Interleaving with soft-decision

Hi,

I have a problem with interleaving and soft-decision. First, i explain
what i have realized.

I have done a 16FSK receiver with FEC decoded Viterbi 1/2 and k=7.

Then, a symbol are 4 bits and i have 16 tones for 16 symbols:

For example: 

1001 is the symbol 9 and the tone number 9.
0010 is the symbol 2 and the tone number 2.

(Actually, i also use Gray Code, but that's is not important now.)

Using the FFT algoritm I can get the energy of each tone, then i have the
METRIC for soft-decision decoder.

But my big problem is: If i use an interleaver for separate the bits, then
i can not recover the metric of original symbol, because thats metric 
depend of four differents symbols.

I explain better: 

Bits original:                       A C E G I J ....etc... 

Bits coded:                 "ABCD EFGH IJKL MOPQ RSTU VWXY Z012 3456....."

Symbols coded(or tones):      4    5    0    12   9   15    13    4 ...  

Diagonal interleaver:

              AEIMQUY...                      
              BFJNRVZ...                      
              CGKOSW0...                      
              DHLPTX1... 

Bits interleaving:            AFKP EJOT INSX MRW1 ...
Symbols(or tones) to send:     5    11   13  7   .... for example   

                       --------TO THE ANTENNA---------

In the receiver i can get metric of symbols: "5", "11", "13", etc...., but
what i need is the metric for symbols "4", "5", "0" etc.... because thats
symbols are just the coded symbols.

¿Any idea?

 

JAlbertoDJ <nietorosdj@yahoo.es> wrote:

>Hi,
>
>Using the FFT algoritm I can get the energy of each tone, then i have the
>METRIC for soft-decision decoder.

Why would one use an FFT here?

>But my big problem is: If i use an interleaver for separate the bits, then
>i can not recover the metric of original symbol, because thats metric 
>depend of four differents symbols.

It is unfortunate that searching groups.google.com is not fuctional,
because I answered a very similar question here several months
ago in some detail.  You need to compute the a-priori probability for 
each received bit.  


Steve

JAlbertoDJ wrote:

> Hi,
> 
> I have a problem with interleaving and soft-decision. First, i explain
> what i have realized.
> I have done a 16FSK receiver with FEC decoded Viterbi 1/2 and k=7.
> Then, a symbol are 4 bits and i have 16 tones for 16 symbols:

This has "homework" written over it.

Just to add to what Steve Pope noted, your channel coding is very 
unoptimal. Consider convolutional base 4 code.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

>Why would one use an FFT here?

Because it is more quickly of correlation for 16 tones. Also, tones are
continuous phase, then with one sample for each tone i have information of
all his energy.

>
>It is unfortunate that searching groups.google.com is not fuctional,
>because I answered a very similar question here several months
>ago in some detail.  You need to compute the a-priori probability for 
>each received bit.  
>
>
>Steve
>

Could you tell me the link to thats information?

thenk you

>
>
>Just to add to what Steve Pope noted, your channel coding is very 
>unoptimal. Consider convolutional base 4 code.
>
>

base 4??? i dont understand. Could you explain better?

JAlbertoDJ wrote:

>>
>>Just to add to what Steve Pope noted, your channel coding is very 
>>unoptimal. Consider convolutional base 4 code.
> 
> base 4??? i dont understand. Could you explain better?

Dual-K codes.
Proakis 8.2.6.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

>
>
>JAlbertoDJ wrote:
>
>>>
>>>Just to add to what Steve Pope noted, your channel coding is very 
>>>unoptimal. Consider convolutional base 4 code.
>> 
>> base 4??? i dont understand. Could you explain better?
>
>Dual-K codes.
>Proakis 8.2.6.
>
>
>Vladimir Vassilevsky
>DSP and Mixed Signal Design Consultant
>http://www.abvolt.com
>
>

Unfortunally, i have not that book and i dont find information about
Dual-K codes.

But i am reading something about TURBO-CODES and Tutbo-Codes have an
interleaver inside.  

Then, could be it the solution?  Are turbo-codes efficients for 16CPFSK?

JAlbertoDJ wrote:

>>
>>JAlbertoDJ wrote:
>>
>>
>>>>Just to add to what Steve Pope noted, your channel coding is very 
>>>>unoptimal. Consider convolutional base 4 code.
>>>
>>>base 4??? i dont understand. Could you explain better?
>>
>>Dual-K codes.
>>Proakis 8.2.6.
>>
> 
> Unfortunally, i have not that book and i dont find information about
> Dual-K codes.
> 
> But i am reading something about TURBO-CODES and Tutbo-Codes have an
> interleaver inside.  
> 
> Then, could be it the solution?  Are turbo-codes efficients for 16CPFSK?

The chasm of your ignorance is really bottomless.

If you don't have The Book Of Proakis, then you should go and get it. 
And don't ask any questions before you read The Book from the first to 
the last page.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

>
>
>JAlbertoDJ wrote:
>
>>>
>>>JAlbertoDJ wrote:
>>>
>>>
>>>>>Just to add to what Steve Pope noted, your channel coding is very 
>>>>>unoptimal. Consider convolutional base 4 code.
>>>>
>>>>base 4??? i dont understand. Could you explain better?
>>>
>>>Dual-K codes.
>>>Proakis 8.2.6.
>>>
>> 
>> Unfortunally, i have not that book and i dont find information about
>> Dual-K codes.
>> 
>> But i am reading something about TURBO-CODES and Tutbo-Codes have an
>> interleaver inside.  
>> 
>> Then, could be it the solution?  Are turbo-codes efficients for
16CPFSK?
>
>The chasm of your ignorance is really bottomless.
>
>If you don't have The Book Of Proakis, then you should go and get it. 
>And don't ask any questions before you read The Book from the first to 
>the last page.
>
>
>Vladimir Vassilevsky
>DSP and Mixed Signal Design Consultant
>http://www.abvolt.com
>
>
>
Thank you very much

Vladimir Vassilevsky <nospam@nowhere.com> wrote in news:MfqdnYYF2sD-
ygHXnZ2dnUVZ_g-dnZ2d@giganews.com:

> 
> The chasm of your ignorance is really bottomless.
> 
> If you don't have The Book Of Proakis, then you should go and get it. 
> And don't ask any questions before you read The Book from the first to 
> the last page.


As a previous victim of your verbal abuse, I don't feel so bad because I 
now know that you didn't specifically single me out. Given your above 
comment, there is no longer any doubt whatsoever my mind that arrogance and 
verbal abuse are the primary, if not the only ingredients that comprise 
your sad and pathetic MO.  The bottomless ignorance, which you assign to 
the persone who naively came here for help, is only overshadowed by your 
unbounded arrogance and your uncontrolled abusive tongue.  It is beyond me 
to comprehend why you are here and why you respond in the offensive manner 
that you do, but then I'm not a psychiatrist.