Cell Relay Archive

Cell Relay Retreat>List Archive>month:1999-Oct> msg00028



[Date Prev][Date Next][Thread Prev][Thread Next]  
  [Date Index][Thread Index][Author Index][Subject Index]

Re: Single bit header error correction in ATM Specific TC

  • From: murray@pa.dec.com (Hal Murray)
  • Date: 10 Oct 1999 06:42:05 GMT
  • Distribution: world
  • Organization: Compaq Systems Research Center



> you are correct in noting that i was assuming random independently
> distributed bit errors.  but clumps of errors should be detectable
> by the crc generator polynomial if it is reasonably specified.

Sorry.  I should have said more...

The context of the discussion was correcting errors in the header.

You said that if the link had a bit error rate of 10^-7 then there
wouldn't be much chance of false corrections.

I was trying to say that if the errors come in clumps (which happens
in many physical systems) then the probability of false corrections
is no longer very low.

-----

I think a good N-bit CRC will detect all error patterns where all
the bit errors are within a clump of N-1 adjacent bits.  (That
is you could fix the packet by XORing with an N-1 bit number that
was aligned correctly.)  Maybe it's N rather than N-1.

-----

Does anybody know of any data on the distribution of errors on
real links?

-- 
These are my opinions, not necessarily my employers.