Forums

Erasures decoding of Reed Solomon codes

Started by Mike McLernon August 31, 2005
On Tue, 24 May 2016 12:14:12 -0500, Tim Wescott
<seemywebsite@myfooter.really> wrote:

>On Mon, 23 May 2016 12:46:22 -0500, Tim Wescott wrote: > >> On Sat, 21 May 2016 14:55:38 -0700, kortas.manel wrote: >> >>> Le mercredi 31 ao&ucirc;t 2005 20:35:16 UTC+1, Mike McLernon a &eacute;crit&nbsp;: >>>> Hi, >>>> >>>> I'm building an erasures decoder for Reed Solomon codes, and I'm >>>> having trouble when the number of erasures (without errors) exceeds >>>> (n-k)/2, the error-correcting capability of the code. When my code >>>> has no errors, and only erasures, I would expect the error locator >>>> polynomial to be uniquely 1. >>>> When I have up to (n-k)/2 erasures, that's exactly what happens. >>>> However, when the number of erasures exceeds (n-k)/2, my error locator >>>> polynomial exhibits unusual behavior. I run into the following >>>> circumstances in these cases: >>>> >>>> 1) the polynomial has no roots in the field of the code, or 2) >>>> the polynomial has repeated roots, or 3) the polynomial's roots are >>>> a subset of the erasure locations, or 4) the polynomial's roots are >>>> different from the erasure locations. >>>> >>>> Scenarios 1-3 seem reasonable, in that I can discard the information >>>> in the error locator polynomial. I already have that information from >>>> the erasure locator polynomial. However, scenario 4 is troublesome, >>>> and I'm not quite sure how to handle that one. >>>> >>>> Any help would be appreciated. >>>> >>>> Mike McLernon >>> can any one provide me the source code of reed solomon in matlab with >>> BM or euclidien or PGZ method please? thx in advance >> >> Not for free, generally, and not if you're a student, ever. We and you >> both benefit from you doing your own homework. If you can't do your own >> homework -- switch majors. > >Boy, that was harsh. And I forgot the rider: we're pretty much >universally opposed to _doing_ your homework for you, but there's quite a >bit of us (myself included) who are delighted to _help_ you to understand >your homework so that you might do it yourself. > >We _won't_ help to raise up a generation of ignorant engineers who spend >a minimal amount of time screwing things up before becoming managers and >_really_ screwing things up. But we're happy to help raise up the next >generation of _competent_ engineers.
That's about as well as I've seen that put for a while.