wifi indoors positioning

I’ve always wondered how wifi positioning worked. A hospital in London was able to incorporate this technique in doors:

http://networks.silicon.com/mobile/0,39 … 876,00.htm

I’ve researched a lot on triangulation in general. And the only viable method is through signal strength detection and mapping the signal strength from an antenna for each room on a floor.

With WiFi, routers are the antennas. Somehow, the routers are to report back to a central computer the signal strength of all the devices they detect. This can only be possible with an alteration of router firmware, which makes WiFi too much of a hassle. However, this is the only solution I’ve come up with, and I’m pretty sure that this isn’t what wifi positioning companies do. Can some describe in detail how this works? I also want to try at wifi positioning. Since theoretically, it requires no additional hardware and only programming. Some companies have specially made wifi tags, which no doubt simplifies the process of positioning. But is it possible to do without and simply track a laptop with wifi?

for position tracking using WiFi infrastructure, there are two main issues:

1- the access points must expose the received signal strength to the position calculation engine and this requires a political relationship between companies. Ekahau is one that does well at this, with Cisco and others.

  1. This signal strength technique relies on an accurate survey and highly overlapping coverage footprints. To get to, say, 5 or 10m of accuracy on x and y, the coverage footprints of 4 or more APs must overlap. Normally, you don’t have that for a typical WiFi system. More overlap = more interference.

I call the above “RF Footprinting”. Note that if major changes in walls/furniture occur after the survey, the accuracy suffers.

The other technique is time-difference-of-arrival. Three receivers, overlapping coverage. Special receivers.

All in all, indoor position tracking is high cost.

Better, are simple electronic mileposts. cheap.

Ping times can be measured in nanoseconds…

stevech, for the first problem, I was considering altering the router firmware. But that seems more problem like its worth. And you’re right on the overlapping. The reason why I was biased towards wifi tracking was because most places that would need it would already have a router network setup. But because you don’t know exactly how many access points they have and how much they overlap, it poses as a problem on accuracy.

Perhaps if I could get accurate ping times, this would be a problem. But as you said, that requires additional expensive equipments.

I’m straddling between bluetooth and wifi tracking right now. And now it seems like bluetooth is better. Since I wont have to deal with some of the problems of wifi and I can establish a relationship easily with a computer.

Would you guys say bluetooth is better suited here?

A way to get RSSI without modifying the WiFi access point is for the tracked device to send a UDP datagram to a particular IP address, containing RSSI of the access point’s transmitter. This is done in some products - esp. when the WiFi access points are of random vendors and non-cooperative (as is the norm).

not sure what/why you are tracking… but one neat new scheme is as follows.

put 802.15.4 modules (cheap) all around - exit signage, emergency battery lights in hallways, ceiling lights, etc. Spaced say, 50 ft. with 1mW modules and a PCB antenna. These are electronic signposts. Set these up as a mesh using ZigBee or some home-brew fixed routing (since these don’t move). You can get RSSI easily from each radio. And do simple estimates of “closest” signpost. Or get brave and do the RF footprinting estimate based on a survey and RSSI correlation.

You can do this with one-way transmitting beacons and receivers all around that use telephone wire pair spares. But installation labor is high.

In a few kinds of buildings, certain floors/areas today’s GPS receivers get a fix.

that was one of the ideas that I had to deal with indoor tracking. However, the method is only accurate to the point of locating the room a device is in. I was originally thinking of using Bluetooth instead of 802.15.4 modules for this purpose. And instead of using a mesh with ZigBee, I was thinking of just going with my paired Bluetooth antennas(which all are able to pair with a central computer on my floor). Paired, these Bluetooth antennas would be able to report back the RSSI. And then its all the same from there.

Wouldn’t you say Bluetooth would be simpler here? Plus, these antennas can be spaced farther and are able to track bluetooth supported devices by default(at least their RSSI).

The idea in using a ZigBee mesh or 802.15.4 modules with hard-coded neighbor-addresses and forwarding software you code (simpler than ZigBee), is that only one such fixed node needs to connect to the root PC/server.

Don’t know how you’d do the same with Bluetooth. And Bluetooth radios are far more expensive and undocumented.