Geolocation with static node helpers?

1

Hi All, I reckon the brains trust can help out here:

Location for lorawan node (call it “Node A”) is derived from the timestamps from gateways that receive a message from “Node A”.

What would happen if you had multiple static nodes (fixed position) in the area that “Node A” is also transmitting? Would it be any advantage computing the static nodes timestamps/rssi to derive a more accurate location for “Node A”?

Cheers
Adam.

2

what is the purpose of these ‘static’ nodes ? I don’t see it :thinking:

3

Yep its hard to explain.

So I think people will own static nodes dotted all over the place in the future/now … eg a temp sensor … door sensor … motion sensor … none of these move location.

if “Node A” was to be in the area of all these other static sensors is it possible to use the data gleamed from the static sensors to give a more accurate location for the moving “Node A”.

4

Triangulation by using signal strength will give a poor indication of location.

All of us will have seen significant variations in signal strength particularly with nodes that are moving or changing orientation. In very approximate terms a variation in signal strength of only 3dB represents a distance change of 50%.

So move a node around a bit and one second you think its 1km from the gateway and the next it is 1.5km away.

1 Like
5

Yep signal strength sucks for location.

But what about timestamps from the static nodes?

6

you know that the gateways accuracy is key (you’ve asked this before)
-Geolocation Gateways?

if you don’t place all these gateways yourself, and have your own backend… you can never do it imho

7

I dont know, what is the resolution and accuracy of the timestamps ?

Radio waves take 3.3uS to travel 1km.

8

Sorry, I thought this was a different question.

9

So anyway… it’s Ok if you don’t know the answer … I’m pretty sure that’s what science is all about, talking through a problem.

I still think there is merit to using static nodes data to increase a moving node (Node A) location accuracy.

Cheers
Adam.

10

I believe what you are describing is technology like http://www.locata.com/

It is essentially ground based GPS nodes (those are your “fixed nodes”)

11

Or it could make it worse, if the given positions of these static nodes was stated incorrectly.

12

Sure it could.

But I think it would increase the location accuracy of the Node A as you can draw from a larger dataset of the static nodes, rather than just the timestamps from Node A.

when we sample from a sensor we don’t just do it once… we take averages from the dataset right?

13

I guess you will find out when you implement it.

The key bit of information is the resolution\stability of the current timestamping, if its say only down to 1mS, then your trying to average a location that could be +- 300km, which I doubt would be meaningful.

14

Is it down to 1ms?

15

I have no idea, thats why I asked you earlier.

16

@johan, @htdvisser, @wienkegiezeman, @RichL is what we are chatting about a viable thing? Love to hear your input.

Cheers
Adam.

17

While I can understand where you aim, the geographical position of static nodes not owned by you is not something you sample to reduce the error - position will be likely set statically (like already you can do in console), and nothing prevents voluntary mispositioning. The same is valid for gateways of course. On the other side, if you plan to build a strictly controlled infrastructure, at least static nodes position will be right (to which you have to add the time resolution issue…).

18

Thanks @UdLoRa, I think I was more on the path of only sampling data of “static node” that we own for this theoretical case study, there for the location of “our” static nodes would be accurate.

Sorry for all the questions, but I still think there is something in it.

Cheers
Adam.

19

To get a decent resolution of postion you need to timestamp to a sufficient resolution also. To then get accuracy you need to sync the gateway time accuratly, that should be possible using the PPS from the GPS.

Once you have done that, and the packet is seen by at least 3 gateways, you should have an accurate geolocation. If you do that then adding extras nodes will add little extra benefit, I would think.

TTN are thinking about these things, I saw this presentation at TTN Conferance 2018;

Perhaps contact Richard direct ?

20

Hey @RichL get in here!!! We have some sciencing to do!!

  • What is the precision of gateway timestamps, ns/ms/s?
  • Whats is the impact of static node location whilst computing geolocation?
  • Are you coming to TTN Conf Adelaide?

Cheers
Adam.