Hello Everyone, As i’m using a sensor node to send packets to TTN Gateway i noticed a strange behavior which i don’t understand, according to the mechanism of LaraWan ,many Gateways can receive the same packet but mostly (depending on the RSSI) the nearest will be the considered one.
The dilemma is that the Gateway ID which receives my packet is 63Km away from my position and what makes it strange is that there’re at least 3 gateways nearby (in the same area around me) but only one of them (2.9Km) that receives my packet (sometimes but not always as the 63km Gateway) , my question is does the location of the gateways (which i see on the TTN application) could be fake ? and if so, how i can possibly measure the distance between my sensor and the gateway that received my packets ?
This is the position of the receiving gateway (63 Km away), My sensor location is on the upper point:
Those are (according to TTN map) the gateways that are on the same city:
This is what i get on the TTN application web page :
I noticed that the altitude is 530m , could it be a balloon and is it possible that it receives better than the near by gateways ?
IMHO this altitude is referenced to MSL. There seams to be a hill with the gateway on top. The antenna of the gateway can look into Torino from above. With SF7/125kHz theoretically 135 km are possible.
A tethered balloon would be fairly obvious I would think.
The only way for you to measure the distance is to know for sure where the Gateway is located, since the co-ordinates could be wrong, and measure on a map.
Calculating the free space path loss gives a difference of 5dB (63Km & 3Km) between the two gateways, that’s a lot in terms of the received power …I believe in what you said but with a smaller difference in distance (not 60 Km !!) …unless the buildings really do harsh stuff for the 3Km away gateway…
So it means that if the owner of the gateway didn’t set correctly the position of the gateway there’s no way for the users to know where it’s …!!..so application that’s build for localization based on Lora & TTN must uses its own gateways …
At best with TTN data you might be able to narrow a location down to a 10km square with the conflicting information - with some local tests you, may be able to tune a model for your area, inputing RSSI in to a Machine Learning system to get down to 1km square. [Note to self, another summer project].
If you need to know the location down to 10’s of meters, you will have to transmit GPS co-ordinates or install four or more high end gateways with the Semtech special facility to compute time of flight data.
If you want a rough idea of location in a metropolitan area, the loverly Google collects data on WiFi networks - so your node can listen for WiFi stations, send that via LoRa, you will have a rough idea of where it is due to the gateway and the WiFi Id can apparently be looked up and give you a better location.
Those are very interesting ideas , but can’t i just follow the RSSI then start to circle around and narrowing the circle until i have the best RSSI which will give me roughly an error of 10th of meters ?
I stored some data on a csv file , extracting the packets which were received by 2 gateways ( the < 3Km and the fake location GW) i got those RSSI which shows that the fake location GW is probably more closer than the other one .
GW: Polito < 3 Km
GW: eui-69c5a8fffe7664c3 (fake location GW)
These values are getting close to the calculation/reporting limits for the solutions (search forum for others reported experience on lowest reported (v.s. theoretical) values on LoRa/TTN) and remember signal strength on direct line of sight follows ISL decay so once you are more than a few tens, certainly hundreds of meters distant a small change in Signal strength is same a large change in distance and if not direct LOS then absorbers can have big impact…e.g. a 4mm pane of glass in a building or greenhouse between your node and gw can make a 3km distance look more like 13km! Also LoRa is quite resilient for reflected or multi-path transmission so again a 3km distant node can look like 20km if signal is coming via a reflection. Use RSSI with extreme caution.
Yeah, sure, if you are looking to find the location. And when you’ve narrowed it down to a dozen or so buildings, then what?
Personally, rather than driving around in circles (although I have done that once with a balloon payload that had lost GPS fix and got very very close after one circuit of the town it had landed in), I’d use a Yagi - which is what I got out just as my phone rang from the person that had found it. So a bit of triangulation will save you a fair amount of bother. Bit unorthodox on a device, but point it in four different directions and see what happens - go somewhere else and repeat.
But I was actually referring to the general idea of using LoRaWAN for figuring out a devices location based on your comment.
