About ibeacon indoor SDK accuracy

Indoor Location
1

Hi,

I am trying to use beacons to set a squared region at outdoor open space to get respective position (up-to 1 meter accuracy) once I enter that region.
I didn’t re-implement trilateration because I believed that indoor location SDK already took care of it.
And the reason that I am using beacons instead of location data is that I need to get data that has lower horizontal accuracy than 5-8 meters.

I have tried using 8 beacons with various set-ups: 20x20 meters, 20x15 meter, 15x15 meters, 40x40 meters, 10x10 meters. All the beacons had intervals around 190-200 ms and power with 4dBm.
For 8 beacons: I set 4 of them at the corners and the other 4 in the middle of each wall. (used indoor location builder)

It seems like 8 beacons with 20x20 meters worked the best (I don’t know why 10x10 didn’t work better than 20x20 though), but I still want to reduce noise and inaccurate data.
For example, it lost track once I started running or walking fast within the region.
Also, sometimes it needed 3-10 seconds until I get the position.

I am wondering how I can get better accuracy: should I add more beacons or reduce the interval, or is there anything that I am missing?

I look forward to hearing from you.

Best regards,

2

Where do you put the beacons, on the ground? Note that at this time, the whole system is calibrated around indoor spaces, with solid walls behind the beacons, and beacons put at around a chest-height.

Decreasing adv interval will yield more data points, so it should improve accuracy—just watch for the battery life. Adding more beacons should also help.

Putting a beacon in the corner will cut it’s radius of operation to a quarter, so it’s usually better to aim for the middle of a perimeter. Consider this example:

image
90f9f89aff062f52de701c655e90a987198be224.jpg2380×2448 788 KB

Not sure why 20x20 worked better than 10x10. What’s the procedure you’re using to measure accuracy? How many samples? Could it be just within a statistical error?

Currently, the algorithms are optimized around normal walking speed, as this is the most common scenario for typical usages of indoor positioning.

3

Dear Piotr Krawiec,

Thank you so much for your quick response and the picture! : )
Yes, I put the beacons on the lawn when I was testing.
For accuracy, I basically had an location position view to check in real-time and also stored the position coordinate to visualize it later on my own.

Thank you so much!

Best regards,

1 Like