Gateway - LTE data usage


Do any one have stats on amount of data there gateway use for the backend connection?

I know there are a few factors that will influence the date consumed, like amount of uplinks and downlinks.

If you could state the actual data consume and the amount of up and down links per month

But I just want to form a general idea, as for us buying fixed monthly packages on LTE are a lot cheaper than flexible packages. (and a ISP connection is to costly and I know I have LTE connection available at the site)

I don’t think theoretical calculations are accurate enough.

If you go fixed, make sure to figure out what will happen if traffic increases and you drastically exceed expectations.

Yes, if the fix package is to small then all the data will be dropped.

If the difference between a fix and flexible package is in the region of a factor between 10 and 20, worth the swap out of the sim.

I have a fair idea of the amount of up and down links, so if someone have some actual stats it will eliminate some of the guess work.

Please search the forum? This question has been asked and answered a number of times over the past years.

Old post 2018 V2 stack

Post with reference to usage on back end but not on up and down links and at 1G or 2G a day seems out of proportion.

@kersing I have looked but cant find anything wit some descent stats, except if I have misread any post

You focus on two numbers @jpmeijers considers unrealistic as well and ignore all the other data quoted.
There likely won’t be any corresponding up-/downlink numbers available because there is no way to get those numbers. TTN only showed cumulative numbers in V2 and only amounts for the current ‘session’ in V3. For V3 the duration of the session is not displayed so a session might be weeks or minutes.

A little searching for more information (it might be embedded in a larger discussion), there is a lot more but I don’t have the time right now to search for it:

Keep in mind that anyone deploying LoRaWAN nodes near the gateway (any LoRaWAN node, regardless of which target network) will impact your gateway traffic. At least one forum member ‘suffered’ from a deployment of hundreds of devices in buildings within gateway reach resulting in on average one LoRaWAN message a second. That quickly increases the traffic from a couple of MB a day to tens of MBs a day.

I have. I will look at it tomorrow


I have several gateways 7249/7258 RAK Wireless with SIM

All these gateways are connected to WisDM Cloud of RAK Wireless so a (great) part of the data is used to communicated with WisDM in the Cloud

Uplink 72.005
Downlink 9.098
MB: 91

Uplink 20.137
Downlink 3.076
MD: 75

Uplink 211.018
Downlink 3.349
MB 107

Uplink 51.229
Downlink 246
MB 89

Uplink 696.478
Downlink 2.697
MB 173


1 Like

Thank you for your response

What time period is this for? (Day, Month)

I take this is the API call that gives you the average of Up/Down links for the last few minutes. (same API as below.)


What makes you believe that? A remote administration solution not being really asked to do anything should really just be putting the occasional keepalive packet through, it would only be “a great part” of the overall data if there was little LoRaWAN packet traffic.

Hi @Johan_Scheepers

Sorry for not be clear or complete.

The MB is per gateway per month. This was December 2021

The number of messages is form the gateway interface itself via WisDM. I check every 1st of the month to totals and I calculate the uplink/downlink messages for that month

Hope I answered your question


It’s a guess. But your remark made be think… before the RAK gateways where on WisDM cloud …

I moved most of them to WisDM cloud ed of Nov 2021

Normally one gateway had about 90 MB per month
Since the migration to WisDM Cloud (2 months history now) about 170 MB

Therefore: for this gateway (696.478 / 2.697, the last in the row) about 90 MB’s per month on data to TTN and 80 MB’s per month for data to WisDM


1 Like

Maybe, but that’s assuming unchanged conditions.

I’d be inclined to look at per-process accounting by the kernel, or to see what a gateway that ran for a day or two unable to receive anything (wrong frequency plan, whatever) was consuming.

If the remote administration were really consuming significant data compared to the traffic, it would be an argument for replacing it with a more efficient scheme.

Hi @MarcVanBracht

Thank you for your feed back, doing a theoretical calculation leave the other bits out (keep alive and management). This gives me better insight. Want to deploy a gateway in a very low density of populated area, less than one person per square kilometer.

So this is 72005 uplinks and 9098 per month? (Can’t be fractions)

I measured the data-usage of my Dragino LIG16 (latest firmware) connected to TTS by a LTE-stick Huawei e3372h-320 and a TP-link WR902 router. The gateway received no nodes.
It used 6MB per 24 hours.

Edit: I use Semtech UDP


:thinking: So approx 180MB/mo with no node traffic?! Seems high if just on UDP - some of my low traffic UDP GW’s see 65-100MB/mo (they are on 150 or 250MB/Mo sim plans)… might be worth investigating what that traffic is and how configured (if you can be bothered! :wink: )

1 Like

…just dont attempt to do Mining with a GW on a cellular connection at the same time by e.g. attempting to dual home on TTN and Helium (or even run solo on Helium on cell connection?) - I hear HNT GW’s currently need to consume around 40-60GB/mo to support the ledger (though have been told on the Helium discussion thread that there are plans for ‘lighter’ Hotspots that will use cloud services - not sure what data consumption will be then) :wink:

Definitely won’t add a gateway to Helium, at this point.

But I want to add a node ad see if there are any coverage in Mt area, as all the gateway have strange names for SA, like …lion…, this is not a norm on how people here name things

All Helium gateways have the What Three Words style format - it does sort of make it easier at a community level as they are based on proofs so visibility is everything.

I don’t know what happens under the hood - maybe one of us should descend in to Dante’s Inferno?

On a seperate matter and point on this (off topic I know!) I really cant understand the point of this system - whilst 3 random words is potentially ‘unique’ to identify a location IMHO its a retrograde step from simple GPS/Lat/Long which is also unique & also has advantage of adding altitude. If I see Apple-Humber-Mountain and Blackbird-Tulip-Firehose (deity of choice knows where these are - I havent checked and chose at random!) I have absoluetly no clue which direction one is relative to the other or an inckling of how far apart they are… Lat1-Long1 vs Lat2-Long2 I will know roughly where in the world, east or west of Greenwich meridian, north or south of equator and depending on size of any delta between relative number I will at least have a clue as to how far apart. WTW tells me nothing where as 51.6134827, -0.644497 to 53.574935, -2.426964 will at least tells me that both locations are west of Greenwich, both are in the Northern hemisphere and the increasing magnitude of numbers tells me the 2nd is approx N./N.West of 1st. From experience of scaling I know the as crow flies distance must be of the order of 150miles - (135-165?). Far more useful than 3 random words can ever be! (BTW the two points are approx Beaconsfield Railway Station and Bolton Railway Station! :wink: ) From that I can assume and estimated ~3.5 hrs (3-4?) motorised travel in UK without allowing for stops and diversions and depending on travel method, where as 3 random words may be 3 minutes away in same village or may be 3 days away on the otherside of the world! Just saying… :man_shrugging: