[aprssig] Paths - Solution proposal

[Kris Kirby]

On Sun, 13 Jul 2014, Tom Hayward wrote:
> ARP, IP (DHCP and BGP make this automatic), DNS, application-layer 
> authentication and databases--all of these layers work together to 
> route your message for you. This is possible because over the last 30 
> years technology has advanced and commercial high speed data networks 
> have been built. Hams have simply been left in the dust, still using 
> technology from the 1980s. If you want to play with modern 
> applications on ham radio, you're going to have to go down a few 
> layers and improve the infrastructure first.

OSLR makes it a little different as well. And we've got the entire 
44.0.0.0/8 space to play with, which was seemingly ignored by HSMM.
 
> I think the point you're getting at is that AX.25 isn't an optimal 
> protocol choice for APRS. There are some other great tactical 
> information protocols out there--I am impressed by many of the 
> features of AIS.

To be fair, we have limitations that the FCC didn't burden Part 90 with, 
or at least ignored enforcement on, such has "FM Modulation index shall 
not exceed 1.0", which was ignored by the FLEX protocol and Motorola's 
trunking protocols as long as the occupied bandwidth matched the 
licensed bandwidth. FSK (GMSK) and AFSK got us this far, it's far 
overdue that we start moving to more bit-dense modulation like 4FSK. 
Telephone modems hit 56K, but with 16- and 24-bit sound cards, we should 
be able to approach channel capacity that over a full-quieting FM link 
(S/N is often 20dB or better). We are generally limited to the amount of 
bandwidth available for a "communications quality" signal, i.e.: similar 
to bandstopped AM for telephonic speech. 

It would be nice to see a nationwide "high-speed" infrastructure on a 
band besides 2m. As I understand it, many amateurs near the Pave Paws 
installations don't have 70cm as an option at all. 70cm would be 
preferable since the required antenna aperture is smaller, and the 
general coverage mostly equal to 2m. 900 MHz tends to be LOS and highly 
cellular, and 222MHz still has "large" antennas. 

For that matter, 2400 or 3600 bps should have been the norm, not 1200 
bps. Both will fit through a radio configured for voice. Flex is 1600 
bps with a deviation of 1600 Hz or 4800 Hz (thus, "four levels" can be 
discerned post-discriminator: a 1/0 from the data slicer and a 
loud/quiet from a comparator -- 4FSK). 

One individual told me he was able to connect using a modem over a 
full-duplex phone patch at 14.4Kbit/s. 

And yes, collisions need to be dealt with somehow or someway. AX25 
results in either a copied packet or a trashed packet. IP is far too 
chatty for simple communications, particularly when tunneled through 
AX25 (it should be UI frames, like APRS), and is limited by the 
subnetting practices of the local network. D-Star does not cope with 
hearing packets addressed to another repeater.

Then there's the hidden node problem... Timeslotting is nice but makes 
things technically complex, and the least common denominator here is a 
TNC-type device attached to a random radio. Surely one can come up with 
a protocol and convince a Chinese manufacturer to implement it. 

It's possible to do what I call "sortacasting", where two sets of 
repeater frequencies are used across a state, alternating the use of a 
frequency to curvature of the earth and repeater coverage. That doesn't 
work in all situations; even the WiFi guys figured out that with three 
non-overlapping channels you could occasionally fit a fourth or fifth 
channel into the mix. We've been lucky that 144.39 is far enough away 
from some repeater inputs, so sites can be collocated, but not all 
repeater inputs are, i.e.: 144.63 MHz. (2m should have never had a 
600KHz split, but I digress.) In less populous states, "sortacasting" 
may work well -- it largely depends on the sites chosen and the terrain 
covered. It's basically two full-duplex SFNs alternating. 

I've done a few cross-country drives, and the least common denominator 
of 1200 bps on 144.39 MHz has paid off in my tracks. I run 25W mobile 
into a 1/4-wave whip, and on the cross-country trips, it was on top of 
the vehicle. My interest is more in seeing where we have coverage and 
where we don't; 25W with 0dBd antenna will get out of valleys on 
occasion where 5W or 10W with a 3dBd antenna won't. 

--
Kris Kirby, KE4AHR
Disinformation Architect