Equating RSSI vals to good/ok/bad wifi?


#1

For our customer manual we’d like to be able to tell them how good their wifi <-> photon connection is, and if it needs work.

Can anyone categorise RSSI values into Excellent / good / ok / poor / hopeless ? I can’t find anything in the docs.


#2

I remember a thread where a similar question was raised and the code snippet from the firmware was quoted that does a 5 level mapping.
You could use the search feature :mag: of the forum to find that thread :wink:


#3

I’ve already searched and seen a thread with code in it, but the values chosen seemed to have been plucked out of the air as reasonable guesses. Does anyone have hard data? I’d have thought Particle’s hardware devs at least would have a good handle on this.


#4

Google is your friend :slight_smile:

I found this in like 10 seconds:


#5

No, a decent datasheet is my friend. I’ve hunted down the BCM43362 datasheet to find the receiver sensitivity specs. They take a bit of interpreting, but in essence I’d say your table is out by about 10dBm - where did you get that from anyway? Does it relate to Particle’s wifi or is it just a generalised hand wavy guideline?

Datasheet summarized:
At 10% packet error rate the sensitivity limits are roughly : slow speed -90dBm; moderate speed -84dBm ; fast speed -75dBm.
So I’d put “poor” as less than -90dBm, “fair” as -75dBm to -90dBm and “good” as -65dBm to -75dBm, with anything above -65dBm as “bloody marvellous”

I hope someone else finds actual numbers useful. Its the sort of thing that would be nice to see in the Photon datasheet under “wifi specifications”. Tx power is only half the story.

From the actual datasheet:
RX sensitivity
(8% PER for 1024 octet PSDU) at
WLAN RF porta
1 Mbps DSSS –95 –97 – dBm
2 Mbps DSSS –92.5 –94.5 – dBm
5.5 Mbps DSSS –90 –92 – dBm
11 Mbps DSSS –87 –89 – dBm
RX sensitivity
(10% PER for 1000 octet PSDU) at
WLAN RF porta
6 Mbps OFDM –88 –90 – dBm
9 Mbps OFDM –88 –90 – dBm
12 Mbps OFDM –86 –88 – dBm
18 Mbps OFDM –84 –86 – dBm
24 Mbps OFDM –82 –84 – dBm
36 Mbps OFDM –79 –81 – dBm
48 Mbps OFDM –75 –77 – dBm
54 Mbps OFDM –73 –75 – dBm
RX sensitivity
(10% PER for 4096 octet PSDU) at
WLAN RF porta. Defined for default
parameters: GF, 800 ns GI, and
non-STBC.
20 MHz channel spacing for all MCS rates
MCS7 –70 –72 – dBm
MCS6 –72.5 –74.5 – dBm
MCS5 –74.5 –76.5 – dBm
MCS4 –78.5 –80.5 – dBm
MCS3 –82 –84 – dBm
MCS2 –84.5 –86.5 – dBm
MCS1 –86.5 –88.5 – dBm
MCS0 –88 –90 – dBm


#6

@ScruffR Here is the post you were referring to earlier about mapping RSSI to a 1-5 reading.


#7

Good to know you have found your hard facts here, especially since the verbal terms chosen for the categories are just as well defined :wink:


#8

Interesting. After doing some tests with this myself, I find that I usually get between -90dBm and -100dBm
(Electron on my desk, clear of anything that could alter the signal)…I live in a large city though and don’t have too many issues with wonky connections - there’s actually a Rogers cell tower about 500 meters away

Wondering if the signal here really is that subpar, or if I should be taking my results with a grain of salt


#9

I usually get 2 to 3 flashes on the Electron when I pushed the mode button to get the cellular signal strength in my area which is suburban with no tall buildings around but with lots of cellular towers. I expected to see a stronger signal.

My Wifi with an external antenna on a Photon will show -60 with my phone hotspot 20 feet away, and -10 to -14 when the phone is right next to the antenna.


#10

I used to apply the 1 to 5 scale to map the RSSI - the results typically were all over the place - giving results as high as 4-5 for devices inside metal containers that had a lot of connectivity issues.

I have only looked at the raw dBm-values today (with only one device) but either the values are off - or low values (e.g. < -95) aren’t as bad as we make them seem to be, as it usually connects in 10-30 seconds.

I’ll do some more testing later (including outside, and on a metal container) and see how this all affects the values.