Prev: I GOT $2500 FROM PAYPAL....
Next: Vocal pitch detection
From: Jason on 3 Aug 2010 10:49
On Aug 3, 10:46 am, "cpshah99" <cpshah99(a)n_o_s_p_a_m.rediffmail.com>
wrote:
> >Selecting a coding scheme is a BIG question and there is a zillion of
> >things to consider, starting from the channel properties and error
> >probabilities and up to the amount of royalties that has to be paid for
> >the intellectual property.
>
> OK. So if it is about royalties, then it is better to avoid Turbo code.
>
> >If someone does what seems obvious and what averybody else does, he will
> >certainly be fucked.
>
> Well that depends on the investor. I prefer long term investments and a
> careful investment in any blue chip stock will deliver profit sooner or
> later.
>
> Cheers,
>
> Chintan

Tell that to the circa-2000 shareholders of Enron.

Jason
From: Clay on 3 Aug 2010 10:58
On Aug 3, 10:31 am, Vladimir Vassilevsky <nos...(a)nowhere.com> wrote:
> cpshah99 wrote:
>
> >>cpshah99 wrote:
>
> >>>Hello All
>
> >>>Given the task to design a receiver which should operate reasonably
>
> > well
>
> >>>under moderate to severe channel conditions, which coding scheme you
>
> > will
>
> >>>select from hardware implementation point of view: Turbo code or LDPC?
> >>>Your opinion matters a lot.
>
> >>>Chintan
>
> >>A young guy and young girl came to the rabbi to ask for his advice.
>
> >>A girl:
>
> >>"Tomorrow is my first night with my fiance. What kind of night dress
> >>should I put on? Should it be a long dress, or a short dress, or an open
> >>dress, or a closed dress, or whatever?"
>
> >>A boy:
>
> >>"I got some money. Will you please give me an advice what to do with it:
> >>maybe, invest into something, or buy some real estate, or gold, or put
> >>it in a bank?"
>
> >>The rabbi:
>
> >>"My dear children, I have one answer for both of you: whatever you do,
> >>you will certainly be fucked."
>
> >>VLV
>
> > %%%
>
> > HI Vlad
>
> > Wassup?? How is life treating you?
>
> > You had posted the same ans long time back when someone asked a question
> > which was stupid according to you.
>
> > Anyways, you guys work on hardware a lot. So which scheme would you select
> > if you were to design a modem.
>
> Selecting a coding scheme is a BIG question and there is a zillion of
> things to consider, starting from the channel properties and error
> probabilities and up to the amount of royalties that has to be paid for
> the intellectual property.
>
> > But looking at current markets, if someone invests in Gold, I think he will
> > make some profit for sure.
>
> If someone does what seems obvious and what averybody else does, he will
> certainly be fucked.
>
> VLV- Hide quoted text -
>
> - Show quoted text -


Concerning gold, in the US under the new healthcare legislation, any
sale of gold over $600 to a business must be reported to the IRS. So
if you buy gold, keep your receipts so you can defend and document a
proper basis price and only pay tax on the increase of the gold's
value when you finally sell it. This means you must 1099 the seller!
While the profits were always subject to tax, now gold dealers will be
1099ing you when you sell them gold. This requirement starts Jan 1,
2012. This is in section 9006 of the healthcare bill. Section 9006
requires any business to 1099 any entity that is paid over $600 in a
year or the business will not be able to write off the costs. Imagine
in your small business, your having to 1099 the rental car companies,
the airlines, hotels, etc for just about all expenses. You only have
to exceed $600 in a year to any entity to need to 1099 that entity.
Talk about a paperwork nightmare.

Clay


From: Muzaffer Kal on 3 Aug 2010 11:39
On Tue, 03 Aug 2010 07:04:20 -0500, "cpshah99"
<cpshah99(a)n_o_s_p_a_m.rediffmail.com> wrote:

>
>Hello All
>
>Given the task to design a receiver which should operate reasonably well
>under moderate to severe channel conditions, which coding scheme you will
>select from hardware implementation point of view: Turbo code or LDPC?

Given that at this point in time, it is pretty much accepted that
Turbo Codes and Gallager codes are practically the same thing:
http://www.ee.usyd.edu.au/~lyh/Publications/Journal%20Paper/PIEEE01.pdf
http://www.ima.umn.edu/csg/bib/bib27.0399mack.pdf

the question which remains is which of these coding schemes are
suitable for your purposes. You ask "from hardware implementation
point of view"; did you answer all other relevant questions properly?

Assuming yes, my answer would be LDPC as it has been shown to have a
lower noise floor. But you should keep in mind that low latency
implementations of LDPC in fully parallel mode are extremely expensive
routing-wise so if you're not doing your hardware in a high-end
process (<= 45nm) you won't get good results. After all there is a
reason Gallager codes were forgotten for more than 25 years at some
point.
--
Muzaffer Kal

DSPIA INC.
ASIC/FPGA Design Services

http://www.dspia.com
From: Steve Pope on 3 Aug 2010 11:48
Muzaffer Kal <kal(a)dspia.com> wrote:

>Assuming yes, my answer would be LDPC as it has been shown to have a
>lower noise floor. But you should keep in mind that low latency
>implementations of LDPC in fully parallel mode are extremely expensive
>routing-wise so if you're not doing your hardware in a high-end
>process (<= 45nm) you won't get good results. After all there is a
>reason Gallager codes were forgotten for more than 25 years at some
>point.

It is true that on average LDPC codes perform slightly better
than parallel-concatenated Turbo codes of similar block length.
But usually the error-floor issue in the Turbo code can be solved
by selecting the right interleaver, and at that point the performance
difference is small, on the order of a tenth of a dB or so.

In my experience the Turbo codes are more flexible -- it is easier
to dial in the code rate and still have close to optimal performance --
and they are less tricky to implement.

Steve
From: Eric Jacobsen on 3 Aug 2010 12:52
On Tue, 3 Aug 2010 15:48:58 +0000 (UTC), spope33(a)speedymail.org (Steve
Pope) wrote:

>Muzaffer Kal <kal(a)dspia.com> wrote:
>
>>Assuming yes, my answer would be LDPC as it has been shown to have a
>>lower noise floor. But you should keep in mind that low latency
>>implementations of LDPC in fully parallel mode are extremely expensive
>>routing-wise so if you're not doing your hardware in a high-end
>>process (<= 45nm) you won't get good results. After all there is a
>>reason Gallager codes were forgotten for more than 25 years at some
>>point.
>
>It is true that on average LDPC codes perform slightly better
>than parallel-concatenated Turbo codes of similar block length.
>But usually the error-floor issue in the Turbo code can be solved
>by selecting the right interleaver, and at that point the performance
>difference is small, on the order of a tenth of a dB or so.
>
>In my experience the Turbo codes are more flexible -- it is easier
>to dial in the code rate and still have close to optimal performance --
>and they are less tricky to implement.
>
>Steve

Turbo Codes and LDPCs are definitely close enough in performance that
comparisons can get very difficult. For the most part the
differences in performance are so small that they're essentially
indistinguishable.

There are some differences that may matter to an architect or
implementer, though:

1. The IP issue has already been mentioned. That's not to say that
certain forms of LDPC are IP-free, just that in general LDPCs are,
IMHO, less encumbered than TCs. Consultation with relevant entities
will be key here.

2. The sort of Turbo Code used may change significantly over block
size, i.e., if your system deals with mainly large blocks, a different
TC may be optimal than if the system deals primarily with small
blocks. For flexible block sizes my personal view is that LDPCs are
much better, but both have been used effectively.

3. LDPCs, in general, don't need an interleaver. This can be an
architectural advantage (and potentially reduce latency), especially
for an OFDM system.

4. LDPCs can be trickier to implement, as Steve mentioned. This is
strictly an implementation issue that may vary greatly as
implementation technology changes.

Both TCs and LDPCs have been used effectively for low SNR channels.
Beware of those who claim a clear winner, but for any given
application the particular system/design/architecture/whatever
constraints may favor one over the other. So it really does depend
a lot on the project specifics.


Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.abineau.com
First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4
Prev: I GOT $2500 FROM PAYPAL....
Next: Vocal pitch detection