Position triangulation accuracy

[EngIP 31]

Does anyone know what level accuracy can be achieved by using triangulation to determine an indoor position using 4 transmit stations? I'm toying with the idea of creating a system which can track the location of an RF device (868MHz) indoors, but as far as I can gather the best accuracy that can be achieved using RSSI triangulation for this is 2m, which might not be good enough for my needs. I am not very clued up on RF or triangulation, so any help here would be appreciated.

 

If I could place a fixed position transmitter in every room, would this be suitable for location (maybe used in conjunction with RSSI triangulation)? E.g. Triangulation calculates the location to be on the kitchen/living room border, the RSSI is strongest from the kitchen transmitter, decision is made that object is in kitchen.

 

Failing that, does anyone have any other suggestions as to how I'd achieve this? The end goal is a method of locating a person to a specific room (in a house), and thus control various environment settings (heating, lighting etc.). Would another technology be better for this, or maybe a different RF system (Bluetooth?)?

 

I'd love to hear thoughts on the feasability/techniques possible to achieve this, so please feel free to throw your ideas in, even if it's just to eliminate a possible method.

 

TIA

 

Guy

[cde 334]

As a not so batshit insane option, rfid readers that can scan an entire doorway. If they are all networked, it's easy to determine which room was entered and which was left. Bonus, no mucking around with wireless horrors of the rf kind.

[bluehash 1,581]

Edit: His paper is here - DIVERSITY TECHNIQUES FOR SIGNAL-STRENGTH BASED INDOOR LOCATION DETERMINATION

 

I would have suggested radio RSSI(received signal strength indicator) to detect location. There was a guy in my school doing his thesis on it.

 

I cannot find his paper, but here is a similar one.

[EngIP 31]

cde - Good suggestion, I did consider RFID but discounted it due to it not being deterministic(?), though if I could develop a system where an RFID tag would only generate a positive when passed through the doorway (and not near to the reader) I could have a fairly robust system. I'll have to dig deeper into whether it's possible to have a specific "2d plane" reading area (i.e. specific to a doorway).

 

BH - Thanks, interesting papers posted there! I was thinking that RSSI in conjunction with another system could give greater accuracy (than just by using RSSI alone), those papers posted are a good starting point for me to get some background on the subject. The second paper esp. has some interesting detail.

 

I think my intended system will have many (12+) reference nodes, so it might be a case of "suck it and see", and improve/refine as I go.

 

As an alternative I was considering image processing, but I've a feeling that could get uglier than RF telemetry in a short space of time!

[kenemon 29]

Sorry to interject, but can PIR detectors be used in this way by scanning at different sensitivities? Ie. by cycling the gain low to high and then computation from several locations could triangulation be achieved?