[e2e] Re: [Tsvwg] Really End-to-end or CRC vs everything else?

[Craig Partridge]

In message <4.3.2.7.2.20010608171151.0e9bfcf0 at mira-sjc5-9.cisco.com>, "Michael 
A. Ramalho" writes:

>I know of no other easily proven statement for
>bit "stuckedness" for Alder (or even other statistical
>or frequency based checking/detection algorithms).

The TCP checksum catches all error patterns that are 15-bits longer or
less and all but one 16-bit long error.  If you did the TCP checksum as
a 32-bit checksum (32-bit adds instead of 16) you'd have the same
property.  If single bit errors are the error pattern, CRC is overkill.

Key point here is we need an error model that is realistic to determine
what checksum to use.

[Side point -- Jonathan apologies if this steals your thunder -- but
Jonathan has a nice little proof that, if all error patterns are equally
likely, that a checksum that just includes a fixed size constant to each
packet is just as effective as any stronger check -- the moral here is, if
you don't know the error pattern, you don't know which checksum to choose].

Craig