yes @descartes and @Jeff-UK , the solution will be that the near farms work toghether and have each one their own gateway, will be good.
For me… i still have a problema, this project is to follow cows and horses that live in the mountain in council land with no internet there.
Still going back to the technical solution… esp32+gps+lora to a esp32+lora central receiver, this one sendind to the nearest gateway? what do you think?
The next issue is the price… i made my own devices …a GPS with lora is expensive for the farmers.
As highlighted earlier I think you will saturate the LoRa P2P collector to LoRaWAN GW link very quickly
Depends on messages (size and update rate), distance (hence SF & time on air - sadly for a tracker you cant realistically use ADR to optimise), number of monitored animals, terrain (BIG influencer on how any system will perform). Also
How will a given node ‘know’ it has gone out of geofence limit to then send a message - are you running the node continuously to monitor itself live doing local GPS processing and position comparison (big battery!?) - also how will you update any Geo fence data (typically needed if you are to proactively move animals around the area, unless geo fence just relates to the fixed outer limits of the estate)
I think you need to start with mapping the area and topology then look at how a system and deployment needs to be developed and amended (yes there will be compromises to be made - this is no one size fits all in such cases), to accomodate what you find on the ground so to speak. Look also at extending reach from deployed GW’s back to where you can effectively pick up internet connection - long range directional p2p Wifi or uWave links?
All comes back to cost as you suggest - I assume you are charging for your time developing this as after all that is in part what drives the cost of off the shelf devices you consider expensive! IME no farmer ever says the price is too low - everying is always too expensive 
I think the only way you’ll know is by trying something because all we are doing is indulging in advanced guessing based on prior experience.
Or alternatively, buying a tried & tested solution. Or even better, paying me & Jeff to run around a field making Moo’ing noises whilst playing with all the kit we’ve not got around to testing.
Farmers are always being sold something and so are rather cynical. It’s going to be about return on investment - if the tracker system demonstrably saves them money you should find it easier. The best way to do that is to try something.
1 Like
There is a testing ground by me, the local playing field, with trees on the edge, so it is perhaps a reasonable simulation of the limits of RSSI localisation, weathers nice tommorow so might give it a whirl. But whilst you can make a LoRa device play tunes (Star Wars for instance) I have yet to make one Moo…
RSSI is often (at least once a week) proposed over in the Arduino forums as a cunning wheeze method of locating stuff by triangulation, yet over the years no-one seems to have come up with a workable method. Last week there was even someone complaining that the RSSI varied so much it made the localization a waste of time, a fix was wanted.
One of the issues with trying to implement RSSI localisation is that antennas are so directive, move the angle of a transmitter that is 500m away around a bit and is may apparently move to 1.5km away. Same thing happens if an object, animal or person, gets in the way.
1 Like
I’d only expect “which field” for values of large fields for LoRa RSSI. Useful for finding things that come running when you shake the food bucket. Or big yellow things (JCB diggers) that are easy to spot.
I did some work a year back with someone who was trying out the Semtech API service for using gateway RSSI data to approximate a location - if it’s heard by at least three or four gateways it wasn’t entirely useless - certainly enough to point the police to where to search for the villain / body / treasure chest.
I guess you could use it for elephant detection in fridges to save on the butter use.
Bluetooth, according to some vendors, is a bit easier to narrow down a location. Apple’s iBeacon can apparently work some level of miracle as the Claris (ex FileMaker, ex Claris, always a subsidiary of Apple) demo along those lines is a gym client management solution that can tell which machine the instructor is standing next to …
I’d also like to have a go at using reverse RTK - rather than sending correction data to a tracker, you take the co-ordinates it sends and correct at the receiving base station (that is located in the locality of the item being tracked) as that would solve all sorts of silliness - like failing to find aforementioned big yellow things because they were behind a wall and numpty only looked on one side of the divide.
Easy enough to check.
All around the UK there are GPS reference stations, where the position is know to a couple of cm.
There is a easy accessible one near me, Sully Island. You can put a tracker actually on the reference station and see how far out the reported location is. There is even a nice view and a pub.
When should we come over!
Like I said, weathers nice tommorow.
I have used the reference point by the pub (The Captains Wife - The Captain's Wife - Wikipedia) to check the accuracy of the distance and direction calcs in the standard Arduino GPS libraries, which can be used on TTN nodes of course, so maybe this is an on topic TTN disscusion …
Some times you need to think outside the box, a place well out side the intended coverage area could sometimes be the correct place to place the gateway.
If I understand your use case you are well out in the country side, I have see much longer ranges possible in rural arears, here the noise factor is lower.
There is a gateway that is 63KM form nodes I have deployed, it is not the intended gateway for the nodes, but it gets the uplinks all the time, the intended gateway is 8km away, it also does it’s job well. (just bear in mind you need to test your specific case, by no means do I say you will get this distance, a lot of factors are to be accounted for)
They don’t just use RSSI, the use TOA as well. The RSSI for there solver can not be >-85, it makes the location invalid.
The technique of TOA is actually called trilateration, a technique used well before GPS for planes to locate themselves.
Fantastic ideas on the table.
I am in Portugal , Viana do castelo, if i were in England will go to the pub to meet you guys. 
if you look at TTN Mapper we have 2 gateways here, one is mine but is indoor.
Lets try something yes.
I found this : LoRa Location Tracker Part 1: Introduction
and this: LoRa Location Tracker Part 2: RSSI
It looks good , the goal is to find the correct position and not only measure the signal
There is an API for gateway RSSI only they provide.
The TOA is built in to appropriately deluxe gateways. RDF, sextants and inertial navigation worked much better for planes as TOA required timepieces that weren’t disturbed by the vibration of such beasts as the Avro Vulcan.
A few tip I follow when looking for locations to place gateways depended on how wide coverage required and terrain.
- TV and radio transmitters, they have mainly Sub GHz transmitters and were placed to get max coverage area (they also have filler transmitters)
- Emergency services use VHF and UHF radios, look where there transmitters are. They also have done terrain studies already, trying to get max coverage. (they also have filler transmitters)
- ISP providers, they use 2.4 - 5.4 Ghz radio to provide service, this band is nearly only LOS. But with some of the newer tech they can use NLOS.
- Cellular providers, the first sites in the area they will look for max coverage, then they will add filler sites.
- Then the last point, is internet available at location, power available, will I get permission.
So look at what other services have covered the area, they most probably have done all the hard work already for you.
2 Likes
@Johan_Scheepers great!
Didn’t knew about LoRa Cloud. TTN has an integration with that, is that correct?
Devices ar from Semtech i believe. have to study that… uau what a goal.
Do you know any example of application running with that tech? An youtube video?
Thanks
You can use any, I used a Temperature sensor with LoRa® TOA/RSSI. The trick is that you need need to be heard by min of 3 gateways and the max rssi needs to be >-85.
You essentially can use a node sending just one byte that is received by 3 or more gateways and then the solver will give you your location. (see the API documentation)
The LR1110-GNSS I have not used yet, you need a node that sends the GNSS information to the solver, read up on the LR1110 chip. I have heard that >1.7Sec you need the node on to get the satellite info and send the packet to the solver. So very fast and this relates to low battery consumption.
The LR1110-GNSS I have not used yet, you need a node that sends the GNSS information to the solver, read up on the LR1110 chip. I have heard that >1.7Sec you need the node on to get the satellite info and send the packet to the solver. So very fast and this relates to low battery consumption.
I think i’ll try this one yes. Just contact the developer team of Lora Cloud (Semtech). Wait for the meeting next week.
Will let you know n the results.
If you know the direction of the gateway you may replace the device antenna with a yagi multielement to increase the signal strengh.
And you would then probably need to reduce the transmit power of the node to keep it legal.
So you have gained zero really. Less than zero actually since the node now can now only really connect in one direction.