RSSI smoothing methods

RE: RSSI smoothing methods

Øyvind Karlsen — Tue, 09 Aug 2016 11:01:03 GMT

To be honest, I am not sure. I'm guessing that some form of averaging will net you better results, but you will have to try this out. Due to the nature of the 2.4 GHz radio signals there will be a lot of fluctuations.

RE: RSSI smoothing methods

arsalanjun — Tue, 09 Aug 2016 10:54:30 GMT

PWN is a good idea Øyvind but the RSSI varies a lot. How do you smooth the values so that you do not get a lot of variations?

RE: RSSI smoothing methods

Øyvind Karlsen — Tue, 09 Aug 2016 05:56:41 GMT

If visuals are important you could PWM a LED when you enter the RSSI range that you want to trigger at, let the LED get more and more lit as you get lower and lower RSSI relative to your decision level. Otherwise just turn on a led when you have 10 samples under a threshold or something.

RE: RSSI smoothing methods

arsalanjun — Mon, 08 Aug 2016 10:57:34 GMT

Thanks for your answer Jan. I a aware that RSSI and BLE are not meant to be used to measure distance but I just want to use it as a proximity purpose. Just like the advertisement of BLE proximity (here), I want to use it to turn on an LED after a specific distance.

How to implement this? Turn on a buzzer/led when link is lost? or turn it on at a specific RSSI dbm?

RE: RSSI smoothing methods

endnode — Sun, 07 Aug 2016 19:18:20 GMT

Hello arsalan,

The suggestion is: don't use BT LE RSSI for anything related to distance measurement because results are very uncertain.

If you go deeper into how 2.4GHz radio works you will see that signal can be reflected and shielded by pretty much everything. The topology can be linked to RSSI only in very isolated (laboratory) conditions, otherwise you can maybe talk about some close/medium/distant zones and accuracy (relating RSSI to such "zones") of less then 100%. If you see all these "BT RSSI vs. distance" or "Wifi RSSI vs. distance" (which is very similar physics) solutions they usually come to only one or two thresholds which is making zone boarders and then they more or less say where the thing is. If you really want something more accurate then you need to work with multiple broadcasters and receivers model, known topology and calibration. Nothing for low cost solutions such as "beacon and mobile phone".

Btw. there are several posts on this topic (search for RSSI) on this forum and conclusions are basically the same.

Cheers Jan