Open 802.11b wireless access points and remailers

[dmolnar]

On Wed, 25 Jul 2001, David Honig wrote:

> When "cell phones" get more programmable, and handle text, an interesting
> "app" could be guerilla-net-like "routing".  If everyone's "phone" is
> a RF repeater/router, its not impossible.

You could probably hack this up now, if you were willing to lose the
cell phone functionality of your cell phone. Maybe you could even get by
with just replacing the web browser on your cell phone. (I still can't
make head or tails out of my phone's browser, but apparently people use
them.)

cell phones do handle text. In fact people are trying to make a
business out of cell phone mailing lists. see www.upoc.com
You'd have to add would be some kind of scripting language for
forwarding text messages on the phone.

 >
> Battery life would probably be the worst impact.  A few airline bottles
> of vodka will keep the fuel cells humming (for the future phone, I mean).

Heh. "One for me, one for my phone."

Batch transmissions every hour on the hour might help with this. No reason
to be up all the time for sending e-mail. You could also play games in
which every phone picks a different minute each hour, then wakes up during
that minute for transmission. Your chance of being in the same minute as
your destination isn't great, but you could transmit the packet to each of
your neighbors in that minute, each of whom tries to relay the packet in
different minutes during the next hour.

One issue with that, though, is how to stop packets from flying around
long after they've been first delivered. A gnutella/freenet-style limit on
the number of tries might help. So might announcements of packets
received; i.e. a phone says "I've received packet X, so you can drop it."
You'd have to be careful about an adversary trying to create packets which
live forever (i.e. the hop limit should not live in the packet, unless the
packet is signed and these announcements had better have some way of
proving they come from the 'intented' sender) (but at the same time, we
should avoid any protocol which requires a PKI for phones or even
public-key crypto on efficiency and speed grounds; it takes 20 seconds
for my phone to negotiate one RSA key exchange).

In order to prevent what Anderson calls "sleep deprivation" attacks, you'd
also want that the number of minutes the phone is up depends weakly or not
at all on parameters under the control of an adversary. like how many
packets received during the previous minute up. Random dropping of
incoming packets might be a way to get around this, since I'm thinking
that every phone broadcasts to every other phone in the same minute
anyway.

(I keep thinking of _Dayworld_ through all of this, but I don't yet see a
good joke or a useful metaphor -- phones are not assigned set minutes
for life, unlike in _Dayworld_, so what's a "dayworld breaker"? a phone
that continues to relay during the entire hour? that would be a good
thing, since it'd speed up packet relay.)

I'd be pretty surprised if people haven't already looked at these sorts of
schemes and come up with much better ones. Although maybe they haven't
been considered with adversaries in mind. Anyone know of references?

 >
> Lots of mil apps for fully distributed RF nets, too.

That's where spread-spectrum came from, isn't it? How much is publically
known about the toys they already have?

-David