I received this reply on the Live555-mailing list:
“FEC is not needed for UDP. The UDP service delivers a packet, or does not.
UDP does not randomly flip some bits inside the packet.
To tolerate packet loss, only redundancy is needed.”
I would like to know if this statement is correct; and please also provide the sources on which your answer are based.
Clarification: I was referring to Forward Erasure Correction. Maybe the person answering in the Live555-list was referring to Forward Error Code.
UDP in itself does nothing to cause or prevent bits being flipped. Depending on how the data is transmitted, bits being flipped is certainly possible.
At the same time, most networking has moved toward error correction taking place at a lower level in the stack, so in a lot of cases, you can probably get by without adding any as a layer on top of UDP. That, however, is far from guaranteed — depending on the physical layer over which you transmit it, bit errors in UDP packets certainly remain a possibility, so if you really need to ensure against that, independent of the lower layers in use, then FEC could provide real utility.
Edit: and yes, I’d guess that absent some context to indicate otherwise, most people using “FEC” mean “forward error correction” (that’s certainly what I thought you were referring to when I wrote the reply).