On Mon, 9 Jul 2001, Ken Clements wrote:
> Be very careful about using geographic data for wireless routing.
> Plenty of folks have done this in the past and been sorry. You can
That doesn't surprise me in the least. It's too early yet. If you're too
early, you get burned.
> get pointers by asking Google for "wireless geographic routing."
> Also have a look at my post:
> Geographic routing works well in places where the physical space maps
> well into the communication link space (Kansas) and is problematic
Of course, it's the prerequisite. That's why I mentioned cellular
wireless, which is orderly as if taken from Christaller.
> when it does not (Manhattan). The shortest physical distance may not
> be the shortest latency. Wireless links are stochastic. Just because
Sure, we don't live in a perfect world. But the point is not to choose the
best possible link (it will come later, when the connectivity map will
resemble the physical map, and routers will measure relativistic signal
TOF latency), the point is to have a progress metric, so you can chuck out
monster routing tables, and overwhelming admin traffic.
> you got a packet through a link quickly this time, does not mean you
> will be able to do so next time. Often the best route in
Yes, ATM did it better. As I said, IP is b0rken.
> communication space looks very strange when you see it represented on
> a map.
If you use cut-through, you can measure the line length inbetween.
-- Eugen* Leitl <a href="http://www.lrz.de/~ui22204/">leitl</a>
ICBMTO : N48 10'07'' E011 33'53'' http://www.lrz.de/~ui22204
57F9CFD3: ED90 0433 EB74 E4A9 537F CFF5 86E7 629B 57F9 CFD3
This archive was generated by hypermail 2b30 : Fri Oct 12 2001 - 14:39:43 MDT