (without noise)  with all final values.  Highlighted in green is the curve  $i_0(t)/E_{\rm B}$  with the final result  $I_0/E_{\rm B} = \ -1$,  which first rises linearly to  $+1$  $($the first bit of  $s_0(t)$  and  $s_3(t)$  in each case coincide$)$  and then falls off over two bit durations.

Tree diagram of the correlation receiver
  • The correct results are thus:
$$I_0/E_{\rm B}\hspace{0.15cm}\underline { = -1},$$
$$I_2/E_{\rm B} \hspace{0.15cm}\underline {= +1}, $$
$$I_4/E_{\rm B} \hspace{0.15cm}\underline {= -3}, $$
$$I_6/E_{\rm B}\hspace{0.15cm}\underline { = -1} \hspace{0.05cm}.$$


(2)  Only the  second solution  is correct:

  • In the presence of noises,  the functions  $i_i(t)$  no longer increase or decrease linearly,  but have a curve as shown in the right graph.
  • As long as  $I_3 > I_{\it i≠3}$,  the correlation receiver decides correctly.
  • In the presence of noise,  $I_0 ≠ I_6$  always holds,  in contrast to the noise-free tree diagram.


(3)  Only the  second statement  is true:

  • Since now the possible transmitted signals  $s_i(t)$  can no longer be composed of isolated horizontal sections,  also the tree diagram without noise does not consist of straight line segments.
  • Since the energies  $E_i$  are different – this can be seen e.g. by comparing the  (red)  signals  $s_0(t)$  and  $s_2(t)$  – it is essential to use the corrected quantities  $W_i$  for the decision.
  • The use of the uncorrected correlation values  $I_i$  can already lead to wrong decisions without noise disturbances.


(4)  Answer 1  is correct:

  • In the case  without intersymbol interference  (blue rectangular signals),  all signals are limited to the range  $0 \ ... \ 3T$.
  • Outside this range the received signal  $r(t)$  is pure noise.  Therefore in this case also the integration over the range  $0 \ \text{...} \ 3T$.
  • In contrast,  when intersymbol interference  (red signals)  is taken into account,  the integrands  $s_3(t) \cdot s_i(t)$  also differ outside this range.
  • Therefore,  if  $t_1 = \ -T$  and  $t_2 = +4T$  are chosen,  the error probability of the correlation receiver is further reduced compared to the integration range  $0 \ \text{...} \ 3T$.