One of the disadvantages of some line codes like NRZ-L or RZ is that if a sudden change in polarity occurs, then all received bits will be erroneous (flipped).
- How can such a sudden change in polarity occur in practice?
- And is it really a problem?
\$\begingroup\$
\$\endgroup\$
9
-
\$\begingroup\$ U are thinking of NRZ-I and this was invented by IBM in 1956 for mag tapes and is obsolete. \$\endgroup\$Tony Stewart EE since 1975– Tony Stewart EE since 19752019-06-21 07:01:00 +00:00Commented Jun 21, 2019 at 7:01
-
\$\begingroup\$ Thanks. Are you implying that polarity inversion is not a problem anymore? I know NRZ-I, and I gave NRZ-L as an example of a line code that is affected by polarity inversion. I am asking about the inversion itself and how it may happen. So are you saying that it is not an issue anymore? \$\endgroup\$osmak– osmak2019-06-21 07:05:03 +00:00Commented Jun 21, 2019 at 7:05
-
\$\begingroup\$ I was saying glitches inverting polarity only occurs in NRZ-I while -L is simply binary code and that -I is obsolete \$\endgroup\$Tony Stewart EE since 1975– Tony Stewart EE since 19752019-06-21 07:06:35 +00:00Commented Jun 21, 2019 at 7:06
-
\$\begingroup\$ I wouldn't say NRZ-I is obsolete, at all. Isn't it in use, for example, by USB? \$\endgroup\$Marcus Müller– Marcus Müller2019-06-21 07:08:56 +00:00Commented Jun 21, 2019 at 7:08
-
\$\begingroup\$ Regardless if they are obsolete or not. My understanding is that NRZ-I almost solves the problem because only one bit will be erroneous unlike NRZ-L. \$\endgroup\$osmak– osmak2019-06-21 07:10:37 +00:00Commented Jun 21, 2019 at 7:10
|
Show 4 more comments
1 Answer
\$\begingroup\$
\$\endgroup\$
1
how can such a sudden change in polarity occur in practice?
In a wireless/bandpass channel, that'd be a 180° phase shift.
I've never seen that occur anywhere on a wired (baseband signaling) link that you'd use NRZ-whatever on, and aside from "blatant bugs in the hardware design", I wouldn't see how this would occur in practice.
and is it really a problem?
Not that I can think of, no.
Even if we came up with a model where we get polarity reversals, they'd be "worst case" errors (our whole line code being designed the fact that the two levels are as strongly not the other as possible, you'd need the maximum physical error to achieve that) – and hence rare (otherwise our line code choice would have been bad, and we should have used a physical layer that is robust against these).
Rare errors on links are typically countered either by checksums and re-requesting broken frames, or by forward error correction. Of the latter, you can definitely devise a scheme that converts a continuous bit flip starting at specific position to a single bit error, which might be well-recoverable with the given code. However, this would have little to do with the line coding.
-
\$\begingroup\$ Comments are not for extended discussion; this conversation has been moved to chat. Any conclusions reached should be edited back into the question and/or any answer(s). \$\endgroup\$Dave Tweed– Dave Tweed2019-06-21 23:23:58 +00:00Commented Jun 21, 2019 at 23:23