A field test in the woods. How is the Helium network holding?
I took a recent trip hiking in the nearby mountains. I have been on this particular path before, it is about 30 km by car from the town I live in, followed by a 9 km hike up in the mountains culminating at a local cottage that is 1500 m+ high. It is not a difficult one, however, I have been postponing it for some time. The reason being that this was to be partially for relaxation, partly for business.
I had this idea to put a weather station high up at the cottage, so people can check the weather up there before they head out, making sure they are not caught off guard. Being a LoRaWAN enthusiast I thought this is going to be the network to be utilized, seeing how it is free to use and has long range. And long it turned out to be, indeed.
So, I took my RAK10701 Helium edition, and an energy drink (I am addicted to the taste the track is not actually that hard so to need it) and set out. Now, I am using the Helium edition for two main reasons:
- This is the latest one I got from RAK (sponsored), which works out of the box, no need to configure it, and you won’t be charged Helium DC.
- Network coverage. I do have a TTN gateway deployed with a good line of sight in town, but there are more than a hundred Helium Hotspots so one would expect better coverage. This is impressive given that this is a small 20k town in a poor, derdeveloping country and the IoT means absolutely nothing to most people.
So, you can think of this try as a mapping campaign, see where there is coverage, where there is not. I have done these before, mostly cycling in a 30km radius around town, you can check a post about it here.
I am just going to start by saying it. THE RESULTS WERE QUITE IMPRESSIVE. Just take a look at the image below. You can distinguish the green hexes as points with coverage and if you take a closer look, you can see that the entire length of the track (the part where I walked in the forest) had coverage.
The second image is a screenshot from the output of my Garmin Fenix 7X smartwatch. Garmin being the king of GPS, I am assuming this is a very good base for comparison and quite accurate. As was to be expected, the Garmin map is more detailed, but this is mostly due to the way it is displayed (in points not in hexes as the Helium mappers UI).
So far so good, GPS data is accurate and there are no large areas where there was no coverage and missed packets.
Let’s take a deeper look and try to interpret some data and gain some insight.
- The color of the hexes represents the signal level, where the more transparent it gets the lower the level. Throughout the entire length of the track, there isn’t a single spot with Received Signal Strength Indicator (RSSI) higher than -110dBm. This is pretty low, but not terribly so for a LoRaWAN network. It is also not expected, I would have assumed signal levels would not go much past -120dBm, given the fact that it is now very lush and most of the path is in the forest, with dense trees blocking the signal. There are a few patches with small meadows, so probably this is where the better signal levels were measured.
- No hex had coverage from less than 10 hotspots, this is quite unexpected, there are not that many in the area as it is mostly comprised of very small villages. This unexpected behavior can be explained by the fact that most of the hotspots that have received the packets are 30km+ away in the bigger cities of the province.
- The nearby town where I came from and I expected most of the hotspots within coverage to be has received almost no packets. I am attributing this to the hilly terrain which lead to them being blocked by land masses, so there was no Line of Sight (LoS).
- At the very end of the track where the highest point is and there is LoS for 100 km+, I expected to have the best coverage. This is true, this is the hex with the most hotspots seen, however to my surprise none of them were in the nearby town, all were very far, some even 140km+ away. In short, this means LoS is king, distance matters little to LoRaWAN.
- Last, but not least we take a more detailed look at the RSSI and Signal to Noise (SNR) levels. The image below lists the coverage at the very end of the trail, the point with the best LoS (open area). The RSSI levels are consistent, more or less the lower they get the further the hotspots are. What is a bit unexpected is the SNR. There are links with RSSI in the -120 dBm range, with SNR in the -5dB range, but there are also some within the same RSSI range, but SNR is worse than -10 dB. This is a lot of fluctuating and also a lot of noise, which one would not expect in a relatively low populated area as this one.
So, let’s sum it up and see what we learned:
- The hike was really fun, everything was very green and verdant.
- There was consistent coverage over the whole area despite the heavy foliage.
- The range was great as expected, hotspots as far as 140km were seen.
- There was more noise than expected, LoRaWAN is getting more and more popular and networks are handling a lot more capacity now, so noise and interference might become an issue.
- The RAK10701 used for the tests was transmitting very reliably, every 60 seconds with on SF10, 125kHz bandwidth, blasting at max power (14dBm EIRP). Given that the spreading factor can be dropped to SF12 we could squeeze a bit more range, but at the expense of network capacity and more interference. This would not be a realistic scenario, so we went with SF10 (also I forgot to check on the default settings).
- There were a lot more hotspots that saw the signal and most of them were further away than expected, I guess I do not know the topology of the area that well, turns out my hometown is not visible from that peak after all.
- The Helium network does seem to be working quite well, say what you will about their incentive model, tokenomics, and unreliable coverage (as they are peer hosted), but I would not have seen any LoRaWAN coverage in this area if I was utilizing TTN for example for sure, the gateway would simply not be visible. I know this for a fact as my hotspots are mounted at the spot in town with the best visibility and this measurement campaign proved it was not within range (due to no LoS).