Hacker News Reader

Comment by CraigJPerry | Hacker News Reader

It’s hard to bet against human determination but on the face of it, it seems out of reach to put satellite voice calls in a package the size of an iPhone.

The power requirements and thus the bulky batteries of existing satellite phones could be reduced by switching to protocols which trade off bandwidth for higher error correction abilities through encoding and I don’t think existing satellite phones use the most bandwidth efficient encoding available so that could be another improvement.

There’s still the free space path loss between the surface of earth and a Globalstar or Starlink which my best guess estimate is in the order of 130db (can anyone more knowledgeable share a better estimate? I assumed bottom end of L-band, a height between Starlink and Globalstar, and used unverified dBi values for antennas at both ends).

Then there’s latency but I don’t think that’s a show stopper for many common use cases of voice and data.

View on HN

> The power requirements and thus the bulky batteries of existing satellite phones could be reduced by switching to protocols which trade off bandwidth for higher error correction abilities

No, you can’t trade bandwidth against error correction. What you can do is use channel coding/error correction to trade effective signal-to-noise strength against data rate (at constant power).

But since your data rate demand is fixed for voice, and the uplink transmit power is fixed too (by 4G/5G specs), you need to raise the received signal strength. The only thing that can help you there are more directional receive antennas on the satellites, which is what these new generations are doing.

To be clear, what's happening here is people are accustomed to using "bandwidth" to mean bits/second but bandwidth has a much more specific meaning in the RF world, akin to calling Ethernet "SSH".

Bandwidth is the actual number of kilo/mega/gigahertz the signal occupies on the spectrum and is largely a function of symbol rate, but the actual data rate has many more variables.

I’m aware of the distinction and I’m using the terms accordingly.

More bandwidth does not help you get a better data rate if you’re power limited. In the case of satellite communications you basically always are, and bidirectionally so when it’s to mobile/battery-powered devices.

Yeah, I posted to clarify for the person you replied to and other non-rf people. You obviously know what's going on but you used a lot of language I figured should be defined for anyone who comes across this.

Ah, sorry, I mixed you up with GP then and thought you were disagreeing with the "power-limited regime" statement!

That is in fact incorrect. Capacity scales as Blog2(1+SNR) so increasing bandwidth at the same power (which would decrease SNR linearly assuming AWGN) will result in increased capacity.

Although pratically you can do that only so far, because you're going to be limited by the bandwidth of your receiver.

It’s only theoretically incorrect.

Practically, satellite communications are power-limited and accordingly will always be so close to the noise floor (i.e. S/N << 1) that the logarithm can be approximated with a constant and the bandwidth dependency is negligible and drops out of the equation.

It only matters in the bandwidth-limited regime, i.e. for S/N >> 1.

This thread is why I like HN. Thanks for sharing the insights both.

Please take all of this with a grain of salt! It's quite possible that the limiting factor here is already spectral efficiency and not power; I don't know the link budgets involved and am mainly basing this on the very low observed data rates of systems I'm familiar with.

I don't have the exact S/N numbers for these systems, and there might well be some potential gain that's not being exploited for reasons of spectral efficiency (after all, these systems want to serve multiple phones at least somewhat efficiently).

The spectrum is limited in any case, both by receiver capabilities and the fact that it's a very scarce resource in useful frequency ranges.

You can talk to satellites over lora with a 5w transceiver (source: done it), which is on the order of magnitude of a cell phone transceiver. Opus goes down to .75kbps. Ham boomerisms don't cleanly apply to modern voice codecs and digital modulation techniques.

How do you do LoRa over a sat? I thought the chirping makes that difficult because you exceed the passband on the sat? Interesting! Which sat did you use?

I used tinygs on a lilygo esp32-based board with a semtech sx1262. Before its untimely demise I used SDSAT, which would rebroadcast messages. I'm not currently aware of any active sats that do that, but you can receive telemetry from tons of them as listed on tinygs.

Thanks! I'll look into it! I live in a busy city now near the ground floor so the hobby is kinda on hold as I can't get any antennas up :( But this might be portable enough to bring out on a hike.

I would be quite happy with texting or very slow internet tbh

Reply on news.ycombinator.com