Exercise 4.08Z: Basics about Interleaving

From LNTwww

Interleaver description

Interleaving is required,  for example,  for a channel with burst error characteristics in order to distribute the errors within the burst over a sufficiently large area so that they can subsequently be largely corrected  $($or at least detected$)$.

For turbo codes based on so-called  RSC encoder  $($"Recursive Systematic Convolutional Encoder"$)$  – and only such make sense – interleaving is essential also with the AWGN channel,  because then there are also always $($some$)$ input sequences,  which deliver only  "zeros"  in the output sequence after quite a few  "ones",  and that to infinity   ⇒   there are output sequences with very small Hamming weight.

If the bits of such input sequences are distributed over a wide range in the second encoder,  the problem can be largely eliminated by the interaction of both component decoders in the case of iterative symbol-wise decoding.

A general distinction is made between

  • block interleaver and
  • random interleaver.


In block interleaving  one fills a matrix with  $N_{\rm C}$  columns and  $N_{\rm R}$  rows column-by-column and reads the matrix row-by-row.  This deterministically scrambles a block of information with  $I_{\rm max} = N_{\rm C} \cdot N_{\rm R}$  bits.

On the right,  two interleavers are indicated and in graphical form by the assignment  $I_{\rm Out}(I_{\rm In})$.  These quantities represent the  "output sequence index"  and the  "input sequence index",  respectively.  It holds:

$$1 \le I_{\rm Out} \le I_{\rm max} \hspace{0.05cm},$$
$$1 \le I_{\rm In} \le I_{\rm max} \hspace{0.05cm}. $$

In the subtask  (1)  it is asked whether this is  "block interleaving"  or  "random interleaving".  The latter are discussed in the  "theory section"  but only very briefly.



Hints:

  • But other  $\rm LNTwww$  books also discuss interleaving,  including the book  "Examples of Communication Systems"  with reference to the



Questions

1

What interleaver type is shown in the graphic on the details page?

Block interleaving,
Random interleaving.

2

How many rows  ($N_{\rm R}$)  and columns  ($N_{\rm C}$)  does the upper "Interleaver matrix 1" have?

$N_{\rm R} \ = \ $

$N_{\rm C} \ = \ $

3

It holds   $\underline{u} = (1001'0001'1101'1101'0010'0111)$.  How does the scrambled sequence  $\underline{u}_{\pi}$ begin?    Note:   The quotation marks serve only as a reading aid.

$\underline{u}_{\pi} = (110'100'100'011'111'110'010'001' \text{...}\ )$,
$\underline{u}_{\pi} = (101'001'000'111'100'101'011'101'\text{...}\ )$.

4

The scrambled sequence be  $\underline{u}_{\pi} = (100'100'011'101'110'100'100'111)$.  What is the sequence after de-interleaving?

$\underline{u} = (1101'0010'0011'1111'1001'0001'\text{...}\ )$,
$\underline{u} = (1010'0100'0111'1001'0101'1101' \text{...}\ )$.


Solution

$4×3$  interleaver matrix

(1)  From the regular structure of the function  $I_{\rm Out}(I_{\rm In})$  one can see that it is a block interleaver   ⇒   Response 1.


(2)  The index  "1"  is output as the first character.  Further applies:

  • The index 5 is output as the second character   ⇒   $\underline{N_{\rm R} = 4}$.
  • The index 2 is output as the fourth character   ⇒   $\underline{N_{\rm C} = 3}$.


The upper graph shows for the  $4×3$  interleaver matrix:

  • the column-by-column write  $($red$)$,
  • the row-by-row readout  $($green$)$.


Interleaving

(3)  Correct is  the proposed solution 2:

  • The matrix is written column-by-column and read row-by-row.
  • After  $12$  bits,  the matrix is cleared and the procedure starts again.
  • The graphic shows that the solution suggestion 2 is correct.


De–interleaving

(4)  Correct is  the proposed solution 1:

  • In de-interleaving,  the matrix is written row-by-row and read column-by-column.
  • The graphic shows that here the solution suggestion 1 is correct.