Forums

Reed solomon with erasures

Started by shimi July 11, 2008
Hello,

I'm interested in using RS- Reed solomon codes with erasures.

I know the RS code can correct 2g+e=n-k, where g is erasure and e is
error.

My system can find the erasures with high probability.
My question is, if the system can detect X erasures but 50% of them are
correct.
Meaning they were mistakenly detected as erasures.
Will I earn something using decoder with erasure or using regular decoding
?

Another question, if from the X erasures 75% are correct.
Will I earn something ?


thanks

Itzic


On Jul 11, 11:53 am, "shimi" <itzic...@yahoo.com> wrote:
> Hello, > > I'm interested in using RS- Reed solomon codes with erasures. > > I know the RS code can correct 2g+e=n-k, where g is erasure and e is > error. > > My system can find the erasures with high probability. > My question is, if the system can detect X erasures but 50% of them are > correct. > Meaning they were mistakenly detected as erasures. > Will I earn something using decoder with erasure or using regular decoding > ? > > Another question, if from the X erasures 75% are correct. > Will I earn something ? > > thanks > > Itzic
Probably depends upon the error rate of the mistakenly-classified bits. If all of the bits that your receiver incorrectly marks as erasures are correct, then you would be best served to include them. If the error rate is high for those bit positions, then you're best off just marking them as erasures. Other error rates give you a case in between the two extremes. Do you have any more information? Jason

shimi wrote:
> Hello, > > I'm interested in using RS- Reed solomon codes with erasures. > > I know the RS code can correct 2g+e=n-k, where g is erasure and e is > error. > > My system can find the erasures with high probability. > My question is, if the system can detect X erasures but 50% of them are > correct. > Meaning they were mistakenly detected as erasures. > Will I earn something using decoder with erasure or using regular decoding > ? > > Another question, if from the X erasures 75% are correct. > Will I earn something ?
In general, any apriori knowledge about the errors improves the operation of the decoder. There is a family of algorithms of different complexity for decoding of the RS codes with the use of the reliability information for every symbol. Depending on the algorithm and the error rate, you can expect the gain of 0.5 to 3dB compared to the hard decision algebraic decoding. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com
On 2008-07-11, shimi <itzic100@yahoo.com> wrote:
> I'm interested in using RS- Reed solomon codes with erasures. > > I know the RS code can correct 2g+e=n-k, where g is erasure and e is > error.
Actually it's 2e+g=n-k. You can correct more erasures than errors. Think about it this way: The error correcting power can be applied to both *finding* and *fixing* (the error case) or just *fixing* if you supply the location information using outside information.
> My system can find the erasures with high probability.
You will always gain from that. -- Ben Jackson AD7GD <ben@ben.com> http://www.ben.com/
Hi All

Sorry to ask different question.

Which is better: Reed Solomon coding or convolutional code with SOVA
decoding?

Thanks

Regards,

Chintan
On 2008-07-11, cpshah99 <cpshah99@rediffmail.com> wrote:
> > Which is better: Reed Solomon coding or convolutional code with SOVA > decoding?
It will depend on all kinds of things. For a primer try the telecom part of NASA's "basics of space flight": http://www2.jpl.nasa.gov/basics/bsf10-1.html -- Ben Jackson AD7GD <ben@ben.com> http://www.ben.com/

cpshah99 wrote:

> Which is better: Reed Solomon coding or convolutional code with SOVA > decoding?
It depends. BTW, Reed Solomon codes can be decoded with SOVA as well. Vladimir Vassilevsky DSP and Mixed Signal Design Consultant http://www.abvolt.com