Unless Im mistaken (entirely possible), an IP enabled phone has 2 distinct
and separate "stacks", the IP stack and the "phone" stack.
As I said, in a NAT'd scenario the IP stack will never see an unsolicited
request and hence not respond to it.
The phone side of course will ring when called. Duh.
GPRS <> VoIP (yet)
Jm
That's the *point*.
You hand the phone a trojan/virus/whatever when it's making an OUTBOUND
connection on the NAT side (for instance, if the PDA side is checking
mail, feed it a trojaned piece of mail). You then have the trojan drop
you a note "Oh, and my phone number is XXX-YYYY".
Then, when it's time to attack somebody, you send the phone a page that
tells the trojan "Hey XXX-YYYY, wake up and pound on victim address <whatever>".
With proper encoding of the page, the phone's owner may even just say
"Damn, more <bleeping> Korean spam in characters I can't read", and not notice
that 45 seconds later, the phone starts chirping away by itself....
The point is that you can contact the phone via *non-NAT* means and have it
launch an attack - the fact you can't wake it up via NAT can be worked around.