[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
ULE, and CRC-32
After some-off list discussions with Alain, I think we can now propose
some modified text for the section in ULE on CRC usage. Basically, this
says follow the well-known practice, and in particular this converges on
what ISO DSM-CC specifies.
New text reads:
SNDU Trailer CRC
Each SNDU MUST carry a 32-bit CRC field in the last four bytes of the
SNDU. This position eases CRC computation by hardware. The CRC-32
polynomial is to be used. This is a 32 bit value calculated according to
the generator polynomial represented 0x04C11DB7 in hexadecimal:
x^32+x^26+x^23+x^22+x^16+x^12+x^11+x^10+x^8+x^7+x^5+x^4+x^2+x^1+x^0.
Examples where this polynomial is also employed include Ethernet, DSM-CC
section syntax [ISO-DSMCC} and AAL5 [ITU356]. The use resembles, but is
different to that in SCTP [RFC3309].
The Encapsulator initialises the CRC-32 accumulator register to the
value 0xFFFF FFFF. It then accumulates a transmit value for the CRC32
including all bytes from the start of the SNDU header to the end of the
SNDU (excluding the 32-bit trailer), and places this in the CRC Field.
The receiver performs an integrity check by independently calculating
the same CRC value and comparing this with the transmitted value in the
SNDU trailer. SNDUs that do not have a valid CRC, are discarded, causing
the Receiver to enter the Idle State.
This description may be suited for hardware implementation, but this
document does not imply any specific implementation. Software-based
table-lookup or hardware-assisted software-based implementations are
also possible.
The primary purpose of this CRC is to protect the SNDU payload from
undetected resassembly errors and errors introduced by unexpected
software / hardware operation while the SNDU is in transit across the
MPEG-2 subnetwork and during processing at the encapsulation gateway
and/or the receiver. It may also detect the presence of uncorrected
errors from the physical link (these may however, in some cases, also be
detected by other means).
Alain:
> Some questions about the CRC-32, because I'm a little bit confused.
> I think many of you are familiars with that CRC-32 stuff, so I
> need their help ;-)
<SNIP>
- Next by Date:
test
- Next by thread:
test
- Index(es):