I’m running into a puzzling issue with gateways on our large college campus. When gateways are installed indoors, they work fine. I can often get 1/2 mile of coverage and they can access multiple floors. However, when I move the gateways outdoors—even using the same models, antennas, and mounting positions—performance drops significantly. The gateway is using a 4g SIM card so it isn’t dependent on any other local network.
Here’s what I’ve tried so far:
Multiple gateway suppliers and models (5 and counting - 3 indoor 2 outdoor models)
Different dB antennas
Trying indoor models outdoors and outdoor models indoors. Outdoor models with higher db antennas don’t work indoors or outdoors. But the indoor models don’t seem to work well outdoors either.
The pattern seems clear: indoors = good reception, outdoors = unreliable.
I suspect this might be related to RF interference. There may be other departments or campus systems operating in the same frequency band, possibly creating noise.
My questions are:
Has anyone experienced a similar issue with outdoor gateways?
Are there practical steps for identifying interference sources on unlicensed bands like this? Like low cost devices I can buy or a service provider/consultant who can help me identify issues.
If interference is confirmed, is there any recourse—e.g., involving the FCC—or is it just part of using unlicensed spectrum?
What devices are you using to assess the ‘performance’ of the GW’s? Are they indoors - in same environment as the GW’s or outdoors? What distances, and with what intervening built environment and screening? Have you analysed the metadata of the received signals at the GW’s to see how the RSSI’s and SNR’s associated with the target devices you are using for the assement? Individual values may not tell you much but analysis of volumes of info and trends may tell you a lot. Remember that whilst RF can be a bit of a black art there are some fundamentals that will likely apply. e.g. RF noise floor is aggregated and is additive……the more sources of interference and the louder they shout the higher the noise floor and consequenty the low the SNR for a given signal. It may be that moving the GW’s outside - especially if moving them higher - exposes them to more sources & higher intereferer levels. When moved inside those external interferers will see higher attenuation from the building materials and therefore the noise floor may fall. If the sensors are reasonably close then from the GW’s perspective it may appear that the sensors signal gains headroom, even if the sensors signal itself is also seeing some additional attenuation. If the sensors signal drops from say -95dbm to -105dbm from a LoRa signal processing perspective that is still a good and very usable signal level, if the noise floor is similarly attenuated by ~10dbm that will improve the SNR and may make the received signal far more useable. Remember that for any given SF there is a limit to how low/high the SNR can be for reliable demodulation, if signal is marginal or occationally droping out at a given SNR then improving the SNR by filtering/attenuating the extenal noise sources can have a remarkable impact on improving the signal capture. e.g. a SF7 signal with SNR @ -10 might not be detected reliably, if at all, but improve that to 0 or even -3 say by improving the noise floor and suddenly you have a very reliable and stable signal for the same RSSI
LoRa’s response and mitigation in the face of other RF sources as interferers, even other LoRa devices will vary by the type of modulation of those sources, or in the case of a LoRa device will depend on the relative SF of the desired target device vs the SF of the interferer(s) (There is information avalable and marketing/research info published on this TL:DR for here - GIYF)
A SDR/spectrum analyser may also be useful in helping you identify the type and relative strength of any interferers or the state of the local noise floor…..
Your ref to FCC suggests you are US based? Sadly whilst there is the positive of higher available Tx power and no duty cycle limits when compared to some other territories, there is also the negative of interferers also having option of higher Tx power and no enforced duty cycle! (Though the latter somewhat mitigated by dwell time limits and associated RF channel hopping, however, poor broadband vs narrow band antennas and limited/cheap RF filtering and device matching may limit the effectiveness of that mitigation)
Thanks for your detailed response - way more than I expected!
What devices are you using to assess the ‘performance’ of the GW’s?
Up until now it was just using my temp/humidity sensors. I have milesight AM102 (internal antenna) and some dragino devices (external antenna and the ability to manually trigger a uplink. I recently purchased a Rak Field Tester and RAK Gateway RAK7268CV2 to try to gather more meaningful data. When testing the end devices I test them both indoors and outdoors. I test them at ground level walking around and they typically get 1/4-1/2 mile range to my indoor gateway. It’s very flat here so no elevation changes while walking around. I have mostly only had luck with line of sight to the building where the gateway is located. I’ll know more once I get out in the field with this RAK field tester, which also has gps so that should hopefully speed up my analysis.
I haven’t done much regarding analyzing the SNR or RSSI but I will with this new Field Tester and RAK gateway. I can report back what I learn. I think my impetus for this post is once I do this analysis is there even anything I can do… You’re right I am based in the USA. So I was hoping if I found an obvious culprit I’d have some pathway to pursue.
My working theory is the same as you mentioned. That when I mount these high on buildings with higher gain antennas I’m exposing them to much more interference.
Like you suggested, I may end up getting a SDR with a YAGI antenna and try to determine if there’s at least a primary direction the interference is coming from.
In the end, the problem I want to solve is connecting up some LoRa modbus dragino devices to random remote energy and water meters that are spread out over campus that are too difficult or expensive to connect. Like water meters in ground wells. So getting 1-3 miles radius of coverage is the goal.
I’ll circle back with some more data once I get out there with the field tester and I’ll use the resources and information you provided to help me understand it better myself.
TL:DR Dont bother as higher gains can cause more problems!
Systems are typically spec’d for use with 2-3dbi ants, so that gives best, closest to isotropic radial coverage. As you go to higher gains you focus more on the horizon, and depending on the implementation of the ant higher gain units start to sacrifice near field and intermediate coverage/signal strength as they typically loos smooth coverage and introduce nulls at various radial distances. Also makes them more suseptable to near range angular loss of signal. If you put a hogh gain ant high up outsde your often find that previously good close in sensors that are below the ant and its reception field start to show issues.
Biggest I set up are 5/5.9db ant and that is under juress, if I ‘must’ reach an outlying device - closer to getting a point to point connection, I then back fill with a smaller GW using lower gain ant to ensure local coverage. Usually for community (or campus!) deployments even coverage for end nodes is the goal - unlike some networks which incentivise GW to GW comms: looking at you Helium! forcing use of stupid gain figures to get to distant GW’s at cost of poor local coverage - so stay low! There is a reason why emergency personell comms systems use low gain ants (0-3dbi) to ensure all round coverage
Check radiation patterns for those antennas and you will know why it’s a bad idea. Apart from having to dial down gateway transmission energy to stay within legal boundaries. (When receiving the pattern will be almost identical to the transmission pattern)
In practical outdoor deployments of LoRaWAN, the effective communication range is significantly influenced by interference and antenna placement. According to a study by the University of Ilmenau, interference can reduce receiver sensitivity by up to 24 dB. Outdoors, there are numerous potential interference sources beyond just the intended signals, which degrade signal quality and reduce range.
Read more: https://www.akoriot.com/mioty-versus-lorawan/
Another critical factor is the Fresnel zone clearance. When sensors are mounted close to the ground, the Fresnel zone is often obstructed, causing a loss of around 6 to 10 dB in signal strength. This reduction is explained in detail in the LPWAN Cookbook, highlighting how ground-level mounting affects radio wave propagation by disrupting the Fresnel zone, thereby weakening the received signal.
Read more: LPWAN Cookbook - Antennity - Integrated IoT antennas
To maximise LoRaWAN range in outdoor settings, careful consideration should be given to minimising interference and ensuring clear Fresnel zones, for example, by mounting sensors and gateways at elevated positions. This approach helps approximate the high link budget that LoRaWAN can theoretically achieve in ideal conditions.
