FPGA Central

Hello, I am honing my skills on the Physical layer of communication
systems. I request the knowledgeable members of this group to help me
with the following topics. I'd appreciate that, and I believe this may
help other people who would want similar knowledge.

1) On what criterion do we select the original data bits (like it is
244 bits for UMTS 12.2 Kbps)

2) What would be the optimal length for CRC for any 'x' bits like x =
244 or 500

3) How should we determine the length of the interleaver and the
interleaving permutation for output
of the interleaver?

4) How can we know how many bits will our convolutional encoder
correct (for block it has been
explained here, I guess)

5) In the models given with Matlab for CDMA, spreading is done after
modulation, however in
standards, it is done after modulation

If anyone has more information like this, It'd be really appreciated
as these are the basic question one asks when self studying...

(I'm a student by the way not a pro comm engr)

On Dec 29, 10:44*am, Communications_engineer
<communications_engin...@yahoo.com> wrote:
> Hello, I am honing my skills on the Physical layer of communication
> systems. I request the knowledgeable members of this group to help me
> with the following topics. I'd appreciate that, and I believe this may
> help other people who would want similar knowledge.
>
> 1) On what criterion do we select the original data bits (like it is
> 244 bits for UMTS 12.2 Kbps)

There are three criteria: cost, schedule, and performance. You get to
dictate any two.

>
> 2) What would be the optimal length for CRC for any 'x' bits like x =
> 244 or 500

The optimal length is that which meets your requirements.

>
> 3) How should we determine the length of the interleaver and the
> interleaving permutation for output
> * * of the interleaver?

In a manner consistent with governing procedures and doctrine.

>
> 4) How can we know how many bits will our convolutional encoder
> correct (for block it has been
> * * explained here, I guess)

By reading a book.

>
> 5) In the models given with Matlab for CDMA, spreading is done after
> modulation, however in
> * * standards, it is done after modulation
>

And your question is?

John

On Dec 29, 10:59*pm, John <sampson...@gmail.com> wrote:
> On Dec 29, 10:44*am, Communications_engineer
>
> <communications_engin...@yahoo.com> wrote:
> > Hello, I am honing my skills on the Physical layer of communication
> > systems. I request the knowledgeable members of this group to help me
> > with the following topics. I'd appreciate that, and I believe this may
> > help other people who would want similar knowledge.
>
> > 1) On what criterion do we select the original data bits (like it is
> > 244 bits for UMTS 12.2 Kbps)
>
> There are three criteria: cost, schedule, and performance. You get to
> dictate any two.
>

Ok, so whats wrong with having 245 bits or 200 bits what would it
effect the cost, schedule, and performance.

>
> > 2) What would be the optimal length for CRC for any 'x' bits like x =
> > 244 or 500
>
> The optimal length is that which meets your requirements.
>

For what kind of requirements would an 8-bit CRC or 16 or 32-bit CRC
suffice

> > 3) How should we determine the length of the interleaver and the
> > interleaving permutation for output
> > * * of the interleaver?
>
> In a manner consistent with governing procedures and doctrine.
>

I understand that the interleaving period is really based on the
coherence time & the channel conditions, right? So coherence time we
can find but what about channel conditions

>
>
> > 4) How can we know how many bits will our convolutional encoder
> > correct (for block it has been
> > * * explained here, I guess)
>
> By reading a book.
>

Would you suggest a good book, something that gets to the point rather
than having to read 500 pages before getting the answer

>
> > 5) In the models given with Matlab for CDMA, spreading is done after
> > modulation, however in
> > * * standards, it is done after modulation
>
> And your question is?
>

sorry, I'll repeat "5) In the models given with Matlab (examples) for
PN sequences, spreading is done after modulation, however in 3GPP
standards (25.212), spreading is done BEFORE modulation"

If there is a good book regarding these questions, please let me know

> John

See you, John

On Wed, 31 Dec 2008 00:42:05 -0800 (PST), Communications_engineer
<[email protected]> wrote:

>On Dec 29, 10:59*pm, John <sampson...@gmail.com> wrote:
>> On Dec 29, 10:44*am, Communications_engineer
>>
>> <communications_engin...@yahoo.com> wrote:
>> > Hello, I am honing my skills on the Physical layer of communication
>> > systems. I request the knowledgeable members of this group to help me
>> > with the following topics. I'd appreciate that, and I believe this may
>> > help other people who would want similar knowledge.
>>
>> > 1) On what criterion do we select the original data bits (like it is
>> > 244 bits for UMTS 12.2 Kbps)
>>
>> There are three criteria: cost, schedule, and performance. You get to
>> dictate any two.
>>
>
>Ok, so whats wrong with having 245 bits or 200 bits what would it
>effect the cost, schedule, and performance.

That depends entirely on your system requirements and how the work is
done.

>> > 2) What would be the optimal length for CRC for any 'x' bits like x =
>> > 244 or 500
>>
>> The optimal length is that which meets your requirements.
>>
>
>For what kind of requirements would an 8-bit CRC or 16 or 32-bit CRC
>suffice

There's no need to use a CRC larger than what is needed to provide the
reliability you desire.

>> > 3) How should we determine the length of the interleaver and the
>> > interleaving permutation for output
>> > * * of the interleaver?
>>
>> In a manner consistent with governing procedures and doctrine.
>>
>
>I understand that the interleaving period is really based on the
>coherence time & the channel conditions, right? So coherence time we
>can find but what about channel conditions

There are models available for different applications, but they're
just models. You can go out and measure propogation in a number of
environments that you think will be representative of your
application, or you can find a model that's close.

If you know the coherence time you must already know something about
the channel.

>> > 4) How can we know how many bits will our convolutional encoder
>> > correct (for block it has been
>> > * * explained here, I guess)
>>
>> By reading a book.
>>
>
>Would you suggest a good book, something that gets to the point rather
>than having to read 500 pages before getting the answer

Yes, I would suggest using a good book rather than one that's too
obtuse.

The sorts of details you're asking about don't have answers that are
easily packaged or spoon-fed. Some system understanding is required,
so it may take more than looking up a page or a paragraph in book or
two. Or three. Or more.

>
>>
>> > 5) In the models given with Matlab for CDMA, spreading is done after
>> > modulation, however in
>> > * * standards, it is done after modulation
>>
>> And your question is?
>>
>
>sorry, I'll repeat "5) In the models given with Matlab (examples) for
>PN sequences, spreading is done after modulation, however in 3GPP
>standards (25.212), spreading is done BEFORE modulation"

I still don't see a question there.

Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.ericjacobsen.org

Blog: http://www.dsprelated.com/blogs-1/hf/Eric_Jacobsen.php

Communications_engineer <[email protected]> wrote:

>I understand that the interleaving period is really based on the
>coherence time & the channel conditions, right? So coherence time we
>can find but what about channel conditions

Actually there are a ton of other factors that might influence
choice of interleaving block size. Off the top of my head:
(1) even on a AWGN channel you may need interleaving to spread
errors among FEC codewords, or spread them around within a
convolutional codeword; (2) latency needs by the application
dictate how long an interleaver you can use.

The interleaver length may or may not be short relative to
the coherence time. I've seen systems designed either way.

Steve

On Jan 5, 3:28*am, spop...@speedymail.org (Steve Pope) wrote:
> Communications_engineer*<communications_engin...@y ahoo.com> wrote:
> >I understand that the interleaving period is really based on the
> >coherence time & the channel conditions, right? So coherence time we
> >can find but what about channel conditions
>
> Actually there are a ton of other factors that might influence
> choice of interleaving block size. *Off the top of my head:
> (1) even on a AWGN channel you may need interleaving to spread
> errors among FEC codewords, or spread them around within a
> convolutional codeword; (2) latency needs by the application
> dictate how long an interleaver you can use. *
>
> The interleaver length may or may not be short relative to
> the coherence time. *I've seen systems designed either way.
>
> Steve

I actually looked up regarding my questions. We design interleavers
(the size of the interleaver is chosen so the interleaver attains its
required depth, and the depth is in turn found by the Average Fade
Time Duration, an analysis of the time the signal is below a power
threshold required for a BER constraint) and FEC correction capability
is dictated by the same i.e. calculate how many bits are in error in
that given Fade Time Duration

Now this fade that we create is based on the channel model we use.
Rayleigh, Rician, Nakagami etc

However steve I'm not too sure about using interleaver with AWGN?

Communications_engineer <[email protected]> wrote:

>However steve I'm not too sure about using interleaver with AWGN?

For a classic example of an interleaver being necessary on
an AWGN channel, look at the decades-old Deep Space Network
concatenated Reed-Solomon / Convolutional coding standards.
You will see a depth 5 interleaver is needed between the
two codes. The channel is AWGN (being deep space), but the
performance definitely suffers if the interleaver is not there.

Steve

On Jan 13, 12:56*am, spop...@speedymail.org (Steve Pope) wrote:
> Communications_engineer*<communications_engin...@y ahoo.com> wrote:
> >However steve I'm not too sure about using interleaver with AWGN?
>
> For a classic example of an interleaver being necessary on
> an AWGN channel, look at the decades-old Deep Space Network
> concatenated Reed-Solomon / Convolutional coding standards.
> You will see a depth 5 interleaver is needed between the
> two codes. *The channel is AWGN (being deep space), but the
> performance definitely suffers if the interleaver is not there.
>
> Steve

Hmm.. well you are right I suppose. But my view was that since White
noise has almost equal noise power for entire bandwidth of the channel
and more so what would be use of interleaver as it protects against
contiguous bursts of noise, and really there aren't any bursts in
AWGN.

Also Steve can you tell me what should be the optimal length of a CRC?
To a beginner, because I find it easier to learn if I already have a
direction to go into rather than going through a lot of stuff that
isn't important (atleast as far as CRC is concerned) An example for
the use of 8bit, 16-bit CRC would be highly appreciated, or some
refernce of book tutorial

Thanks

Communications_engineer <[email protected]> wrote:

>On Jan 13, 12:56*am, spop...@speedymail.org (Steve Pope) wrote:

>> Communications_engineer*<communications_engin...@y ahoo.com> wrote:

>> >However steve I'm not too sure about using interleaver with AWGN?

>> For a classic example of an interleaver being necessary on
>> an AWGN channel, look at the decades-old Deep Space Network
>> concatenated Reed-Solomon / Convolutional coding standards.
>> You will see a depth 5 interleaver is needed between the
>> two codes. *The channel is AWGN (being deep space), but the
>> performance definitely suffers if the interleaver is not there.

>Hmm.. well you are right I suppose. But my view was that since White
>noise has almost equal noise power for entire bandwidth of the channel
>and more so what would be use of interleaver as it protects against
>contiguous bursts of noise, and really there aren't any bursts in
>AWGN.

Right, in the example I gave the channel has no bursts of
noise, but there are still bursts of errors at the convolutional
decoder output, requiring interleaving.

Somewhat similarly, symbol interleaving in an OFDM system isn't
protecting agaist burts of errors either.

>Also Steve can you tell me what should be the optimal length of a CRC?

Too general a question to answer. But if you know the block
length and the acceptable probability of a failure to detect
errors, that would allow you to calculate the minimum CRC
field length, based purely on combinatorics.

Steve

On Tue, 13 Jan 2009 17:57:38 +0000 (UTC), [email protected]
(Steve Pope) wrote:

>Communications_engineer <[email protected]> wrote:
>
>>On Jan 13, 12:56*am, spop...@speedymail.org (Steve Pope) wrote:
>
>>> Communications_engineer*<communications_engin...@y ahoo.com> wrote:
>
>>> >However steve I'm not too sure about using interleaver with AWGN?
>
>>> For a classic example of an interleaver being necessary on
>>> an AWGN channel, look at the decades-old Deep Space Network
>>> concatenated Reed-Solomon / Convolutional coding standards.
>>> You will see a depth 5 interleaver is needed between the
>>> two codes. *The channel is AWGN (being deep space), but the
>>> performance definitely suffers if the interleaver is not there.
>
>>Hmm.. well you are right I suppose. But my view was that since White
>>noise has almost equal noise power for entire bandwidth of the channel
>>and more so what would be use of interleaver as it protects against
>>contiguous bursts of noise, and really there aren't any bursts in
>>AWGN.
>
>Right, in the example I gave the channel has no bursts of
>noise, but there are still bursts of errors at the convolutional
>decoder output, requiring interleaving.

Exactly. Even with distributed input errors a Viterbi decoder tends
to produce clumpy output errors at low SNR. The interleaver helps
the outer RS clean those up.

>Somewhat similarly, symbol interleaving in an OFDM system isn't
>protecting agaist burts of errors either.

Hmmm...I think it is, but the mechanism is a bit different. i.e., at
the input of the inner-most FEC decoder the errors will be less clumpy
when a symbol interleaver is used than without. Kind of a roundabout
way of saying BICM is better than no interleaving with OFDM from that
perspective.

>>Also Steve can you tell me what should be the optimal length of a CRC?
>
>Too general a question to answer. But if you know the block
>length and the acceptable probability of a failure to detect
>errors, that would allow you to calculate the minimum CRC
>field length, based purely on combinatorics.
>
>Steve
Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.ericjacobsen.org

Blog: http://www.dsprelated.com/blogs-1/hf/Eric_Jacobsen.php