[ih] Internet in the Air: Was Re: Internet at Sea

[Karl Auerbach]

Thinking of Internet at Sea, there is also "Internet in the Air" (there 
is also "Internet in automobiles", which has some similar issues.)

Several years back we did some work with the FAA and Boeing who were 
trying to figure out how to improve air traffic control over the 
mid-Pacific.  At that time there was not solid voice connectivity to 
aircraft way out in the middle of the Pacific. (There were some lower 
frequency radios that could do the job, much of the time, but they were 
not particularly favored.)

We put modified Cisco routers and other gear onto some commercial 
trans-Pacific aircraft and played.  Because pilots are used to 
push-to-talk systems and long response times, we could cache voice 
spurts and interleave those with other traffic.  That gave us a lot of 
flexibility about adding things like redundancy in case of RF noise.

We began our experiments with geo-synch satellites.  We intended to move 
to low earth orbit satellites, and then aircraft-to-aircraft relays 
(with each airplane acting as an ever-moving IP router) but we ran out 
of funding.  (It can be expensive working with trans-oceanic capable 
aircraft.)

For pilots the geo-synch links worked.  (I wanted to experiment with 
tokenized voice like what had been done earlier at SDC for 
communications with certain kinds of manned undersea vehicles. ATC 
communications are highly stylized with a small core vocabulary.  This 
would have allowed common words to be converted to nice short tokens.  
The voice of a given speaker would not be reproduced accurately, but the 
words would be synthetically generated at the receiving end and 
generally were rather more clear to the listener than typical ATC voice.)

The geo-synch path worked wasn't so great for passengers.  As usual a 
lot of onboard caching helped.

By-the-way, one of the lessons I took from the DARPA Robotics Challenge 
(I worked on that for several years) is that we networking people can 
learn a lot from the undersea sound/communications people at places at 
MBARI and Woods Hole.  I was amazed at how they were able to pull a 
usable signal from a very noisy channel even without forward error 
correction.

(On the geosync system we were using access was moderated via a ground 
station in Texas.  One got to that moderator using Aloha style access.  
The moderator came back with a time slot (usually a few hundred 
milliseconds beginning at a specified time.)  So, apart from the need 
for well synchronized clocks on the aircraft the typical access time to 
the main channel could be several seconds.  Again, that was OK for the 
pilots, but not for passengers.)

There are, of course, issues that are too often overlooked when using a 
single bent-pipe link via a geosync satellite, such as solar blanking 
(when either the satellite transits the face of the sun from the point 
of the view of the sending or receiving ground station or when the 
satellite's view of a ground station is blinded because or a reflection 
of the sun off of the earth.  At that time tracking low earth satellites 
from a moving platform was not well developed - At Sun we had designed 
some highly portable antenna capabilities to track low earth satellites 
from Steve Robert's bicycle, but for that project we were aiming only at 
about 32kbits/second, which is OK, but marginal, for non-tokenized voice.

I of course suggested a technology we created on the Interop Show net 
back in 1998: "Gaganet", trans-relativistic networking:

https://www.cavebear.com/cb_catalog/techno/gaganet/

(Some people actually believe that this was real, and not a joke.)

         --karl--