meshnetworks

Bill Kearney wkearney99@hotmail.com
Thu, 7 Feb 2002 18:22:16 -0500


> I think this assumption, that long-haul traffic usually has
> to reach fiber, could break down.
>
> If we reach a point where there are several paths between
> Texas and San Francisco with unbroken cell-coverage, why
> not an unbroken ribbon of wireless-traffic relays?

This ignores (or fails to mention) an important issue; latency.

Each of those hops is going to introduce latency.  While it would be great
to have ubiquitous wireless I wouldn't want to use it exclusively.  The
nodes could/should get the data back into terrestrial lines for more
efficient delivery to their destinations.  Look at a satellites, the latency
to get up and back from a bird is /almost/ too slow for decent quality voice
traffic.  Thus terrestrial and trans-oceanic fiber still exists (and
thrives).

Things like streaming and telnet or other per-character sessions all suffer
when latency mounts up.

Your cell phone call during your travel from TX to SF does not use the
wireless mesh to carry the conversation.  At the earliest point it exits the
wireless mesh and uses regular SS7 network routes, just like landline calls.

Another thing about wireless, it's not just about speed, it's about
capacity.  Given that wireless is effectively a shared network like ethernet
hubs it's unlikely to reach anywhere near 100% of possible bandwidth.
Regardless of how fast you can get TO the tower you're still at the mercy of
how many others are trying to do exactly the same thing.

Hey, don't get me wrong, I think wireless is great.  But it's not without
some serious issues that have much harder solutions than those needed for
wired access.

-Bill Kearney