How many bits of error can be corrected by LDPC method?
md3848 
How many bits of error can be corrected by LDPC method?
Author
Message
md3848 ki0bk
How many miles per gallon can I get with my toyota pick up?
Answer: it depends on how it is driven!
I don't think your question can be answered given the info provided
Jim
Last Edited: Tue. Dec 31, 2019 - 01:32 PM
awneil 
Level: 10k+ Postman
Joined: Fri. Jul 1, 2005
Posts: 25278 View posts
Location: Basingstoke, Hampshire, UK
md3848 wrote:
LDPC method
You mean this: https://en.wikipedia.org/wiki/Low-density_parity-check_code ?
Which, as Jim says, leads to the answer: "It depends".
md3848 awneil wrote:
You mean this: https://en.wikipedia.org/wiki/Low-density_parity-check_code ?
+
Well, Depends on what? Would you explain a little more?
Last Edited: Tue. Dec 31, 2019 - 04:34 PM
awneil
Level: 10k+ Postman
Joined: Fri. Jul 1, 2005
Posts: 25278 View posts
Location: Basingstoke, Hampshire, UK
You're the one that's asking - what research have you done?
From what I can tell from that article, "LDPC" is not just one single thing - it's a whole class of a certain type of codes.
Therefore, the performance is going to depend on the particular type of LPDC you choose, and how you implement it.
So your question is a bit like saying "how many bits of error can a CRC detect" - that depends on which particular CRC you choose!
ka7ehk
This is kind of interesting... historically, parity has been a very weak error management process. When applied byte by byte, a parity violation could tell you that there has been at least a single bit error. But, it was very possible to have errors of two or more bits in which parity is preserved.
LDPC seems to be potentially a lot stronger than that (re: above referenced Wikipedia article). And, there appears to be several classes of applications in which it is superior to turbo codes.
I wish that I understood this stuff better. Maybe, some time, if I have nothing better to do, ... ha, ha, ha!
Jim
ron_sutherland
ka7ehk wrote:
I wish that I understood this stuff better
Me too. I use to work with ATCA-9305:
https://www.smartembedded.com/ec/product-datasheets/1398193020-ds.pdf/Artesyn-ATCA-9305.pdf
And a few boards like it, it had the LDPC 10GBase-T network fabric on the backplane in which it plugged in. I mostly recall struggles with the error correction code (ECC) memory that would self-correct (LDPC?). To detect if errors were happening during burn-in, we had to load a bare metal test application from the CPU manufacturer rather than running the Linux that was used for other tests. It needed some timing values that caused the struggle. At any rate, once the timing was figured out it was flawless, I don't recall seeing errors, and I probably unloaded 5000 of those things from burn-in.