Gateway recommendation with low bandwidth usage

Hi, there is a vague plan to cover the Hawaiian islands with LoRaWAN. Essentially up two 100 gateways could be deployed around the islands, ideally offering TTN LoRaWAN coverage on all mountains and in all valleys.

The deployment would use existing cellular connections which are limited to about 20MB/month. Would that be sufficient for gateways? I’m interested in the possible traffic handling as well as telemetry used by the gateways.

From my understanding a LoRaWAN package received by a gateway is forwarded to TTN with additional metadata information. How git is one of those?

Also, are there gateways known to use low/high telemetry or “phone to home” bandwidth? If a gateway would automatically install a 20MB firmware upgrade, the cellular connection would be overloaded.

Are the TTN Outdoof Gateways the best choice for long term deployments or would it make sense to buy the cheaper alternatives like MikroTiks wAP LR9 kit?

Thanks for all further information!

Sounds like a brilliant community plan with a huge amount of potential.

I’d strongly recommend that you get yourself two or three gateways and half a dozen devices so you can see the detail for yourself because there are many many variables - topographically you could have two gateways covering one major conurbation but find you need three just in one valley alone, so being able to run some test installs will give you a much better picture.

Generally with gateway deployment you start out with some obvious high places and then do some signal mapping to find out where the dead spots are - and consider if they need filling in. If there is a resident in the area who’s happy to host on their own internet connection this may be a good compromise rather than having something in the middle of no-where that’s not really providing any benefit.

It is sort of useful to work backwards from a connection budget, unfortunately this won’t work too well. If a gateway on cellular has many devices around it, it will get a lot of traffic - the devices may be doing very very important & useful sensing - but then come to a grinding halt as soon as the 20Mb limit is reached.

If you search the forum you’ll find a number of discussions on the elusive answer to “how many gateways, how much data” - most of which conclude with “situations vary” - but they will highlight many of the details.

I would strongly recommend these two strategies:

  1. Run two or three significantly useful proof of concept trials - this should help with funding.
  2. Commission a northern UK consultant to come out to help run the trials …

Maybe the local telecoms could provide free data plans when they see it’s not a huge amount of data and they get good PR?

I’ve got a DLOS8 deployed with reverse ssh for management. There are just a few nodes in its coverage area (looking at traffic in TTN console). According to the billing by the Telco the gateway uses 130-200MB a month.
Another (TTOG) with 3G and no remote management but with a few dozen nodes in the coverage area uses about the same.
Based on my experience 20MB/month will not work, you need at least 10 times that amount for coverage area with a few nodes sending just a few times each day.

1 Like

Commission a northern UK consultant to come out to help run the trials …I gather Jeff-UK may be available and his costs are reasonable

There finished that for you! :rofl:

He also knows from experience that putting a GW high on a hill isnt always the best way to monitor that hill or the valley bottom below…and the gw is perhaps best placed to monitor the hill on the other side of the valley, and a GW placed on the opposite side may be best for monitoring original hill territory etc. If the valley meandres then a GW placed on a hill/hillside at the top of a meander may be well placed to cover valley bottom down stream as well as the flanking sides… :wink:

@spooren Paul, I have a number of 3g/4g enabled deployments and most come in at around 55-85MB per month, some up to 120-130MB, so most of mine are deployed with 100MB plans (with small risk of overtopping - I can use PAYG to top up if/when needed just in case) or 150MB/250MB plans, as with Jac @kersing one sees regular bursts from 160->200MB so is on a 250MB plan - the higher plans headroom is also good if field service eng needs to piggy back on connection briefly to say do GW or device in field s/w update, download a datasheet or look up internet advice etc. or simply to log into TTN Console to view traffic and assist debug operational checks etc. (Though I/they usually equipped with other cellular connections when on site so as not to burn through plans). If you disable auto firmware updates for the GW then that removes risk of saturating as data then mostly limited to GW status updates (typically every 30seconds depending on config - so very deterministic for a given month) and then the variable of just how many nodes are visible to the GW - and associated variable payload sizes/update ratespotential downlinks etc… Ironically the higher the GW the greater the likelyhood the better the coverage and hence the greater the number of nodes seen - even picking up nodes from outside the areas that GW is intentionally focussed on. You can’t control this easily. Also remember GW’s are transparent and pass on to the NS ANY and ALL valid LoRaWAN messages they see - its the NS that then drops any unwanted or ‘foreign’/off network messages. If there are private or other public LoRaWAN networks deployed in the coverage area your gateways will see, handle, and backhaul all that traffic as well! So actual data amount may be well in excess of what you expect from your own/communities nodes… Also finally. remember, whatever you plan for today if you are successful in depoying and then providing a utility others can enjoy then future nodes may significantly increase traffic, also even if no other network deployed today that doesnt mean one wont appear next month, next year or whatever - so adopt a flexibile/upgradable billing plan just in case. (One of my old clients deployed a 3g enabled GW a couple of years back, recently the local council contracted with a street lighting vendor to ‘upgrade’ old system to newer energy efficient LED lumiares, monitored and managed by a private LoRaWAN Network…their data jumped from steady 56-58MB/mo to >120MB/mo over a 2 month period as lighting was commissioned - they then had 2 months where they ran out of data a few weeks into billing period and as plan was capped vs move to variable charging they lost some data, whilst they figured out what had gone wrong. With variable they would have still received their own data albeit with a small financial hit. I hear they have since moved on to a (IIRC) 200/250B plan, with variable overshoot option, just in case…)

1 Like

If you are going to operate that many gateways, I really suggest looking into remote management methods, so you don’t have to go there physically when you need to change a setting/update the firmware. I’m looking into this myself aswell, maybe will go for reverse ssh as well like @kersing. This will however add to the data usage.

Maybe a tip for some people who manage multiple 4G gateways:
Some of the commercial SIM providers have options to pool the data of the sims together.
So for example if I buy 10 sims with 150MB of data, it all gets added into a large pool of 1.5GB which they will all draw from.
This way it doesn’t really matter if one gateway uses a little more, as long as the average over all the devices in the pool is <= 150MB.

PS:
I’ve seen a gateway use 14MB per DAY at our office with 70+ devices around

1 Like

Thanks everyone for the quick response! I see that 25MB/m is no good option, the current vendor seem to offer a slightly more expensive 200MB/m package, seems like the way to go. For the last day I ran a TTIG which received a 30Byte package every minute, daily gateway traffic summed up to about one MegaByte.

Looking at the basicstation documentation it seems an “easier” solution to use CUPS instead of a SSH connection. From my understanding this would allow both firmware upgrades as well as remote management?

Since this is for now an altruistic effort to pave the ground for some future scientists to simplify their research, a contractor isn’t in the budget. However, thank you very much for the kind offer and recommendation. If that changes I’ll reach out again!

I’m sure you will be aware of the Fair Use Policy that this comprehensively breached by a fair margin!

https://avbentem.github.io/airtime-calculator/ttn/us915/30

I don’t understand why not - if I sweeten the deal by leaving some kit behind when I finish, as well as pay for my airfare plus all living expenses and several rounds of drinks, can I still come over to help. :pleading_face: