5 ms·

Why IP over HF Radio Should Be Avoided (2014)

2 points by whereistimbo 6y ago

h2odragon 6y ago
> HF Radio is very slow with typical rates around 1200 bits per second. This slow speed is usually perceived as the primary difficulty with using HF. That's certainly enough to pretty much hose a TCP/IP for practical use because of protocol overhead and the latency the link speed enforces. I did some unwise experiments with really long serial links that topped out in that neighborhood; several layers less complicated than anything real world useable. However, we still have the whole layer of protocols that evolved for quite similar conditions; which include serial terminal / command line / file transfer protocols from the modem days. While they're not suited for broadcast, the lessons and designs knowledge they embody largely apply.
- whereistimbo 6y ago
  Thats interesting. Do you think there is a potential of a better protocol suited for HF compared to TCP/IP or even just IP alone. I would also want to hear more about your experiment like what hardware were you using at that time.
  - h2odragon 6y ago
    "Is there a better protocol" ... absolutely, but design for the use. I had a ~500m link across a campus (heavily wooded) for a while. The main link was often coax cable or fiber, but would get cut several times a year, so there was always need for a backup. I experimented with several rs232 serial port type methods in that time, from directly wired over 500m on cat3 cable (12 pairs per signal worked well), to modems and my own "POTS" telephone line emulation, with a couple of efforts at optical and radio links. Hardware was all scrounged trash and lashed together zombie stuff. Somewhere in the piles I still have one of the 486 motherboards that hosted an end of one of those links during a lightning storm. I soldered patch wires onto the ISA slots after the power traces were vaporized by a close strike and that system lived for some time after that.
jlgaddis 6y ago
Meh, this "whitepaper" is just thinly-veiled spam. I'll sum it up for you: > "IP over HF is really inefficient and should be avoided whenever possible." > "You wanna know what's SUPER-DUPER EFFICIENT over HF, though?" > "STANAG 5066!" > "You know what's really crazy, though? It just so happens that WE SELL STANAG 5066 PRODUCTS! To governments and militaries! Would you believe that? I mean, what's the chances, right!?" (Also, FWIW, the "whitepaper" is at least six years old.)

2 points by whereistimbo 6y ago

> HF Radio is very slow with typical rates around 1200 bits per second. This slow speed is usually perceived as the primary difficulty with using HF. That's certainly enough to pretty much hose a TCP/IP for practical use because of protocol overhead and the latency the link speed enforces. I did some unwise experiments with really long serial links that topped out in that neighborhood; several layers less complicated than anything real world useable. However, we still have the whole layer of protocols that evolved for quite similar conditions; which include serial terminal / command line / file transfer protocols from the modem days. While they're not suited for broadcast, the lessons and designs knowledge they embody largely apply.

whereistimbo 6y ago
Thats interesting. Do you think there is a potential of a better protocol suited for HF compared to TCP/IP or even just IP alone. I would also want to hear more about your experiment like what hardware were you using at that time.
- h2odragon 6y ago
  "Is there a better protocol" ... absolutely, but design for the use. I had a ~500m link across a campus (heavily wooded) for a while. The main link was often coax cable or fiber, but would get cut several times a year, so there was always need for a backup. I experimented with several rs232 serial port type methods in that time, from directly wired over 500m on cat3 cable (12 pairs per signal worked well), to modems and my own "POTS" telephone line emulation, with a couple of efforts at optical and radio links. Hardware was all scrounged trash and lashed together zombie stuff. Somewhere in the piles I still have one of the 486 motherboards that hosted an end of one of those links during a lightning storm. I soldered patch wires onto the ISA slots after the power traces were vaporized by a close strike and that system lived for some time after that.

jlgaddis 6y ago

Meh, this "whitepaper" is just thinly-veiled spam. I'll sum it up for you: > "IP over HF is really inefficient and should be avoided whenever possible." > "You wanna know what's SUPER-DUPER EFFICIENT over HF, though?" > "STANAG 5066!" > "You know what's really crazy, though? It just so happens that WE SELL STANAG 5066 PRODUCTS! To governments and militaries! Would you believe that? I mean, what's the chances, right!?" (Also, FWIW, the "whitepaper" is at least six years old.)