💾 Archived View for rawtext.club › ~sloum › geminilist › 007655.gmi captured on 2023-09-28 at 16:15:24. Gemini links have been rewritten to link to archived content

-=-=-=-=-=-=-

[Off Topic] WiFi HaLow, Mesh, and Low Powered Devices

John Goerzen jgoerzen at complete.org

Thu Dec 2 20:24:21 GMT 2021

- - - - - - - - - - - - - - - - - - -

On Thu, Dec 02 2021, Andrew Singleton wrote:

Look. I live Simi-Rural. Meaning not so far out that you have hundreds

of acres for cattle, or farms or whatever. Just an eight acre piece of

land that's more of a family hobby project than anything. And yet that

is still a fairly sizeable chunk of land so far as traditional wifi is

concerned.

Hi Andrew,

I've been interested in these protocols for a long time. I've writtensome stuff that runs atop LoRA and XBee SX, for instance, including PPPor tun/tap IP/ethernet stacks.

HaLow looks like an interesting middle ground; not as long of a range(1km vs 10km) but more easily integrated.

I hear you on the LoRA bandwidth. XBee SX is better, and I've seeneffective rates of 50Kbps out of the thing. My xbnet program athttps://github.com/jgoerzen/xbnet covers running TCP/IP using tun/tap aswell as PPP over it athttps://github.com/jgoerzen/xbnet/blob/master/doc/xbnet.1.md and I alsocover running UUCP over the thing. The XBee SX also has apoint-to-point "modem emulation" mode in which it looks behaves like aregular modem to the OS (no translation code necessary). That removessome of the possibilities of direct node-to-node communication, ofcourse, but could, eg, host a number of PPP connections forconnectivity.

The XBee devices are also interesting for a couple of other reasons.One, is they support a self-assembling mesh in firmware, so you get meshrouting without having to do anything on the software side. Very handy,though of course it reduces available bandwidth. Secondly, they supportTX power up to 1W which is what allows them to achieve near-LoRAdistance while having a much faster bitrate. However, they are moreexpensive than LoRA devices, and obviously in the case of the 1W ones,more power-hungry.

Since we're talking about old protocols, there is, of course, UUCP.There is a more modern replacement called NNCP that has quite a few niceproperties. It is still asynchronous, but it is fully encrypted andauthenticated, supports tor-style onion routing and async multicast, andso forth. I have made some blog posts about it and also offer Usenetfeeds over NNCP. UUCP or NNCP would be perfect fits for something likethis.

Another interesting program is Syncthing. Think of it like a serverlessDropbox. It is fully distributed, and every peer in a network canpropagate changes.

Both Syncthing and NNCP support a "traveler" device -- say a phone thatgoes to civilization periodically and syncs up things that propagateto/from a remote location.

And, there is NNTP, the protocol behind Usenet. Usenet can beexchanged via NNTP, but also via asynchronous means (NNCP, UUCP, etc).It can be done in a very low-bandwidth way and could propagate posts,(small) files, etc.

Finally, don't forget email. It can be routed in all sorts of ways, too.

I have been writing and coding around some of this stuff. See:

https://changelog.complete.org/archives/tag/nncphttps://github.com/jgoerzen/nncp-toolshttps://github.com/jgoerzen/gitsync-nncp/blob/main/gitsync-nncp.org

I also have a 5-minute lightning talk video I should get uploaded one ofthese days...

- John