Re: NEW TOPIC! : What about an AX.25 Bridge?

[Phil Karn <karn@unix.ka9q.ampr.org>]

I think the most practical way to implement the hardware is to adopt
Part 15 spread spectrum equipment. The ARRL is currently working on a
proposal to the FCC to liberalize the amateur spread spectrum rules,
and it's specifically written with this in mind.

In fact, if you're willing to stay under 1W -- plenty of power for an
intelligently designed radio in such a network -- then you can just
stick to Part 15. There are enormous political benefits, such as a
complete lack of content and use restrictions. Combine Part 15 for
local use with the Internet for the long haul stuff and traditional
ham radio starts to look awfully irrelevant. Several companies are
already doing just this, including Metricom's Ricochet service and
Dewayne Hendricks' Tetherless Access Limited.

Min-energy routing is not all that new. I first heard of it many years
ago when Dave Mills gave a talk on DARPA's Multiple Satellite System
research project. The problem is to find the path through a random
collection of low-earth-orbit satellites such that the total
transmitter energy from all of the satellites required to forward a
packet to its destination is minimized. This is not the same as
minimizing the number of hops or minimizing the total distance
traveled, because RF propagation is inverse-square. A series of many
closely-spaced hops can be much cheaper energy-wise than a few long
hops even if the total distance traveled is greater.

This is done partly to minimize DC power requirements on the
satellites, but more importantly it maximizes system
throughput. That's because different satellites can be transmitting at
the same time on the same channel if they are only talking quietly to
their nearby neighbors.

Same thing on the ground. If you implement automatic transmitter power
control, and if you pick min-energy routes, then you increase the
chances that the spectrum can be in simultaneous use in different
parts of the network. At the 1991 ARRL digital conference I proposed a
simple modification of min-energy routing for terrestrial use.
Instead of minimizing total transmitter energy, your routing metric is
the number of stations you'd capture in order to talk to a particular
neighbor. Your nearest neighbor requires the least power, and if he's
really close then you could probably talk to him so quietly that no
one else would hear you. That gives him a metric of 1. Somebody
farther away (but still directly reachable) might require so much
transmitter power that you'd have to capture a dozen other nearer
stations in the process. So his metric would be 13. And so on. Just
run your routing algorithm with these metrics, and you have a
self-organizing network that automatically maximizes its total
throughput.

Phil