I want to start a project where i measure some stuff like water level in a forest and transmit it via lorawan to my cabin, where i would like to install a gateway and connect the data to TTN.
The signal would have to travel about 600m with at least 100m of pretty dense vegetation and about 500m of plains. (see image)
Is this coverage possible? and if so, can i assemble my own gateway with the RAK2287 module connected to a raspberry Pi & transmit from a simple Arduino Uno with the Dragino shield (RFM95W)?
Or would i require more expensive stuff like The Things outdoor gateway for coverage in the forest?
I suppose that The Things Indoor gateway would definitely not work for this?
The end node would have to be pretty close to the ground for the water level measurement, but maybe i could attach the antenna in a tree? or would this only increase the interference as it needs to pass through the leaves as well?
With some trying things out and tests and putting antennas on poles and so forth, 600m should be possible.
If I had a customer enquiry like this, Iād say that weād have to try things out, do tests, put antennas on poles and so forth.
So I guess thatās where you are heading.
As well as checking on mobile internet speed - LoRaWAN really needs a responsive connection - so high latency can stop the gateway processing some messages properly.
Also consider that if the trees are deciduous, they will add a whole new level of signal blocking technology in spring.
Note that is those tests the transmitter was lying in the ground, which is far from ideal. The tests were at 434Mhz, and estimations of distances possible are given at the ISM band limit of 10mW. At 868Mhz (TTN) although free space loss is more (so shorter distances) the power used is 4dBm higher so overall distances will be about the same.
Thanks for the replies! The lora coverage in the forest experiment gave me some confidence.
I will order the components and keep you up to date after some testing. fingers crossed
Most of the path you highlight above is actually not woodedā¦this will be a no brainer from my own tests arounded wooded areas (e.g. Burnham Beeches, Kielder Forest, Portman Estate, etc.) here in UK and in woodland around the Arno Valley in Italy or the LĆ¼neburger Heide in Germany (covering lots of different tree and moisture types between these various area)ā¦ I would regularly get >750m to >>1.8km through dense forestry and woodland. Real limit for any given path of such short duration more often topographic. When wandering in deep wood/forestry it is often difficult to tell if end points (GW & Node) are hidden from each other by even gentle/slow undulations in the terrain. Assuming you can get the GW up high enough then even that can often be mitigated against.
Some examples:
That was over 1km where >50% of distance between was wooded
The gapping in the N.West part of the track in 2nd picture was caused by the path between samples falling some 30-50m below the level of the GW and becoming āterrain maskedāā¦
Another example where terrain was approx 50% wooded between the node & GWā¦
Up in the Kielder Forest and around Kielder Water I have successfully seen ranges between 3 & 5km with no issues. (Was looking at tracking potential for workers moving between 500m & 2.5km from a base vehicle and so didnt push out much furtherā¦a test for next year! )
Half my High Altitude Balloons have ended up in trees and they happily relay pictures of the canopy tops, so it works in reverse as well, tx up, rx wandering around in the brambles.
Part of the reason behing the test, was to do a direct comparison of an identical link and between a dry forest and a wet one. So many times I had heard the comment about the dire affect it being wet has on range, the reality appeared to be not so bad, 5db.
I dunno, sometimes a little water can really spoil your range:
There being a pun in there about range - it landed just metres in the sea just off the old RAF range at Skegness, didnāt stop us looking for it for an hour. Was found on the beach the next day.
I will let you into a secret, and this is relavent to TTN users, LoRa (and radio stuff in general) does not go very far if the transmitter is under water.
Not tried that one yet. Maybe some sort of buoy on a string to get the antenna to the surface?
This one floated just fine, itās just it wasnāt 100% sealed and the salt water went to town on the batteries.
As for forest based devices, one HAB team did have to follow their payload until it had stopped moving - it was found by teenagers who were rather surprised when asked politely for the box back.
I find it very variable - depending on the planting density of the trees, stage of growth (how big the trunk), type of tree, some variance based on how āwetā the forest is (both in terms of rain covering external but also how recently heat stressed the trees are as internal water content/conductivity varies). Generally allowing for RSSI starting at around -30ā40dBm when node close to GW then moving away to point where RSSI is say -117ā121dbm, (SNR in these remote areas is ofetn favourable also helping achieve good range) that will equate to a range well in excess of 1km and often 3-5km suggesting a lower per100m loss than you have found. Also find plantations often better than dense wild woodlands due to fact former typically has much lower under growth (better managed?) which also impacts low height sensors (most of my tests done as walking tests with sensor ARO 1-2m from ground level).
I did not try to find out the maximum distance. The gateway was abt 40km away. I measured the difference of the RSSI-value between being in front of the forest and being inside the forest (SF7/125kHz). Because my node was more then 1000m MSL high, it could āseeā the gateway from above. The Line of Sight was disturbed only by the trees.
I fully agree with you, 10dB/100m is only an estimation. The forest was dry. BTW: there is an interesting document dealing with this problem āEstimating Low-power Radio Signal Attenuation in Forestsā published by DISI, University of Trento, Italy
If the Gateway was already 40km away, you would probably have been receiving signals at the non-linear part of the RSSI readings. Using a LoRa devices RSSI to measure the differences in reception is fraught with problems really, its just not consistent enough, especially at weak signal levels.
The testing I have done measures the actual transmit power needed for a transmission to succeed and thus avoids the issues around using RSSI as a measurement.
Interesting thread, Iām investigating if LoRaWAN sensors can be used to monitor bat boxes, for motion (bat moving in and out) and temperature/humidity reading. Currently weāre doing this for a bat box in a residential area (we got some of the first signals today, BTW!) but we may also deploy it in a wooded area. Bat boxes are generally mounted at an altitude of at least 4m.
Hi there,
IĀ“m working on wildlife tracking in forested areas and struggle with
coverage. We do have our first wild boar tagged and hunt for signal.
Great to see your results.
Greetings,
Robin
Another document I recommend: ITU-R P.833-4 āAttenuation in vegetationā. There is a very interesting diagram concerning āSpecific Attenuation due to woodlandā.
That was over 1km where >50% of distance between was wooded
)
.