Rf trilateration solution advice

[jtu]

Hi,

I'm looking into trilateration/triangulation project. There is a need for marking and then locating points within <700m x <300m area with ~0.3m accuracy. Area is not allways a rectangle, but mostly. Outside, mostly flat, meaning direct visibility between transceivers.

I'm electronics hobyist myself. I can theoretically imagine how to calculate the solution, but I'm no electronics engineer. Going into theory feels like a big effort. My first impulse is to trow together a prototype combining some cheap 433Mhz (because of free band) rf chips with RSSI function and Arduino for calculations.

For example, I found this: https://www.aliexpress.com/item/32922127422.html?spm=a2g0o.productlist.0.0.6e307b0faUSURO&algo_pvid=7d2c2a82-64e6-4cb1-a076-ac810008b621&algo_expid=7d2c2a82-64e6-4cb1-a076-ac810008b621-10&btsid=0ab6d69515962199877266383e6e0d&ws_ab_test=searchweb0_0,searchweb201602_,searchweb201603_

Anybody with with a professional EE intuition - should I try it or I'm on the wrong path?

Veiksmi,
Jānis

[Benta]

Not really viable IMHO. Signal strength is a very bad indicator of distance.
Why don't you just use a DGPS module? Dirt cheap these days.

[jtu]

I had no clue something like that existed 😀 I have an impression that with pure sattelite signal / without some local positioning it is impossible to achieve accuracy level I need...

Do you mean something like this? https://www.aliexpress.com/item/10000085614992.html?spm=a2g0o.productlist.0.0.11065113SRKvh4&algo_pvid=bcc18d78-34ba-463f-8945-2b6684b876f8&algo_expid=bcc18d78-34ba-463f-8945-2b6684b876f8-13&btsid=0ab50f0815962219229676316ebce8&ws_ab_test=searchweb0_0,searchweb201602_,searchweb201603_

[David Hess]

Triangulation has limited accuracy even with calibration because even small reflections cause phase errors.  This is how Doppler radio direction finding systems work.

Time of flight systems, like GPS, do better because reflections can be discriminated against using time.

I had no clue something like that existed I have an impression that with pure sattelite signal / without some local positioning it is impossible to achieve accuracy level I need...

Differential GPS can be easily accurate enough even with short term measurements.  It requires a close by reference receiver at a defined location and all receivers use the same satellites so the errors cancel out for relative measurements to the reference receiver.

[jtu]

Thanks for educating me!

Short term means what in this context? My marked points needs to be found for up to 5 years with accuracy mentioned.

Can the reference receiver also be DYI project or something industrial is ment?

[David Hess]

Short term means what in this context? My marked points needs to be found for up to 5 years with accuracy mentioned.

It means the reference receiver has to be tied to an exact location, so 5 years later when you take new measurements, the reference receiver needs to be placed in exactly the same spot.

Can the reference receiver also be DYI project or something industrial is ment?

I think you can get pretty good results with multiple identical receivers as long as you can guarantee that they are using the same satellites.  Real differential GPS systems have the reference receiver transmit a set of corrections for each satellite to the other receivers, and the reference receiver can be 10s of miles away.

Turnkey DGPS systems are available but with some programming, I think you can do it yourself like I described over a short distance.  Offhand I do not know if 0.3 meter accuracy is reasonable but I think it would be very close.

[Benta]

DGPS accuracy can easily go to single-digit cm, but normally not on home-brew equipment.
If you want to go there, you'll probably need a subscription to a DGPS service that will provide the correction information in real time. This is country specific, so you'll need to check which service providers are in your area. There are often free services available.
Note that it doesn't have to be GPS; Gallileo or Glonass could also be options.

[robca]

Look also at RTK GPS. There are quite a few available modules that are affordable even for hobbyists https://learn.sparkfun.com/tutorials/what-is-gps-rtk/all https://store-drotek.com/302-rtk (not as cheap as an uBlox 8, but still in the affordable category). You would need 2 boards, one acting as base station, one as mobile unit. cm-level accuracy

I built and tested a multilateration solution using 5 microphones to detect a sound source (and using TDOA and hyperbolic solution), and worked incredibly well in simulations (i.e. when I was generating the timing differences using simple math and using the calculated values in the algorithm), horribly in real life due to multipath reception and issues in cross correlating audio signals. I doubt you could get the precision you need even using radio signals

[awallin]

there's a summary of different approaches at the end of this doc: https://www.decawave.com/wp-content/uploads/2018/12/Ultra-Wideband-Wireless-Positioning-Systems_2014-03-27.pdf

[hagster]

As others have said. RTK of DGPS is a great solution. However you need to have at least one known reference point to place your reference station. If you want the lat lon to be globally accurate then this could be expensive, but I guess you only need it to be locally accurate. In which case you will need a fixed marker that will last at least 5 years. Many systems let you do a long term average to get the best accuracy for a single point. GPS tends to wander around as the satellites move overhead, but the long term average position is pretty good.

Also you can improve accuracy with a choke ring antenna. Basically a corrugated metal ring around a patch antenna. If you can find some dimensions online, you could easily 3d print something and spray it with a conductive paint(mg chem supershield). These choke rings block low angle reflections from the ground and other structures.

An alternative is to use a UWB time of flight system. Decawave make some low cost modules that will do it. These are cm level accurate.

Using RSSI won't work at all. Nowhere near accurate enough.

[beampower]

CRFS has a solution to geolocation using power on arrival (POA).
https://www.crfs.com/geolocation/
https://www.crfs.com/videos/software-poa-geolocation/
Probably way out of reach of general hobbyists and students.
Perhaps someone has published some code on GitHub?
Here's one example:
https://github.com/morriswmz/doa-tools