Difference between revisions of "Channel Coding/Basics of Convolutional Coding"

From LNTwww
 
(71 intermediate revisions by 8 users not shown)
Line 1: Line 1:
 
   
 
   
 
{{Header
 
{{Header
|Untermenü=Faltungscodierung und geeignete Decoder
+
|Untermenü=Convolutional Codes and Their Decoding
|Vorherige Seite=Fehlerwahrscheinlichkeit und Anwendungsgebiete
+
|Vorherige Seite=Error Probability and Areas of Application
|Nächste Seite=Algebraische und polynomische Beschreibung
+
|Nächste Seite=Algebraic and Polynomial Description
 
}}
 
}}
  
== Voraussetzungen und Definitionen ==
+
== # OVERVIEW OF THE THIRD MAIN CHAPTER # ==
 
<br>
 
<br>
Wir betrachten in diesem Kapitel eine unendlich lange binäre Informationssequenz $\underline{u}$ und unterteilen diese in Informationsblöcke $\underline{u}_i$ zu je $k \ {\rm Bit}$. Man kann diesen Sachverhalt wie folgt formalisieren:
+
This third main chapter discusses &nbsp; &raquo;'''convolutional codes'''&laquo;, &nbsp; first described in 1955 by&nbsp; [https://en.wikipedia.org/wiki/Peter_Elias $\text{Peter Elias}$]&nbsp; [Eli55]<ref name='Eli55'>Elias, P.:&nbsp; Coding for Noisy Channels.&nbsp; In:&nbsp; IRE Conv. Rec. Part 4, pp. 37-47, 1955.</ref>.&nbsp;
  
:<math>\underline{\it u} = \left ( \underline{\it u}_1, \underline{\it u}_2, ... \hspace{0.1cm}, \underline{\it u}_i , ... \hspace{0.1cm}\right ) \hspace{0.3cm}{\rm mit}\hspace{0.3cm} \underline{\it u}_i = \left ( u_i^{(1)}, u_i^{(2)}, ... \hspace{0.1cm}, u_i^{(k)}\right )\hspace{0.05cm},</math>
+
*While for linear block codes&nbsp; $($see&nbsp; [[Channel_Coding/Objective_of_Channel_Coding#.23_OVERVIEW_OF_THE_FIRST_MAIN_CHAPTER_.23|$\text{first main chapter}$]]$)$&nbsp; and Reed-Solomon codes&nbsp; $($see&nbsp; [[Channel_Coding/Some_Basics_of_Algebra#.23_OVERVIEW_OF_THE_SECOND_MAIN_CHAPTER_.23|$\text{second main chapter}$]]$)$&nbsp; the code word length is&nbsp; $n$&nbsp; in each case,&nbsp;
  
:<math>u_i^{(j)}\in {\rm GF(2)}\hspace{0.3cm}{\rm f\ddot{u}r} \hspace{0.3cm}1 \le j \le k  
+
*the theory of convolutional codes is based on&nbsp; "semi-infinite"&nbsp; information and encoded sequences.&nbsp; Similarly,&nbsp; "maximum likelihood decoding"&nbsp; using the&nbsp; [[Channel_Coding/Decoding_of_Convolutional_Codes#Viterbi_algorithm_based_on_correlation_and_metrics|$\text{Viterbi algorithm}$]]&nbsp; per se yields the entire sequence.
 +
 
 +
 
 +
Specifically,&nbsp; this chapter discusses:
 +
 
 +
#Important definitions for convolutional codes: &nbsp;  &raquo;code rate&laquo;,&nbsp; &raquo;memory&laquo;,&nbsp; &raquo; influence length&laquo;, &nbsp;&raquo; free distance&laquo;,
 +
#&raquo;similarities&laquo;&nbsp; and &nbsp;&raquo;differences&laquo;&nbsp; to linear block codes,
 +
#&raquo;generator matrix&laquo;&nbsp; and &nbsp;&raquo;transfer function matrix&laquo;&nbsp; of a convolutional code,
 +
#&raquo;fractional-rational transfer functions&laquo;&nbsp; for systematic convolutional codes,
 +
#description with&nbsp; &raquo;state transition diagram&laquo;&nbsp; and &nbsp;&raquo;trellis diagram&laquo;,
 +
#&raquo;termination&laquo;&nbsp; and &nbsp;&raquo;puncturing&laquo;&nbsp; of convolutional codes,
 +
#&raquo;decoding&laquo;&nbsp; of convolutional codes &nbsp; ⇒ &nbsp; &raquo;Viterbi algorithm&laquo;,
 +
#&raquo;weight enumerator functions&laquo;&nbsp; and approximations for the &nbsp;&raquo;bit error probability&laquo;.
 +
 
 +
 
 +
 
 +
== Requirements and definitions ==
 +
<br>
 +
We consider in this chapter an infinite binary information sequence&nbsp; $\underline{u}$&nbsp; and divide it into information blocks&nbsp; $\underline{u}_i$&nbsp; of&nbsp; $k$&nbsp; bits each. One can formalize this fact as follows:
 +
 
 +
::<math>\underline{\it u} = \left ( \underline{\it u}_1, \underline{\it u}_2, \hspace{0.05cm} \text{...} \hspace{0.05cm}, \underline{\it u}_i , \hspace{0.05cm} \text{...} \hspace{0.05cm}\right ) \hspace{0.3cm}{\rm with}\hspace{0.3cm} \underline{\it u}_i = \left ( u_i^{(1)}, u_i^{(2)}, \hspace{0.05cm} \text{...} \hspace{0.05cm}, u_i^{(k)}\right )\hspace{0.05cm},</math>
 +
 
 +
::<math>u_i^{(j)}\in {\rm GF(2)}\hspace{0.3cm}{\rm for} \hspace{0.3cm}1 \le j \le k  
 
\hspace{0.5cm}\Rightarrow \hspace{0.5cm} \underline{\it u}_i \in {\rm GF}(2^k)\hspace{0.05cm}.</math>
 
\hspace{0.5cm}\Rightarrow \hspace{0.5cm} \underline{\it u}_i \in {\rm GF}(2^k)\hspace{0.05cm}.</math>
  
Im Englischen bezeichnet man eine solche Sequenz ohne negative Indizes als <i>semi&ndash;infinite</i>.<br>
+
Such a sequence without negative indices is called&nbsp; &raquo;'''semi&ndash;infinite'''&laquo;.<br>
 +
 
 +
{{BlaueBox|TEXT= 
 +
$\text{Definition:}$&nbsp; In a&nbsp; &raquo;<b>binary convolutional code</b>&laquo;,&nbsp; a code word&nbsp; $\underline{x}_i$&nbsp; consisting of&nbsp; $n$&nbsp; code bit is output at time&nbsp; $i$:
  
{{Definition}}''':''' Bei einem <b>binären Faltungscode</b> (englisch: <i>Binary Convolutional Code</i>)  wird zu dem Taktzeitpunkt $i$ ein Codewort $\underline{x}_i$ bestehend aus $n$ Codebits ausgegeben:
+
::<math>\underline{\it x}_i = \left ( x_i^{(1)}, x_i^{(2)}, \hspace{0.05cm} \text{...} \hspace{0.05cm}, x_i^{(n)}\right )\in {\rm GF}(2^n)\hspace{0.05cm}.</math>
  
:<math>\underline{\it x}_i = \left ( x_i^{(1)}, x_i^{(2)}, ... \hspace{0.1cm}, x_i^{(n)}\right )\in {\rm GF}(2^n)\hspace{0.05cm}.</math>
+
This results according to
 +
*the&nbsp; $k$&nbsp; bit of the current information block&nbsp; $\underline{u}_i$, and <br>
  
Dieses ergibt sich entsprechend
+
*the&nbsp; $m$&nbsp; previous information blocks&nbsp; $\underline{u}_{i-1}$, ... , $\underline{u}_{i-m}$.<br>}}<br>
*den $k$ Bit des aktuellen Informationsblockes $\underline{u}_i$, und<br>
 
  
*den $m$ vorherigen Informationsblöcken $\underline{u}_{i&ndash;1}, \ ... \ \underline{u}_{i&ndash;m}$.<br>{{end}}<br>
+
[[File:EN_KC_T_3_1_S1_neu.png|right|frame|Dependencies for a convolutional encoder with&nbsp; $m = 2$|class=fit]]
  
Die folgende Grafik verdeutlicht diesen Sachverhalt für die Parameter $k = 4, n = 7, m = 2$ und $i = 4$. Die $n = 7$ zum Zeitpunkt $i = 4$ erzeugten Codebits $x_4^{(1)}, \ ... \ , x_4^{(7)}$ können (direkt) von den $k \cdot (m+1) = 12$ rot markierten Informationsbits abhängen und werden durch Modulo&ndash;2&ndash;Additionen erzeugt.<br>
+
The diagram opposite illustrates this fact for the parameters&nbsp;
 +
:$$k = 4, \ n = 7, \ m = 2,\ i = 4.$$  
  
[[File:P ID2590 KC T 3 1 S1 v1.png|Abhängigkeiten bei einem Faltungscodierer mit <i>m</i> = 2|class=fit]]<br>
+
The&nbsp; $n = 7$&nbsp; code bits&nbsp; $x_4^{(1)}$, ... , $x_4^{(7)}$&nbsp; generated at time&nbsp; $i = 4$&nbsp; may depend&nbsp; $($directly$)$&nbsp; on the&nbsp; $k \cdot (m+1) = 12$&nbsp; information bits marked in red and are generated by modulo-2 additions.<br>
  
Gelb eingezeichnet ist zudem ein $(n, k)$&ndash;Faltungscodierer. Zu beachten ist, dass sich der Vektor $\underline{u}_i$ und die Sequenz $\underline{u}^{(i)}$ grundlegend unterscheiden. Während $\underline{u}_i = (u_i^{(1)}, u_i^{(2)}, \ ... \ , u_i^{(k)})$ die $k$ zum Zeitpunkt $i$ parallel anliegenden Informationsbits zusammenfasst, bezeichnet $\underline{u}^i = (u_i^{(i)}$, $u_2^{(i)}, \ ...)$ die (horizontale) Sequenz am $i$&ndash;ten Eingang des Faltungscodierers. Für den Faltungscode gelten folgende Definitionen:
+
Drawn in yellow is a&nbsp; $(n, k)$&nbsp; convolutional encoder.&nbsp; Note that the vector&nbsp; $\underline{u}_i$&nbsp; and the sequence&nbsp; $\underline{u}^{(i)}$&nbsp; are fundamentally different:
*Die <b>Coderrate</b> ergibt sich wie bei den Blockcodes zu $R = k/n$.<br>
+
*While&nbsp; $\underline{u}_i = (u_i^{(1)}, u_i^{(2)}$, ... , $u_i^{(k)})$&nbsp; summarizes&nbsp; $k$&nbsp;  parallel information bits at time&nbsp; $i$,
  
*Man bezeichnet $m$ als das <b>Gedächtnis</b> (englisch: <i>Memory</i>) des Codes und den <i>Convolutional Code</i>  selbst mit ${\rm CC}(n, k, m)$.<br>
+
* $\underline{u}^i = (u_1^{(i)}$, $u_2^{(i)}, \ \text{...})$&nbsp; denotes the&nbsp; $($horizontal$)$&nbsp; sequence at&nbsp; $i$&ndash;th input of the convolutional encoder.  
  
*Daraus ergibt sich die <b>Einflusslänge</b> (englisch: <i>Constraint Length</i>) zu $\nu = m + 1$.<br>
 
  
*Für $k > 1$ gibt man diese Parameter oft auch in Bit an: $m_{\rm Bit} = m \cdot k$ bzw. $\nu_{\rm Bit} = (m + 1) \cdot k$.<br>
+
{{BlaueBox|TEXT=
 +
$\text{Some definitions related to convolutional codes:}$&nbsp; 
 +
*The &nbsp; &raquo;<b>code rate</b>&laquo; &nbsp; results as for the block codes to&nbsp; $R = k/n$.<br>
  
== Gemeinsamkeiten und Unterschiede gegenüber Blockcodes ==
+
*Perceive &nbsp; $m$&nbsp; as the&nbsp; &raquo;<b>memory</b>&laquo;&nbsp; of the code and the&nbsp; "convolutional code"&nbsp; itself with&nbsp; ${\rm CC} \ (n, k, m)$.<br>
<br>
 
Aus der [[Kanalcodierung/Grundlagen_der_Faltungscodierung#Voraussetzungen_und_Definitionen| Definition]] auf der letzten Seite ist ersichtlich, dass ein binärer Faltungscode mit $m = 0$ (also ohne Gedächtnis) identisch wäre mit einem binären Blockcode wie in Kapitel 1 beschrieben. Wir schließen diesen Grenzfall aus und setzen deshalb für das Folgende stets voraus:
 
*Das Gedächtnis $m$ sei größer oder gleich $1$.<br>
 
  
*Die Einflusslänge $\nu$ sei größer oder gleich $2$.<br><br>
+
*This gives the&nbsp; &raquo;<b>constraint length</b>&laquo;&nbsp; $\nu = m + 1$&nbsp; of the convolutional code.<br>
  
{{Beispiel}}''':''' Bei einem $(7, 4)$&ndash;Blockcode hängt das Codewort $\underline {x}_4$ nur vom Informationswort $\underline{u}_4$ ab, nicht jedoch von $\underline{u}_2$ und $\underline{u}_3$, wie bei dem [http://en.lntwww.de/Kanalcodierung/Grundlagen_der_Faltungscodierung#Voraussetzungen_und_Definitionen beispielhaften Faltungscodes] (mit $m = 2$) auf der letzten Seite.<br>
+
*For&nbsp; $k > 1$&nbsp; one often gives these parameters also in bits: &nbsp; $m_{\rm bit} = m \cdot k$ &nbsp; &nbsp;resp.&nbsp; &nbsp; $\nu_{\rm bit} = (m + 1) \cdot k$.<br>}}
  
[[File:P ID2592 KC T 3 1 S2 v2.png|Abhängigkeiten bei einem (7, 4)–Blockcode zum Zeitpunkt <i>i</i> = 4|class=fit]]<br>
+
== Similarities and differences compared to block codes ==
 +
<br>
 +
From the&nbsp; [[Channel_Coding/Basics_of_Convolutional_Coding#Requirements_and_definitions| $\text{definition}$]]&nbsp; in the previous section,&nbsp; it is evident that a binary convolutional code with&nbsp; $m = 0$&nbsp; $($i.e., without memory$)$&nbsp; would be identical to a binary block code as described in the first main chapter.&nbsp; We exclude this limiting case and therefore assume always for the following:
 +
*The memory&nbsp; $m$&nbsp; be greater than or equal to&nbsp; $1$.<br>
  
Gilt beispielsweise $x_4^{(1)} = u_4^{(1)}, \ x_4^{(2)} = u_4^{(2)}, \ x_4^{(3)} = u_4^{(3)}, \ x_4^{(4)} = u_4^{(4)}$ sowie
+
*The influence length&nbsp; $\nu$&nbsp; be greater than or equal to&nbsp; $2$.<br><br>
  
:<math>x_4^{(5)} = u_4^{(1)}+ u_4^{(2)}+u_4^{(3)}\hspace{0.05cm} ,\hspace{0.3cm}
+
{{GraueBox|TEXT=
x_4^{(6)} = u_4^{(2)}+ u_4^{(3)}+u_4^{(4)}\hspace{0.05cm} ,\hspace{0.3cm}
+
$\text{Example 1:}$&nbsp; For a&nbsp; $(7, 4)$&nbsp; block code,&nbsp; the code word&nbsp; $\underline {x}_4$&nbsp; depends only on the information word&nbsp; $\underline{u}_4$&nbsp; but not on&nbsp; $\underline{u}_2$&nbsp; and&nbsp; $\underline{u}_3$, as in the&nbsp; [[Channel_Coding/Basics_of_Convolutional_Coding#Requirements_and_definitions| $\text{example convolutional codes}$]]&nbsp; $($with&nbsp; $m = 2)$&nbsp; in the last section.<br>
x_4^{(7)} = u_4^{(1)}+ u_4^{(2)}+u_4^{(4)}\hspace{0.05cm} ,</math>
 
  
so handelt es sich um einen so genannten <i>systematischen Hamming&ndash;Code</i> $(7, 4, 3)$ entsprechend [http://en.lntwww.de/Kanalcodierung/Beispiele_bin%C3%A4rer_Blockcodes#Hamming.E2.80.93Codes_.282.29 Kapitel 1.3.] In der Grafik sind diese speziellen Abhängigkeiten für $x_4^{(1)}$ und $x_4^{(7)}$ rot eingezeichnet.{{end}}<br>
+
[[File:EN_KC_T_3_1_S2.png|right|frame|Dependencies on a&nbsp; $(7, 4)$&nbsp; block code at time&nbsp; $i = 4$|class=fit]]
  
In gewisser Weise könnte man auch einen $(n, k)$&ndash;Faltungscode mit Gedächtnis $m &#8805; 1$ als Blockcode interpretieren, dessen Codeparameter $n' >> n$ und $k' >> k$ allerdings sehr viel größere Werte annehmen müssten als die des vorliegenden Faltungscodes. Aufgrund der großen Unterschiede in der Beschreibung, in den Eigenschaften und insbesondere bei der Decodierung betrachten wir aber Faltungscodes in diesem Lerntutorial als etwas völlig Neues. Hierfür sprechen folgende Gründe:
+
For example, if
*Ein Blockcodierer setzt Informationsworte der Länge $k \ \rm Bit$ blockweise in Codeworte mit $n \ \rm Bit$ um. Der Blockcode ist dabei um so leistungsfähiger, je größer seine Codewortlänge $n$ ist. Bei gegebener Coderate $R = k/n$ erfordert dies auch eine große Informationswortlänge $k$.<br>
+
:$$x_4^{(1)} = u_4^{(1)}, \ x_4^{(2)} = u_4^{(2)},$$
 +
:$$x_4^{(3)} = u_4^{(3)}, \ x_4^{(4)} = u_4^{(4)}$$  
 +
as well as
 +
:$$x_4^{(5)} = u_4^{(1)}+ u_4^{(2)}+u_4^{(3)}\hspace{0.05cm},$$
 +
:$$x_4^{(6)} = u_4^{(2)}+ u_4^{(3)}+u_4^{(4)}\hspace{0.05cm},$$
 +
:$$x_4^{(7)} = u_4^{(1)}+ u_4^{(2)}+u_4^{(4)}\hspace{0.05cm},$$
  
*Dagegen wird die Korrekturfähigkeit eines Faltungscodes im wesentlichen durch sein Gedächtnis $m$ bestimmt. Die Codeparameter $k$ und $n$ werden hier meist sehr klein gewählt $(1, \ 2, \ 3, \ ...)$. Von praktischer Bedeutung sind somit nur ganz wenige und zudem sehr einfache Faltungscodes.<br>
+
applies then it is a so-called&nbsp; [[Channel_Coding/Examples_of_Binary_Block_Codes#Hamming_Codes| $\text{systematic Hamming code}$]]&nbsp; $\text{HS (7, 4, 3)}$.&nbsp; In the graph,&nbsp; these special dependencies for&nbsp; $x_4^{(1)}$&nbsp; and&nbsp; $x_4^{(7)}$&nbsp; are drawn in red.}}<br>
  
*Auch schon bei kleinen Werten für $k$ und $n$ überführt ein Faltungscoder eine ganze Sequenz von Informationsbits $(k' &#8594; &#8734;)$ in eine sehr lange Sequenz von Codebits $(n' = k'/R)$. Ein solcher Code bietet somit oft ebenfalls eine große Korrekturfähigkeit.<br>
+
In a sense,&nbsp; one could also interpret a&nbsp; $(n, k)$&nbsp; convolutional code with memory&nbsp; $m &#8805; 1$&nbsp; as a block code whose code parameters&nbsp; $n\hspace{0.05cm}' \gg n$&nbsp; and&nbsp; $k\hspace{0.05cm}' \gg k$&nbsp; would have to assume much larger values than those of the present convolutional code.&nbsp;
  
*Es gibt effiziente Faltungsdecoder &#8658; [http://en.lntwww.de/Kanalcodierung/Decodierung_von_Faltungscodes#Blockschaltbild_und_Voraussetzungen Viterbi&ndash;Algorithmus] bzw. [http://en.lntwww.de/Kanalcodierung/Soft%E2%80%93in_Soft%E2%80%93out_Decoder#BCJR.E2.80.93Decodierung:_Vorw.C3.A4rts.E2.80.93R.C3.BCckw.C3.A4rts.E2.80.93Algorithmus BCJR&ndash;Algorithmus.] Diese können <i>Soft&ndash;Decision&ndash;Input</i> (Zuverlässigkeitsinformationen über den Kanal) einfach verarbeiten und liefern auch <i>Soft&ndash;Decision&ndash;Output</i> (Zuverlässigkeitsinformation über das Decodierergebnis).<br>
+
However,&nbsp; because of the large differences in description,&nbsp; properties,&nbsp; and especially decoding,&nbsp; we consider convolutional codes to be something completely new in this learning tutorial.&nbsp; The reasons for this are as follows:
 +
# &nbsp; A block encoder  converts block-by-block information words of length&nbsp; $k$&nbsp; bits into code words of length&nbsp; $n$&nbsp; bits each.&nbsp; In this case,&nbsp; the larger its code word length&nbsp; $n$,&nbsp; the more powerful the block code is.&nbsp; For a given code rate&nbsp; $R = k/n$&nbsp; this also requires a large information word length&nbsp; $k$.<br>
 +
# &nbsp; In contrast,&nbsp; the correction capability of a convolutional code is essentially determined by its memory&nbsp; $m$.&nbsp; The code parameters&nbsp; $k$&nbsp; and&nbsp; $n$&nbsp; are mostly chosen very small here&nbsp; $(1, \ 2, \ 3, \ \text{...})$.&nbsp; Thus,&nbsp; only very few and,&nbsp; moreover,&nbsp; very simple convolutional codes are of practical importance.<br>
 +
# &nbsp; Even with small values for&nbsp; $k$&nbsp; and&nbsp; $n$,&nbsp; a convolutional encoder transfers a whole sequence of information bits&nbsp; $(k\hspace{0.05cm}' &#8594; &#8734;)$&nbsp; into a very long sequence of code bits&nbsp; $(n\hspace{0.05cm}' = k\hspace{0.05cm}'/R)$.&nbsp; Such a code thus often provides a large correction capability as well.<br>
 +
# &nbsp; There are efficient convolutional decoders,&nbsp; for example the&nbsp; [[Channel_Coding/Decoding_of_Convolutional_Codes#Viterbi_algorithm_based_on_correlation_and_metrics|$\text{Viterbi algorithm}$]]&nbsp; and the&nbsp; [[Channel_Coding/Soft-in_Soft-Out_Decoder#BCJR_decoding:_Forward-backward_algorithm| $\text{BCJR algorithm}$]],&nbsp; which can process reliability information about the channel &nbsp; &rArr; &nbsp; "soft decision input"&nbsp; and provide reliability information about the decoding result &nbsp; &rArr; &nbsp; "soft decision output".<br>
 +
# &nbsp; Please note:&nbsp; The terms&nbsp; "convolutional code"&nbsp; and&nbsp; "convolutional encoder"&nbsp; should not be confused:
 +
::*The&nbsp; "convolutional code"&nbsp; ${\rm CC} \ (n, \ k, \ m)$ &nbsp; &#8658; &nbsp; $R = k/n$&nbsp; is understood as&nbsp; "the set of all possible encoded sequences&nbsp; $\underline{x}$&nbsp; $($at the output$)$&nbsp; that can be generated with this code considering all possible information sequences&nbsp; $\underline{u}$&nbsp; $($at the input$)$".
 +
 +
::*There are several&nbsp; "convolutional encoders"&nbsp; that realize the same&nbsp; "convolutional code".<br>
  
== Rate–1/2–Faltungscodierer (1) ==
+
== Rate 1/2 convolutional encoder ==
 
<br>
 
<br>
Die Grafik zeigt einen $(n = 2, k = 1)$&ndash;Faltungscodierer. Zum Taktzeitpunkt $i$ liegt das Informationsbit $u_i$ am Codereingang an und es wird ein 2&ndash;Bit&ndash;Codeblock $\underline{x}_i = (x_i^{(1)}, \ x_i^{(2)})$ ausgegeben.<br>
+
[[File:EN_KC_T_3_1_S3ab.png|right|frame|Convolutional encoder&nbsp; $(n = 2, \hspace{0.05cm} k = 1)$&nbsp; for one bit&nbsp; $u_i$&nbsp; $($upper graphic$)$&nbsp; and for  the information sequence&nbsp; $\underline{u}$&nbsp;  $($lower graphic$)$|class=fit]]
 +
 
 +
The upper graph shows a&nbsp; $(n = 2, \hspace{0.05cm} k = 1)$&nbsp; convolutional encoder.&nbsp;
  
[[File:P ID2593 KC T 3 1 S3a v1.png|Faltungscoder (<i>k</i> = 1, <i>n</i> = 2) für ein Informationsbit <i>u<sub>i</sub></i> |class=fit]]<br>
+
*At clock time&nbsp; $i$&nbsp; the information bit&nbsp; $u_i$&nbsp; is present at the encoder input and the&nbsp; $2$&ndash;bit code block&nbsp; $\underline{x}_i = (x_i^{(1)}, \ x_i^{(2)})$&nbsp; is output.&nbsp;
  
Unter Berücksichtigung der (halb&ndash;unendlich) langen Informationssequenz $\underline{u}$ ergibt sich folgendes Modell:<br>
+
*To generate two code bits&nbsp; $x_i^{(1)}$&nbsp; and&nbsp; $x_i^{(2)}$&nbsp; from a single information bit,&nbsp; the convolutional encoder must include at least one memory element:
 +
:$$k = 1\hspace{0.05cm}, \ n = 2 \hspace{0.3cm}\Rightarrow \hspace{0.3cm} m \ge 1
 +
\hspace{0.05cm}.$$
  
[[File:P ID2594 KC T 3 1 S3b v1.png|class=fit|Faltungscoder (<i>k</i> = 1, <i>n</i> = 2) für die Informationssequenz <u><i>u</i></u> ]]<br>
+
*Taking into account the&nbsp; $($half&ndash;infinite$)$&nbsp; long information sequence&nbsp; $\underline{u}$&nbsp; the model results according to the lower graph.
 +
<br clear=all>
 +
{{GraueBox|TEXT=
 +
$\text{Example 2:}$&nbsp; The graphic shows a convolutional encoder for the parameters&nbsp; $k = 1, \ n = 2$&nbsp; and&nbsp; $m = 1$.&nbsp;
  
Um aus einem einzigen Informationsbit $u_i$ zwei Codebits $x_i^{(1)}$ und $x_i^{(2)}$ generieren zu können, muss der Faltungscodierer mindestens ein Speicherelement beinhalten:
+
[[File:EN_KC_T_3_1_S3c.png|right|frame|Convolutional encoder with&nbsp; $k = 1, \ n = 2, \ m = 1$&nbsp; and example sequences; &nbsp; the square with label&nbsp; $D$&nbsp;  $($"delay"$)$&nbsp; indicates a memory element. |class=fit]]
  
:<math>k = 1\hspace{0.05cm}, n = 2 \hspace{0.3cm}\Rightarrow \hspace{0.3cm} m \ge 1
+
*This is a&nbsp; "systematic"&nbsp; convolutional encoder,&nbsp; characterized by&nbsp;
\hspace{0.05cm}.</math>
+
:$$x_i^{(1)} = u_i.$$
 +
*The second output returns&nbsp; $x_i^{(2)} = u_i + u_{i-1}$.  
  
<i>Anmerkung</i>: Die beiden Begriffe <i>Faltungscodierer</i> und <i>Faltungscoder</i> werden in unserem Lerntutorial synonym verwendet und können beliebig ausgetauscht werden. Beide Begriffe bezeichnen die konkrete Umsetzung einer Informationssequenz $underline{u}$ in eine Codesequenz $\underline{x}$.<br>
 
  
Die Begriffe <i>Faltungscodierer</i> und <i>Faltungscode</i> sollte man allerdings nicht verwechseln. Unter einem Faltungscode ${\rm CC}(k, \ n, \ m)$ &nbsp;&#8658;&nbsp; $R = k/n$ versteht man die Menge aller möglichen Codesequenzen $\underline{x}$, die mit diesem Code unter Berücksichtigung aller möglichen Informationssequenzen $\underline{u} (am Eingang) generiert werden kann. Es gibt verschiedene Faltungscodierer, die den gleichen Faltungscode realisieren.<br>
+
*In the example output sequence&nbsp; $\underline{x}$&nbsp; after multiplexing 
 +
:*all&nbsp; $x_i^{(1)}$&nbsp; are labeled red,
  
{{Beispiel}}''':''' Nachfolgend ist ein Faltungscodierer für die Parameter $k = 1, n = 2$ und $m = 1$ dargestellt. Das gelbe Quadrat kennzeichnet ein Speicherelement. Dessen Beschriftung  $D$ ist von <i>Delay</i> abgeleitet.<br>
+
:*all&nbsp; $x_i^{(2)}$&nbsp; are labeled blue.}}<br>
  
[[File:P ID2595 KC T 3 1 S3c v1.png|Faltungscodierer mit <i>k</i> = 1, <i>n</i> = 2, <i>m</i> = 1 sowie Beispielsequenzen|class=fit]]<br>
+
{{GraueBox|TEXT=
 +
$\text{Example 3:}$&nbsp;
 +
The graphic shows a&nbsp; $(n = 2, \ k = 1)$&nbsp; convolutional encoder with&nbsp; $m = 2$&nbsp; memory elements:
 +
 
 +
*On the left is shown the equivalent circuit.
  
Es handelt sich hier um einen <i>systematischen</i> Faltungscodierer, gekennzeichnet durch $x_i^{(1)} = u_i$. Der zweite Ausgang liefert $x_i^{(2)} = u_i + u_{i&ndash;1}$. In der beispielhaften Ausgangssequenz nach <i>Multiplexing</i> sind alle $x_i^{(1)}$ rot und alle $x_i^{(2)}$ blau beschriftet.{{end}}<br>
+
*On the right you see a realization form of this encoder.  
  
== Rate–1/2–Faltungscodierer (2) ==
+
*The two information bits are:
<br>
+
[[File:EN_KC_T_3_1_S3d_neu.png|right|frame|Convolutional encoder&nbsp; $(k = 1, \ n = 2, \ m = 2)$&nbsp; in two different representations|class=fit]]
Nachfolgend sehen Sie links das Ersatzschaltbild eines (<i>n</i> = 2, <i>k</i> = 1)&ndash;Faltungscodierers, aber nun mit <i>m</i> = 2 Speicherelementen. Die beiden Informationsbits lauten:
 
  
:<math>x_i^{(1)} \hspace{-0.15cm}  = \hspace{-0.15cm} u_{i} + u_{i-1}+ u_{i-2} \hspace{0.05cm},</math>
+
::<math>x_i^{(1)} = u_{i} + u_{i-1}+ u_{i-2} \hspace{0.05cm},</math>
:<math>x_i^{(2)} \hspace{-0.15cm}  = \hspace{-0.15cm} u_{i} + u_{i-2} \hspace{0.05cm}.</math>
+
::<math>x_i^{(2)} = u_{i} + u_{i-2} \hspace{0.05cm}.</math>
  
Wegen <i>x<sub>i</sub></i><sup>(1)</sup> &ne; <i>u<sub>i</sub></i> handelt es sich hier nicht um einen systematischen Code.<br>
+
*Because&nbsp; $x_i^{(1)} &ne; u_i$&nbsp; this is not a systematic code.
  
[[File:P ID2596 KC T 3 1 S3d v1.png|Faltungscoder (<i>k</i> = 1, <i>n</i> = 2, <i>m</i> = 2) in zwei verschiedenen Darstellungen|class=fit]]<br>
 
  
Rechts angegeben ist eine Realisierungsform dieses Coders. Man erkennt:
+
It can be seen:
*Die Informationssequenz <u><i>u</i></u> wird in einem Schieberegister der Länge <i>L</i> = <i>m</i> + 1 = 3 abgelegt.<br>
+
# The input sequence&nbsp; $\underline{u}$&nbsp; is stored in a shift register of length&nbsp; $L = m + 1 = 3$&nbsp;.<br>
 +
#At clock time&nbsp; $i$&nbsp; the left memory element contains the current information bit&nbsp; $u_i$,&nbsp; which is shifted one place to the right at each of the next clock times.<br>
 +
#The number of yellow squares gives the memory&nbsp; $m = 2$&nbsp; of the encoder.<br><br>
  
*Zum Taktzeitpunkt <i>i</i> beinhaltet das linke Speicherelement das aktuelle Informationsbit <i>u<sub>i</sub></i>, das zu den nächsten Taktzeitpunkten jeweils um eine Stelle nach rechts verschoben wird.<br>
+
&rArr; &nbsp; From these plots it is clear that&nbsp; $x_i^{(1)}$&nbsp; and&nbsp; $x_i^{(2)}$&nbsp; can each be interpreted as the output of a&nbsp; [[Theory_of_Stochastic_Signals/Digital_Filters|$\text{digital filter}$]]&nbsp; over the Galois field&nbsp; ${\rm GF(2)}$&nbsp; where both filters operate in parallel with the same input sequence&nbsp; $\underline{u}$&nbsp;.  
  
*Aus der Anzahl der gelben Quadrate ergibt sich wieder das Gedächtnis <i>m</i> = 2 des Coders.<br><br>
+
&rArr; &nbsp; Since&nbsp; $($in general terms$)$,&nbsp; the output signal of a filter results from the&nbsp; [[Signal_Representation/The_Convolution_Theorem_and_Operation#Convolution_in_the_time_domain|$\text{convolution}$]]&nbsp; of the input signal with the filter impulse response,&nbsp; this is referred to as&nbsp; "convolutional coding".<br>}}
  
Aus den beiden Darstellungen wird deutlich, dass  <i>x<sub>i</sub></i><sup>(1)</sup> und <i>x<sub>i</sub></i><sup>(2)</sup> jeweils als der Ausgang eines Digitalen Filters über dem Galoisfeld GF(2) interpretiert werden kann, wobei beide Filter parallel mit der gleichen Eingangsfolge <u><i>u</i></u> arbeiten. Da sich (ganz allgemein) das Ausgangssignal eines Filters aus der [http://en.lntwww.de/Signaldarstellung/Faltungssatz_und_Faltungsoperation#Faltung_im_Zeitbereich Faltung] des Eingangssignals mit der Filterimpulsantwort ergibt, spricht man von <i>Faltungscodierung</i>.<br>
+
== Convolutional encoder with two inputs ==
 +
<br>  
 +
[[File:P ID2598 KC T 3 1 S4 v1.png|right|frame|Convolutional encoder with&nbsp; $k = 2$&nbsp; and&nbsp; $n = 3$]]
 +
Now consider a convolutional encoder that generates&nbsp; $n = 3$&nbsp; code bits from&nbsp; $k = 2$&nbsp; information bits.
 +
*The information sequence&nbsp; $\underline{u}$&nbsp; is divided into blocks of two bits each.
 +
 +
*At the clock time&nbsp; $i$: &nbsp; The bit&nbsp; $u_i^{(1)}$&nbsp; is present at the upper input and&nbsp; $u_i^{(2)}$ at the lower input.  
  
== Faltungscodierer mit k = 2 Eingängen ==
+
*Then holds for the&nbsp; $n = 3$&nbsp; code bits at time&nbsp; $i$:
<br>
+
::<math>x_i^{(1)} = u_{i}^{(1)} + u_{i-1}^{(1)}+ u_{i-1}^{(2)} \hspace{0.05cm},</math>
Nun betrachten wir einen Faltungscodierer, der aus <i>k</i> = 2 Informationsbits <i>n</i> = 3 Codebits generiert. Die Informationssequenz <u><i>u</i></u> wird in Blöcke zu je zwei Bit aufgeteilt. Zum Taktzeitpunkt <i>i</i> liegt am oberen Eingang das Bit <i>u<sub>i</sub></i><sup>(1)</sup> an und am unteren Eingang  <i>u<sub>i</sub></i><sup>(2)</sup>. Für die <i>n</i> = 3 Codebits zum Zeitpunkt <i>i</i> gilt dann:
+
::<math>x_i^{(2)} = u_{i}^{(2)} + u_{i-1}^{(1)} \hspace{0.05cm},</math>
[[File:P ID2598 KC T 3 1 S4 v1.png|rahmenlos|rechts| Faltungscodierer mit <i>k</i> = 2 und <i>n</i> = 3]]
+
::<math>x_i^{(3)} = u_{i}^{(1)} + u_{i}^{(2)}+ u_{i-1}^{(1)} \hspace{0.05cm}.</math>
  
:<math>x_i^{(1)} \hspace{-0.15cm}  =  \hspace{-0.15cm} u_{i}^{(1)} + u_{i-1}^{(1)}+ u_{i-1}^{(2)} \hspace{0.05cm},</math>
+
In the graph,&nbsp; the info&ndash;bits&nbsp; $u_i^{(1)}$&nbsp; and&nbsp; $u_i^{(2)}$&nbsp; are marked red resp. blue,&nbsp; and the previous info&ndash;bits&nbsp; $u_{i-1}^{(1)}$&nbsp; resp.&nbsp; $u_{i-1}^{(2)}$&nbsp; are marked green and brown, respectively
:<math>x_i^{(2)} \hspace{-0.15cm}  =  \hspace{-0.15cm} u_{i}^{(2)} + u_{i-1}^{(1)} \hspace{0.05cm},</math>
 
:<math>x_i^{(3)} \hspace{-0.15cm}  =  \hspace{-0.15cm} u_{i}^{(1)} + u_{i}^{(2)}+ u_{i-1}^{(1)} \hspace{0.05cm}.</math>
 
  
In der Grafik sind die Info&ndash;Bits <i>u<sub>i</sub></i><sup>(1)</sup> und <i>u<sub>i</sub></i><sup>(2)</sup> rot bzw. blau gekennzeichnet, und die vorherigen Info&ndash;Bits <i>u</i><sub><i>i</i>&ndash;1</sub><sup>(1)</sup> und <i>u</i><sub><i>i</i>&ndash;1</sub><sup>(2)</sup> grün bzw. braun.Anzumerken ist:
+
To be noted:
*Das <b>Gedächtnis</b> <i>m</i> ist gleich der maximalen Speicherzellenzahl in einem Zweig &nbsp;&nbsp;&#8658;&nbsp;&nbsp; hier <i>m</i> = 1.
+
# &nbsp; The&nbsp; "memory"&nbsp; $m$&nbsp; is  equal to the maximum memory cell count in a branch &nbsp;&nbsp;&#8658;&nbsp;&nbsp; here&nbsp; $m = 1$.
 +
# &nbsp; The&nbsp; "influence length"&nbsp; $\nu$&nbsp; is equal to the sum of all memory elements &nbsp;&nbsp;&#8658;&nbsp;&nbsp; here&nbsp; $\nu = 2$.<br>
 +
# &nbsp;All memory elements are set to zero at the beginning of the coding&nbsp; $($"initialization"$)$.
 +
# &nbsp;The code defined herewith is the set of all possible encoded sequences&nbsp; $\underline{x}$,&nbsp; which result when all possible information sequences&nbsp; $\underline{u}$&nbsp; are entered.  
 +
# &nbsp;Both&nbsp; $\underline{u}$&nbsp; and&nbsp; $\underline{x}$&nbsp; are thereby (temporally) unbounded.<br>
  
*Die <b>Einflusslänge</b> <i>&nu;</i> ist gleich der Summe aller Speicherelemente &nbsp;&nbsp;&#8658;&nbsp;&nbsp; hier <i>&nu;</i> = 2.<br>
 
  
*Alle Speicherelemente seien zu Beginn der Codierung (<i>Initialisierung</i>) auf Null gesetzt.<br><br>
+
{{GraueBox|TEXT= 
 +
$\text{Example 4:}$&nbsp; Let the information sequence be &nbsp;$\underline{u} = (0, 1, 1, 0, 0, 0, 1, 1, \ \text{ ...})$.&nbsp; This gives the subsequences&nbsp;
 +
*$\underline{u}^{(1)} = (0, 1, 0, 1, \ \text{ ...})$,
 +
*$\underline{u}^{(2)} = (1, 0, 0, 1, \ \text{ ...})$.  
  
Der hiermit definierte Code ist die Menge aller möglichen Codesequenzen <u><i>x</i></u>, die sich bei Eingabe aller möglichen Informationssequenzen <u><i>u</i></u> ergeben. Sowohl <u><i>u</i></u> als auch <u><i>x</i></u> seien dabei (zeitlich) unbegrenzt.<br>
 
  
{{Beispiel}}''':''' Die Informationssequenz sei <u><i>u</i></u> = (0, 1, 1, 0, 0, 0, 1, 1, ...). Daraus ergeben sich die beiden Teilsequenzen <u><i>u</i></u><sup>(1)</sup> = (0, 1, 0, 1, ...) und <u><i>u</i></u><sup>(2)</sup> = (1, 0, 0, 1, ...). Mit der Festlegung <i>u</i><sub>0</sub><sup>(1)</sup> = <i>u</i><sub>0</sub><sup>(2)</sup> = 0 folgt aus den obigen Gleichungen für die <i>n</i> = 3 Codebits:
+
With the specification&nbsp; $u_0^{(1)} = u_0^{(2)} = 0$&nbsp; it follows from the above equations for the&nbsp; $n = 3$&nbsp; code bits.
*im ersten Codierschritt (<i>i</i> = 1):
+
*at the first coding step&nbsp; $(i = 1)$:  
  
 
::<math>x_1^{(1)} = u_{1}^{(1)} = 0  \hspace{0.05cm},\hspace{0.4cm}
 
::<math>x_1^{(1)} = u_{1}^{(1)} = 0  \hspace{0.05cm},\hspace{0.4cm}
Line 137: Line 197:
 
x_1^{(3)} = u_{1}^{(1)} + u_{1}^{(2)} = 0+1 = 1 \hspace{0.05cm},</math>
 
x_1^{(3)} = u_{1}^{(1)} + u_{1}^{(2)} = 0+1 = 1 \hspace{0.05cm},</math>
  
*im zweiten Codierschritt (<i>i</i> = 2):
+
*at the second coding step&nbsp; $(i = 2)$:
  
::<math>x_2^{(1)} \hspace{-0.15cm}  = \hspace{-0.15cm} u_{2}^{(1)} + u_{1}^{(1)}+ u_{1}^{(2)} = 1 + 0 + 1 = 0\hspace{0.05cm},\hspace{0.4cm}
+
::<math>x_2^{(1)} =u_{2}^{(1)} + u_{1}^{(1)}+ u_{1}^{(2)} = 1 + 0 + 1 = 0\hspace{0.05cm},\hspace{0.4cm}
x_2^{(2)} = u_{2}^{(2)} + u_{1}^{(1)} = 0+0 = 0\hspace{0.05cm},</math>
+
x_2^{(2)} = u_{2}^{(2)} + u_{1}^{(1)} = 0+0 = 0\hspace{0.05cm},\hspace{0.4cm}
::<math>x_2^{(3)} \hspace{-0.15cm}  = \hspace{-0.15cm} u_{2}^{(1)} + u_{2}^{(2)}+ u_{1}^{(1)} = 1 + 0+0 =1\hspace{0.05cm},</math>
+
x_2^{(3)} = u_{2}^{(1)} + u_{2}^{(2)}+ u_{1}^{(1)} = 1 + 0+0 =1\hspace{0.05cm},</math>
  
*im dritten Codierschritt (<i>i</i> = 3):
+
*at the third coding step&nbsp; $(i = 3)$:
  
::<math>x_3^{(1)} \hspace{-0.15cm}  = \hspace{-0.15cm} u_{3}^{(1)} + u_{2}^{(1)}+ u_{2}^{(2)} = 0+1+0 = 1\hspace{0.05cm},\hspace{0.4cm}
+
::<math>x_3^{(1)} =u_{3}^{(1)} + u_{2}^{(1)}+ u_{2}^{(2)} = 0+1+0 = 1\hspace{0.05cm},\hspace{0.4cm}
x_3^{(2)} = u_{3}^{(2)} + u_{2}^{(1)} = 0+1=1\hspace{0.05cm},</math>
+
x_3^{(2)} = u_{3}^{(2)} + u_{2}^{(1)} = 0+1=1\hspace{0.05cm},\hspace{0.4cm}
::<math>x_3^{(3)} \hspace{-0.15cm}  = \hspace{-0.15cm} u_{3}^{(1)} + u_{3}^{(2)}+ u_{2}^{(1)} = 0+0+1 =1\hspace{0.05cm},</math>
+
x_3^{(3)} =u_{3}^{(1)} + u_{3}^{(2)}+ u_{2}^{(1)} = 0+0+1 =1\hspace{0.05cm},</math>
  
*und schließlich im vierten Codierschritt (<i>i</i> = 4):
+
*and finally at the fourth coding step&nbsp; $(i = 4)$:
  
::<math>x_4^{(1)} \hspace{-0.15cm}  = \hspace{-0.15cm} u_{4}^{(1)} + u_{3}^{(1)}+ u_{3}^{(2)} = 1+0+0 = 1\hspace{0.05cm},\hspace{0.4cm}
+
::<math>x_4^{(1)} = u_{4}^{(1)} + u_{3}^{(1)}+ u_{3}^{(2)} = 1+0+0 = 1\hspace{0.05cm},\hspace{0.4cm}
x_4^{(2)} = u_{4}^{(2)} + u_{3}^{(1)} = 1+0=1\hspace{0.05cm},</math>
+
x_4^{(2)} = u_{4}^{(2)} + u_{3}^{(1)} = 1+0=1\hspace{0.05cm},\hspace{0.4cm}
::<math>x_4^{(3)} \hspace{-0.15cm}  = \hspace{-0.15cm} u_{4}^{(1)} + u_{4}^{(2)}+ u_{3}^{(1)} = 1+1+0 =0\hspace{0.05cm}.</math>
+
x_4^{(3)}= u_{4}^{(1)} + u_{4}^{(2)}+ u_{3}^{(1)} = 1+1+0 =0\hspace{0.05cm}.</math>
  
Damit lautet die Codesequenz nach dem Multiplexer: &nbsp; <u><i>x</i></u> = (0, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, ...).{{end}}<br>
+
Thus the encoded sequence is after the multiplexer: &nbsp; &nbsp; $\underline{x} = (0, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, \ \text{...})$.}}<br>
  
== Aufgaben ==
+
== Exercises for the chapter==
 
<br>
 
<br>
[[Aufgaben:3.1 Analyse eines Faltungscoders|A3.1 Analyse eines Faltungscoders]]
+
[[Aufgaben:Exercise_3.1:_Analysis_of_a_Convolutional_Encoder|Exercise 3.1: Analysis of a Convolutional Encoder]]
 +
 
 +
[[Aufgaben:Exercise_3.1Z:_Convolution_Codes_of_Rate_1/2|Exercise 3.1Z: Convolution Codes of Rate 1/2]]
  
[[Aufgaben:3.1Z_Faltungscodes_der_Rate_1/2|Zusatzaufgaben:3.1 Faltungscodes der Rate 1/2]]
+
==References==
 +
<references/>
  
 
{{Display}}
 
{{Display}}

Latest revision as of 17:02, 7 December 2022

# OVERVIEW OF THE THIRD MAIN CHAPTER #


This third main chapter discusses   »convolutional codes«,   first described in 1955 by  $\text{Peter Elias}$  [Eli55][1]

  • the theory of convolutional codes is based on  "semi-infinite"  information and encoded sequences.  Similarly,  "maximum likelihood decoding"  using the  $\text{Viterbi algorithm}$  per se yields the entire sequence.


Specifically,  this chapter discusses:

  1. Important definitions for convolutional codes:   »code rate«,  »memory«,  » influence length«,  » free distance«,
  2. »similarities«  and  »differences«  to linear block codes,
  3. »generator matrix«  and  »transfer function matrix«  of a convolutional code,
  4. »fractional-rational transfer functions«  for systematic convolutional codes,
  5. description with  »state transition diagram«  and  »trellis diagram«,
  6. »termination«  and  »puncturing«  of convolutional codes,
  7. »decoding«  of convolutional codes   ⇒   »Viterbi algorithm«,
  8. »weight enumerator functions«  and approximations for the  »bit error probability«.


Requirements and definitions


We consider in this chapter an infinite binary information sequence  $\underline{u}$  and divide it into information blocks  $\underline{u}_i$  of  $k$  bits each. One can formalize this fact as follows:

\[\underline{\it u} = \left ( \underline{\it u}_1, \underline{\it u}_2, \hspace{0.05cm} \text{...} \hspace{0.05cm}, \underline{\it u}_i , \hspace{0.05cm} \text{...} \hspace{0.05cm}\right ) \hspace{0.3cm}{\rm with}\hspace{0.3cm} \underline{\it u}_i = \left ( u_i^{(1)}, u_i^{(2)}, \hspace{0.05cm} \text{...} \hspace{0.05cm}, u_i^{(k)}\right )\hspace{0.05cm},\]
\[u_i^{(j)}\in {\rm GF(2)}\hspace{0.3cm}{\rm for} \hspace{0.3cm}1 \le j \le k \hspace{0.5cm}\Rightarrow \hspace{0.5cm} \underline{\it u}_i \in {\rm GF}(2^k)\hspace{0.05cm}.\]

Such a sequence without negative indices is called  »semi–infinite«.

$\text{Definition:}$  In a  »binary convolutional code«,  a code word  $\underline{x}_i$  consisting of  $n$  code bit is output at time  $i$:

\[\underline{\it x}_i = \left ( x_i^{(1)}, x_i^{(2)}, \hspace{0.05cm} \text{...} \hspace{0.05cm}, x_i^{(n)}\right )\in {\rm GF}(2^n)\hspace{0.05cm}.\]

This results according to

  • the  $k$  bit of the current information block  $\underline{u}_i$, and
  • the  $m$  previous information blocks  $\underline{u}_{i-1}$, ... , $\underline{u}_{i-m}$.


Dependencies for a convolutional encoder with  $m = 2$

The diagram opposite illustrates this fact for the parameters 

$$k = 4, \ n = 7, \ m = 2,\ i = 4.$$

The  $n = 7$  code bits  $x_4^{(1)}$, ... , $x_4^{(7)}$  generated at time  $i = 4$  may depend  $($directly$)$  on the  $k \cdot (m+1) = 12$  information bits marked in red and are generated by modulo-2 additions.

Drawn in yellow is a  $(n, k)$  convolutional encoder.  Note that the vector  $\underline{u}_i$  and the sequence  $\underline{u}^{(i)}$  are fundamentally different:

  • While  $\underline{u}_i = (u_i^{(1)}, u_i^{(2)}$, ... , $u_i^{(k)})$  summarizes  $k$  parallel information bits at time  $i$,
  • $\underline{u}^i = (u_1^{(i)}$, $u_2^{(i)}, \ \text{...})$  denotes the  $($horizontal$)$  sequence at  $i$–th input of the convolutional encoder.


$\text{Some definitions related to convolutional codes:}$ 

  • The   »code rate«   results as for the block codes to  $R = k/n$.
  • Perceive   $m$  as the  »memory«  of the code and the  "convolutional code"  itself with  ${\rm CC} \ (n, k, m)$.
  • This gives the  »constraint length«  $\nu = m + 1$  of the convolutional code.
  • For  $k > 1$  one often gives these parameters also in bits:   $m_{\rm bit} = m \cdot k$    resp.    $\nu_{\rm bit} = (m + 1) \cdot k$.

Similarities and differences compared to block codes


From the  $\text{definition}$  in the previous section,  it is evident that a binary convolutional code with  $m = 0$  $($i.e., without memory$)$  would be identical to a binary block code as described in the first main chapter.  We exclude this limiting case and therefore assume always for the following:

  • The memory  $m$  be greater than or equal to  $1$.
  • The influence length  $\nu$  be greater than or equal to  $2$.

$\text{Example 1:}$  For a  $(7, 4)$  block code,  the code word  $\underline {x}_4$  depends only on the information word  $\underline{u}_4$  but not on  $\underline{u}_2$  and  $\underline{u}_3$, as in the  $\text{example convolutional codes}$  $($with  $m = 2)$  in the last section.

Dependencies on a  $(7, 4)$  block code at time  $i = 4$

For example, if

$$x_4^{(1)} = u_4^{(1)}, \ x_4^{(2)} = u_4^{(2)},$$
$$x_4^{(3)} = u_4^{(3)}, \ x_4^{(4)} = u_4^{(4)}$$

as well as

$$x_4^{(5)} = u_4^{(1)}+ u_4^{(2)}+u_4^{(3)}\hspace{0.05cm},$$
$$x_4^{(6)} = u_4^{(2)}+ u_4^{(3)}+u_4^{(4)}\hspace{0.05cm},$$
$$x_4^{(7)} = u_4^{(1)}+ u_4^{(2)}+u_4^{(4)}\hspace{0.05cm},$$

applies then it is a so-called  $\text{systematic Hamming code}$  $\text{HS (7, 4, 3)}$.  In the graph,  these special dependencies for  $x_4^{(1)}$  and  $x_4^{(7)}$  are drawn in red.


In a sense,  one could also interpret a  $(n, k)$  convolutional code with memory  $m ≥ 1$  as a block code whose code parameters  $n\hspace{0.05cm}' \gg n$  and  $k\hspace{0.05cm}' \gg k$  would have to assume much larger values than those of the present convolutional code. 

However,  because of the large differences in description,  properties,  and especially decoding,  we consider convolutional codes to be something completely new in this learning tutorial.  The reasons for this are as follows:

  1.   A block encoder converts block-by-block information words of length  $k$  bits into code words of length  $n$  bits each.  In this case,  the larger its code word length  $n$,  the more powerful the block code is.  For a given code rate  $R = k/n$  this also requires a large information word length  $k$.
  2.   In contrast,  the correction capability of a convolutional code is essentially determined by its memory  $m$.  The code parameters  $k$  and  $n$  are mostly chosen very small here  $(1, \ 2, \ 3, \ \text{...})$.  Thus,  only very few and,  moreover,  very simple convolutional codes are of practical importance.
  3.   Even with small values for  $k$  and  $n$,  a convolutional encoder transfers a whole sequence of information bits  $(k\hspace{0.05cm}' → ∞)$  into a very long sequence of code bits  $(n\hspace{0.05cm}' = k\hspace{0.05cm}'/R)$.  Such a code thus often provides a large correction capability as well.
  4.   There are efficient convolutional decoders,  for example the  $\text{Viterbi algorithm}$  and the  $\text{BCJR algorithm}$,  which can process reliability information about the channel   ⇒   "soft decision input"  and provide reliability information about the decoding result   ⇒   "soft decision output".
  5.   Please note:  The terms  "convolutional code"  and  "convolutional encoder"  should not be confused:
  • The  "convolutional code"  ${\rm CC} \ (n, \ k, \ m)$   ⇒   $R = k/n$  is understood as  "the set of all possible encoded sequences  $\underline{x}$  $($at the output$)$  that can be generated with this code considering all possible information sequences  $\underline{u}$  $($at the input$)$".
  • There are several  "convolutional encoders"  that realize the same  "convolutional code".

Rate 1/2 convolutional encoder


Convolutional encoder  $(n = 2, \hspace{0.05cm} k = 1)$  for one bit  $u_i$  $($upper graphic$)$  and for the information sequence  $\underline{u}$  $($lower graphic$)$

The upper graph shows a  $(n = 2, \hspace{0.05cm} k = 1)$  convolutional encoder. 

  • At clock time  $i$  the information bit  $u_i$  is present at the encoder input and the  $2$–bit code block  $\underline{x}_i = (x_i^{(1)}, \ x_i^{(2)})$  is output. 
  • To generate two code bits  $x_i^{(1)}$  and  $x_i^{(2)}$  from a single information bit,  the convolutional encoder must include at least one memory element:
$$k = 1\hspace{0.05cm}, \ n = 2 \hspace{0.3cm}\Rightarrow \hspace{0.3cm} m \ge 1 \hspace{0.05cm}.$$
  • Taking into account the  $($half–infinite$)$  long information sequence  $\underline{u}$  the model results according to the lower graph.


$\text{Example 2:}$  The graphic shows a convolutional encoder for the parameters  $k = 1, \ n = 2$  and  $m = 1$. 

Convolutional encoder with  $k = 1, \ n = 2, \ m = 1$  and example sequences;   the square with label  $D$  $($"delay"$)$  indicates a memory element.
  • This is a  "systematic"  convolutional encoder,  characterized by 
$$x_i^{(1)} = u_i.$$
  • The second output returns  $x_i^{(2)} = u_i + u_{i-1}$.


  • In the example output sequence  $\underline{x}$  after multiplexing
  • all  $x_i^{(1)}$  are labeled red,
  • all  $x_i^{(2)}$  are labeled blue.


$\text{Example 3:}$  The graphic shows a  $(n = 2, \ k = 1)$  convolutional encoder with  $m = 2$  memory elements:

  • On the left is shown the equivalent circuit.
  • On the right you see a realization form of this encoder.
  • The two information bits are:
Convolutional encoder  $(k = 1, \ n = 2, \ m = 2)$  in two different representations
\[x_i^{(1)} = u_{i} + u_{i-1}+ u_{i-2} \hspace{0.05cm},\]
\[x_i^{(2)} = u_{i} + u_{i-2} \hspace{0.05cm}.\]
  • Because  $x_i^{(1)} ≠ u_i$  this is not a systematic code.


It can be seen:

  1. The input sequence  $\underline{u}$  is stored in a shift register of length  $L = m + 1 = 3$ .
  2. At clock time  $i$  the left memory element contains the current information bit  $u_i$,  which is shifted one place to the right at each of the next clock times.
  3. The number of yellow squares gives the memory  $m = 2$  of the encoder.

⇒   From these plots it is clear that  $x_i^{(1)}$  and  $x_i^{(2)}$  can each be interpreted as the output of a  $\text{digital filter}$  over the Galois field  ${\rm GF(2)}$  where both filters operate in parallel with the same input sequence  $\underline{u}$ .

⇒   Since  $($in general terms$)$,  the output signal of a filter results from the  $\text{convolution}$  of the input signal with the filter impulse response,  this is referred to as  "convolutional coding".

Convolutional encoder with two inputs


Convolutional encoder with  $k = 2$  and  $n = 3$

Now consider a convolutional encoder that generates  $n = 3$  code bits from  $k = 2$  information bits.

  • The information sequence  $\underline{u}$  is divided into blocks of two bits each.
  • At the clock time  $i$:   The bit  $u_i^{(1)}$  is present at the upper input and  $u_i^{(2)}$ at the lower input.
  • Then holds for the  $n = 3$  code bits at time  $i$:
\[x_i^{(1)} = u_{i}^{(1)} + u_{i-1}^{(1)}+ u_{i-1}^{(2)} \hspace{0.05cm},\]
\[x_i^{(2)} = u_{i}^{(2)} + u_{i-1}^{(1)} \hspace{0.05cm},\]
\[x_i^{(3)} = u_{i}^{(1)} + u_{i}^{(2)}+ u_{i-1}^{(1)} \hspace{0.05cm}.\]

In the graph,  the info–bits  $u_i^{(1)}$  and  $u_i^{(2)}$  are marked red resp. blue,  and the previous info–bits  $u_{i-1}^{(1)}$  resp.  $u_{i-1}^{(2)}$  are marked green and brown, respectively

To be noted:

  1.   The  "memory"  $m$  is equal to the maximum memory cell count in a branch   ⇒   here  $m = 1$.
  2.   The  "influence length"  $\nu$  is equal to the sum of all memory elements   ⇒   here  $\nu = 2$.
  3.  All memory elements are set to zero at the beginning of the coding  $($"initialization"$)$.
  4.  The code defined herewith is the set of all possible encoded sequences  $\underline{x}$,  which result when all possible information sequences  $\underline{u}$  are entered.
  5.  Both  $\underline{u}$  and  $\underline{x}$  are thereby (temporally) unbounded.


$\text{Example 4:}$  Let the information sequence be  $\underline{u} = (0, 1, 1, 0, 0, 0, 1, 1, \ \text{ ...})$.  This gives the subsequences 

  • $\underline{u}^{(1)} = (0, 1, 0, 1, \ \text{ ...})$,
  • $\underline{u}^{(2)} = (1, 0, 0, 1, \ \text{ ...})$.


With the specification  $u_0^{(1)} = u_0^{(2)} = 0$  it follows from the above equations for the  $n = 3$  code bits.

  • at the first coding step  $(i = 1)$:
\[x_1^{(1)} = u_{1}^{(1)} = 0 \hspace{0.05cm},\hspace{0.4cm} x_1^{(2)} = u_{1}^{(2)} = 1 \hspace{0.05cm},\hspace{0.4cm} x_1^{(3)} = u_{1}^{(1)} + u_{1}^{(2)} = 0+1 = 1 \hspace{0.05cm},\]
  • at the second coding step  $(i = 2)$:
\[x_2^{(1)} =u_{2}^{(1)} + u_{1}^{(1)}+ u_{1}^{(2)} = 1 + 0 + 1 = 0\hspace{0.05cm},\hspace{0.4cm} x_2^{(2)} = u_{2}^{(2)} + u_{1}^{(1)} = 0+0 = 0\hspace{0.05cm},\hspace{0.4cm} x_2^{(3)} = u_{2}^{(1)} + u_{2}^{(2)}+ u_{1}^{(1)} = 1 + 0+0 =1\hspace{0.05cm},\]
  • at the third coding step  $(i = 3)$:
\[x_3^{(1)} =u_{3}^{(1)} + u_{2}^{(1)}+ u_{2}^{(2)} = 0+1+0 = 1\hspace{0.05cm},\hspace{0.4cm} x_3^{(2)} = u_{3}^{(2)} + u_{2}^{(1)} = 0+1=1\hspace{0.05cm},\hspace{0.4cm} x_3^{(3)} =u_{3}^{(1)} + u_{3}^{(2)}+ u_{2}^{(1)} = 0+0+1 =1\hspace{0.05cm},\]
  • and finally at the fourth coding step  $(i = 4)$:
\[x_4^{(1)} = u_{4}^{(1)} + u_{3}^{(1)}+ u_{3}^{(2)} = 1+0+0 = 1\hspace{0.05cm},\hspace{0.4cm} x_4^{(2)} = u_{4}^{(2)} + u_{3}^{(1)} = 1+0=1\hspace{0.05cm},\hspace{0.4cm} x_4^{(3)}= u_{4}^{(1)} + u_{4}^{(2)}+ u_{3}^{(1)} = 1+1+0 =0\hspace{0.05cm}.\]

Thus the encoded sequence is after the multiplexer:     $\underline{x} = (0, 1, 1, 0, 0, 1, 1, 1, 1, 1, 1, 0, \ \text{...})$.


Exercises for the chapter


Exercise 3.1: Analysis of a Convolutional Encoder

Exercise 3.1Z: Convolution Codes of Rate 1/2

References

  1. Elias, P.:  Coding for Noisy Channels.  In:  IRE Conv. Rec. Part 4, pp. 37-47, 1955.