Regarding nRF51822 RSSI to Distance Calculation

RE: Regarding nRF51822 RSSI to Distance Calculation

Dan Danknick — Mon, 19 May 2014 18:54:43 GMT

It's just a simple infinite impulse response filter using fast integer math.

The equivalent with floating point, which you more typically see, is:

strength = 0.0625f; signal = (1.0f - strength)signal + strengthnew_signal;

Dan

Nazmus Sakib — Sat, 17 May 2014 22:55:36 GMT

Hi Dan.

Thanks for your reply.

I know its quite challenging task plus its value will be approximate.

Could you please explain more about your given formula?

Regards.

Dan Danknick — Sat, 17 May 2014 20:55:36 GMT

I've experimented with this quite a bit. If you want to believe RSSI-to-distance then more/good data is your friend.

More = increased advertising or connected measurements per second Good = throwing away outliers and smoothing what remains

To start, I suggest tossing the bottom and top 10% of readings. And then applying an IIR filter with a ratio like:

signal = (15*signal + new_signal) >> 4;

giving you a good confidence after say, 30 iterations.

RSSI is messy and noisy due to a myriad of physical phenomenae. Welcome to the world of converting noisy sensors into meaningful data!

Dan