New LoRa RF distance record | Hacker News Reader

New LoRa RF distance record

thethingsnetwork.org

181 pointswienke135 comments

LoRa runs in the 868 MHz band (in Europe) - a frequency range known for being line of sight only as the ionosphere doesn't reflect radio waves of that high frequency.

However sometimes you can get Troposheric propagation on these high frequency bands. This used to be common when I was a teenager living in the south of England in Kent and we could occasionally receive racy TV channels from Europe.

https://en.wikipedia.org/wiki/Tropospheric_propagation

From wikipedia:

> Tropospheric propagated signals travel in the part of the atmosphere adjacent to the surface and extending to some 25,000 feet (8 km). Such signals are thus directly affected by weather conditions extending over some hundreds of miles. During very settled, warm anticyclonic weather (i.e., high pressure), usually weak signals from distant transmitters improve in strength.

Radio Amateurs love this kind of propagation. It is very variable, doesn't last long, but can give you contacts over great distances on frequencies which are usually line of sight. I think the record for radio amateurs on the 70 cm band (430-440 MHz) is over 4,500 km!

Could the signal bounce off airplanes as well, or is that effect not relevant in practice? (I think I've heard such stories before from radio amateurs...)

It can bounce off of meteor trails or the moon. It's been done many times in the ham community. Signals do bounce off of aircraft too, there are even passive military radars that track airborne objects using DVB-T or other strong signal reflections. It could be used if the conditions are just right.

All of this has really nothing to do with LoRa and everything to do with physical properties of the emissions (frequency, power, polarization) and physical properties of whatever happens to exist in the world at the time.

Yes, of course this is related to the physical properties of the emissions, but LoRa uses specific frequencies of the ISM band (868 MHz in Europe), and this discussion was about the reasons why that specific LoRa transmission worked even though there wasn't a line of sight. Airplanes could be an explanation, although your sibling comment thinks there isn't sufficient transmission power for it to be feasible.

Ah. It's most likely tropospheric ducting, I thought the top comment explained it.

The path loss for moonbounce is approximately 390 dB. You're not doing that by accident. In addition to using weak signal modes, you'll need directional antennas and lots of power to overcome that.

It's not that high. You have to use the radar equation with the moon being a very lossy but huge target cross section. The path loss at 432 MHz and average moon distance of 384,400 km is 261.6 dB.

It's still not going to happen accidentally.

Would it be closer to the other value if you factored in the atmosphere, and whatever else is in the way?

As an aside, would radio via the moon ever be practical? Without pumping out so much transmit power that you pick up the signal anyway, without pointing a receiver at the moon. I think it would be neat, if only for the sake of novelty.

No, there's very little loss from the atmosphere at 432 MHz. The 390 dB figure is a misunderstanding of how the path loss is calculated. The poster "wl" took the one way path loss (195 dB) and doubled it. But the moon is a huge reflector and provides "gain".

Here's a C program with the correct equation.

  #include <stdio.h>
  #include <stdlib.h>
  #include <math.h>

  int main(int argc, char **argv)
  {
    double  d, f, lambda, loss;

    if (argc != 3) {
      fprintf(stderr, "usage: moon <km> <frequency(MHz)>\n");
      exit(-1);
    }

    d = atof(argv[1]) * 1000.0;
    f = atof(argv[2]);

    lambda = 299792458.0 / (f * 1000000.0);
    loss = (0.065 * (1.738e6 * 1.738e6) * (lambda * lambda)) / (631.65468167 * (d * d * d * d));
    printf("EME path loss = %f dB\n", 10 * log10(loss));
    return 0;
  }

1.738e6 is the radius of the moon in meters and 0.065 is the reflection efficiency.

Satellites make moonbounce impracticable.

For doing meteor- and airplane-scatter stuff you need quite a bit of transmission power. The max. tx power of LoRa transmitters is way too low.

Sure, that's how you get doppler shifted frequencies and enter the realm of radars.

> LoRa runs in the 868 MHz band (in Europe)

No, LoRa can run on two bands in Europe, either EU433 from 433.05 to 434.79 MHz or EU863 from 863 to 870 MHz.

See e.g., https://www.everythingrf.com/community/lora-frequency-in-eur... or https://www.thethingsnetwork.org/forum/t/is-there-an-agreed-... or for a more dense, but definitive source https://docdb.cept.org/download/4316 ("EUROPEAN TABLE OF FREQUENCY ALLOCATIONS AND APPLICATIONS IN THE FREQUENCY RANGE 8.3 kHz to 3000 GHz (ECA TABLE")

Albeit I'm not sure what band they used here.

There is essentially no use of LoRaWAN at 433MHz in Europe, this would have been at 868

Just to expand upon this, tropospheric propagation happens typically where there is a temperature inversion, meaning, air near the surface is cooler than higher up in altitude. That warmer air aloft acts as a layer that radio waves bounce off of.

CTO & co-founder of The Things Network here! I've been working with LoRa for over eight years now, and I keep getting surprised by measurements like this. What started off as "line of sight works" went to a successful "LoRa moon bounce" [0] and now a sea level reception over 830 miles.

So, any physics people here care to comment how this works? Is this pure atmospheric refraction or is there something else?

[0] https://www.camras.nl/en/blog/2021/first-lora-message-bounce...

I've just disconnected a BobCat miner so I have a spare 1.1m 8.5dBi 868MHz antenna on my roof. Can you recommend a good gateway to get started on TTN?

Recently somebody created this guide to turn your Helium gateway (running on a RPi) to a TTN gateway: https://medium.com/@friedkiwi/repurposing-helium-iot-hotspot...

A CTO that doesn't understand RF and has been working with it for 8 years? <surprised Pikachu face>

To some extent it's statistics. Send enough messages and the odds are one will eventually get though.

> providing a standardized and objective measure of the technology's capabilities

Not really. The environment is still variable. A short message could have bounced off some random short duration reflector somewhere (aeroplane, meteor, lightning, ...) or have been refracted by some short term effect. Standardised and objective is an anechoic chamber or a cabled in attenuators/channel simulator.

Yes, this is an outlier and it took around 8 years for this to be emerge and be detected by the community run TTN mapper initiative. The main fun thing about this is more that it shows how cool it is to have a global community network where we enjoy sending messages for the sake of it just to learn technology. Like kids with 2 cans and a string between them.

Hi. Cool work. For this particular record, could you confirm that this was on 868MHz?

Yes, this was 868MHz

There is never any mention of bitrate when talking about LoRa. It's infuriating from someone on the outside.

From what I gather, the map says a "SF:12" which means a spread factor of 12, at a bandwidth of 125khz. With a code rate of '1' according to https://www.rfwireless-world.com/calculators/LoRa-Data-Rate-...

this means about 292 bits-per-second transmission rate for the packet. But I'm unsure if this is correct.

How hard would it have been to put that in the post? I feel like there is some conspiracy to never talk about bitrates/transfer rates in the LoRa world. Like, ever.

Just say it. It helps people understand what this tech is for.

Good point. We'll add that to the next post if there is a new record. Also what can be confusing is that the business viable part is even much smaller than the technical viable part. The low power operations is really only at lower SF7 or SF8. And we push our partners and ecosystem to keep the payload as small as possible as every byte counts. It is a completely different way of thinking than WiFi or cellular. Yet I do agree it is presented with the same terms and paradigm and that is confusing.

> It is a completely different way of thinking...

On that point, have you looked at Myriota's research? (https://myriota.com/) In developing their system they went right back to the fundamentals and reworked everything around the idea of short messages. For example, an implicit assumption in the Shannon bound is that the message length is long, so they figured out that their short message system doesn't obey the Shannon bound and designed the coding and so on accordingly. It's fascinating stuff if you're into information theory.

You can do a lot with 292 bits per second if you use it well.

Yeah we now support 1.5M devices with our community network and enterprise network. And the most successful ones sent the least amount of data. Here is a list of all the supported devices: https://www.thethingsnetwork.org/device-repository/

fwiw lora is meant for extreme low-bw applications, like a few key-value pairs per hour.

maybe not mentioning the bitrate discourages misappropriation?

The things people are able to do these days with low rank approximation is crazy

This has nothing to do with low-rank approximation.

https://en.m.wikipedia.org/wiki/LoRa

Pretty sure this is meant as a joke...

> This has nothing to do with low-rank approximation.

Literal-minded people are an endless source of joy.

With all the naive AI zealotry on HN can you really fault me?

They're referring to this:

https://arxiv.org/abs/2106.09685

There exists technology that ISN'T Ai...

no jokes allowed on HN, sorry

I'll be quiet and good

In the last few years, the RC market (drones, airplanes, etc) has shifted to LoRa for the control link. A tiny, battery-powered transmitter/receiver pair lets me go 50km easily, which is just amazing.

ExpressLRS has taken everything by storm, partly because it's a well-governed open source project, and it manages to innovate many times faster than the old companies manage to fix bugs in their own products.

So I know LoRA: Low-Rank Adaptation of Large Language Models and Using LoRA for Efficient Stable Diffusion Fine-Tuning. But from the text, I don't think this is about it? It seems it's this: https://en.wikipedia.org/wiki/LoRa

Correct. It’s a low bandwidth/ long distance RF protocol over specifics bands. It’s also kinda proprietary.

check out ham radio WSPR for an indication of the sorts of things that can be done

WSPR is very different than LoRa. WSPR is usually used at HF, and has very low bandwidth. LoRa is used at UHF frequencies, and has very wide bandwidth, as it's a spread spectrum modulation scheme.

There's more differences but those are the ones relevant to this discussion.

Not super familiar with LoRa, can you explain what makes it properietary? Is it the protocol?

They've taken out a bunch of patents for instance:

https://patents.google.com/patent/US20160094269A1/en

That's right. Microsoft must have used Bing to see if the name was taken, and got no results, when they named their thing LoRA.

The added this line to their LoRA Github though: "(For the radio communication technique, see LoRa.)"

LoRa vs LoRA. The former is what the link is about which is some cool radio stuff.

How does this work?

At this distance you need a 130km-high tower for line of sight due to earth’s curvature. The previous record was set on a balloon flying at 38km height. LoRa uses high frequencies so shouldn’t benefit from the ground effect?

probably the same amateur radio people can connect to people halfway across the globe; bouncing off an atmospheric layer.

That doesn't work well on 433MHz oder 868MHz. Ionosphere reflection stuff usually happens <30MHz.

On VHF/UHF it's usually atmospheric ducting where radio signals are getting reflected between two layers of air.

99% this was atmospheric ducting. Also LoRa is an incredibly robust modulation.

It's a one-off, you won't be able to achieve this with any degree of reliability in practice, but it's a nice milestone achievement.

Nice, but don't forget that LoRa® LoRaWAN® and The Things Network® is all proprietary and patented stuff. The article is advertising.

This is not entirely accurate. LoRa is a patented RF technology owned by Semtech. LoRaWAN is a standard maintained by the LoRa Alliance (which is not very different from other standards and marketing bodies). The Things Network (founded by @wienke and myself) is a developer community around LoRaWAN and a free to use cloud service intended for R&D and non-commercial use cases.

Yes, so GP is essentially correct.

Please don't astroturf a corporate driven "developer community"

> marketing bodies

There you go.

What is astroturf? If I Google it I see fake grass.

A "grassroots" initiative comes from the people. "Astroturf" is commercial or political interests that masquerade as grassroots.

You can get a LoRa transmitter/receiver for like $5 and setup your own network. Documentation and datasheets are freely available. They are really not printing money with that technology.

There isn't any opensource modulation scheme that comes even close to LoRa.

> They are really not printing money with that technology.

Not until it becomes widespread. For-profit companies are not charities and they patent things for a reason.

So it takes more than 5 dollars. You also need a gateway but you can get these for 99 dollars. Still this technology has a very low barrier to entry for such a long range and low power capability.

For people reading this since I'm certain wienke knows this, you can also do point to point or mesh LoRA between cheap devices (e.g. Meshtastic). You don't necessarily need a multi-channel gateway, depends on what you want to do.

You only need a gateway for LoRaWAN, you can do point to point or mesh LoRa without a gateway.

Yeah... But WiFi doesn't work when I'm sitting on the toilet two rooms down the corridor. Crazy times.

No, to be honest: Very impressive. It's kind of a "dream" of me, to bring Internet to the river/picknick area two kilometers away from me. LoRa would probably be the way to go, will have to look further into that.

For sure would be great but LoRa has limitations as it is designed for small sensors that sent very little data: https://www.thethingsnetwork.org/docs/lorawan/limitations/

Ah, I see. Seems like directional antennas for WiFi would be what I need then.

Yes. 2.4GHz ones will do fine (and better than 5GHz for that range).

You're looking LoRaWan which is not the same as LoRa. In fact I believe web browsing to be quite feasible if you just use LoRa with your own network stack.

How would that exactly work?

I did this once as a small experiment. Basically use a small TCP stack with LoRa as layer 1 and then expose the whole thing as a Linux kernel module. It was a bit shitty but it seemed promising.

For some reason, I don't find LoRaWan at all alluring. As far as I can tell, in the end it's in the hands of one company and that really defeats the spirit for me. I mostly just use LoRa for point-to-point connections with my own shitty networking stack.

At least on 868 MHz, if you stick to the 1% duty cycle limitations that the ISM band allows, I doubt that you'll be happy with the web surfing performance.

(For those that don't know duty cycle limitations: https://www.thethingsnetwork.org/docs/lorawan/duty-cycle/)

This may be an EU duty cycle vs US time per channel thing? In the US you can frequency hop to get higher performance (the limit here is no more than 400ms per 20s per channel), which seems more aggressive than the 1% power sub-band EU rule (but ianal and I haven't read the EU rules carefully).

That is pretty cool. You might want to try again with 2.4Ghz LoRa. Range is shorter but still longer than 868mhz. If you search 2.4Ghz LoRa on our Youtube channel you'll find some interesting content on that.

Not at all. The effective speed is extremely low due to the duty cycle.

LoRa has pretty strict limits on how often you may use the frequency. That severely limits use cases. Browsing the internet is completely unfeasible.

From the relatively little I understand the lower the radio frequency the lower the bandwidth but more range it has/can pass through, while conversely the higher frequency the more bandwidth but shorter range/harder to penetrate obstacles (see eg: 5G). LoRa is the former and used for very low bandwidth applications, so not really suitable for browsing websites but has been used for for remote IoT stats, text messaging, etc.

(Fwiw I wasn't the one who downvoted but thought it'd be worth mentioning)

It's not so much about the frequency (though what you say is true), as there's LoRa at 2.4 GHz as well. I'm guessing the one in the article was 868/915 MHz, though.

Define "internet". LoRA is more like an occasional SMS service than a network connection, so I doubt you could get the average modern 100 MB website to render through it. Would be more practical to get a Starlink dish set up or a directional wifi antenna on both ends, those can do a few km on a good day.

Or just you know, use the 4G that is probably there already.

> Or just you know, use the 4G that is probably there already.

Yeah... Projects like open WiFi are probably becoming less and less useful, as everybody is carrying 4G/5G modems in their pocket.

At some point I was bullish on LoRA + Gemini or scruttlebutts.

A different, low resolution internet basically. But LoRA is too slow for a lot of things. Too.

Still nice.

In the end, our butts remained unscruttled.

The difference is the 160bps they have with this isn't going to be too much fun for web browsing!

A good explanation of LoRa, and in particular of the "link budget" idea which makes LoRa different from other wireless tech:

https://www.youtube.com/watch?v=hMOwbNUpDQA&list=PL3XBzmAj53...