Discussion:
[core] RSSI in SenML
Cullen Jennings
2017-03-22 18:47:12 UTC
Permalink
I got a request to add RSSI to the table in 12.1 of of SenML

Symbol: RSSI

Description: Received signal strength indication

Type: float


Any comments on this ?
Peter Saint-Andre - Filament
2017-03-22 18:50:12 UTC
Permalink
Post by Cullen Jennings
I got a request to add RSSI to the table in 12.1 of of SenML
Symbol: RSSI
Description: Received signal strength indication
Type: float
Any comments on this ?
+1, definitely of interest.

Peter
Carsten Bormann
2017-03-22 18:52:14 UTC
Permalink
Post by Cullen Jennings
I got a request to add RSSI to the table in 12.1 of of SenML
Symbol: RSSI
Description: Received signal strength indication
Type: float
RSSI is a quantity (an ill defined one).

Table 12.1 is about units.

Is there a unit that RSSI tends to use?

(For certain Cisco devices, that would be “percent”.
Other devices try to express their signal strength indications in dBm (typically a negative value approximately from -30 to -120).
But I don’t think there is any industry-wide way to do this.)

Grüße, Carsten
Peter Saint-Andre - Filament
2017-03-22 18:54:14 UTC
Permalink
Post by Carsten Bormann
Post by Cullen Jennings
I got a request to add RSSI to the table in 12.1 of of SenML
Symbol: RSSI
Description: Received signal strength indication
Type: float
RSSI is a quantity (an ill defined one).
Table 12.1 is about units.
Is there a unit that RSSI tends to use?
(For certain Cisco devices, that would be “percent”. Other devices
try to express their signal strength indications in dBm (typically a
negative value approximately from -30 to -120).
dBM is what I've seen.
Post by Carsten Bormann
But I don’t think
there is any industry-wide way to do this.)
If you know what device you're getting it from, does it matter that we
don't have an industry standard?

Peter
Carsten Bormann
2017-03-22 18:59:28 UTC
Permalink
Post by Peter Saint-Andre - Filament
dBM is what I've seen.
Post by Carsten Bormann
But I don’t think
there is any industry-wide way to do this.)
If you know what device you're getting it from, does it matter that we
don't have an industry standard?
As long as you know that you don’t know...
Maybe having a unit name for “device specific RSSI units” is fine.

Maybe adding dBm ("dB relative to 1 mW") would also not be a bad idea.
(We already have dB, but that is a measure of power ratio, not of power.)

Grüße, Carsten
Peter Saint-Andre - Filament
2017-03-22 20:05:44 UTC
Permalink
Post by Carsten Bormann
Post by Peter Saint-Andre - Filament
dBM is what I've seen.
Post by Carsten Bormann
But I don’t think
there is any industry-wide way to do this.)
If you know what device you're getting it from, does it matter that we
don't have an industry standard?
As long as you know that you don’t know...
Exactly.
Post by Carsten Bormann
Maybe having a unit name for “device specific RSSI units” is fine.
Maybe adding dBm ("dB relative to 1 mW") would also not be a bad idea.
WFM.

Peter
Cullen Jennings
2017-03-22 19:57:10 UTC
Permalink
Post by Carsten Bormann
Post by Cullen Jennings
I got a request to add RSSI to the table in 12.1 of of SenML
Symbol: RSSI
Description: Received signal strength indication
Type: float
RSSI is a quantity (an ill defined one).
Table 12.1 is about units.
Is there a unit that RSSI tends to use?
(For certain Cisco devices, that would be “percent”.
Oh, despite what wikipedia claims, Cisco has different systems that implement pretty much every variant of this you can imagine :-) When you get down to it RSSI is really more or less defined by device and software that is loaded on it. I've seen, uh, grade inflation on RSSI on some devices because a competitors device reported better RSSI in same conditions. But it is still one of the primary metrics to understand what is going on with radios.
Post by Carsten Bormann
Other devices try to express their signal strength indications in dBm (typically a negative value approximately from -30 to -120).
But I don’t think there is any industry-wide way to do this.)
Grüße, Carsten
Christian Amsüss
2017-03-22 20:30:49 UTC
Permalink
Post by Carsten Bormann
RSSI is a quantity (an ill defined one).
Table 12.1 is about units.
By the name, but not by the use. The review guide after the table (esp.
point 2) indicate that what is called unit in the document is what SI
calls quantities, and the presence of Hz, 1/s and Bq shows that as well.

Maybe that should be clarified in the registry introduction; straw man
text: "The definitions given here are called units based on common
language use. In the terminology of the International System of Units,
they are special unit names, and not only indicate the unit but also the
described quantity."

Best regards
Christian
--
To use raw power is to make yourself infinitely vulnerable to greater powers.
-- Bene Gesserit axiom
Carsten Bormann
2017-03-22 21:54:58 UTC
Permalink
Post by Christian Amsüss
Post by Carsten Bormann
RSSI is a quantity (an ill defined one).
Table 12.1 is about units.
By the name, but not by the use. The review guide after the table (esp.
point 2) indicate that what is called unit in the document is what SI
calls quantities, and the presence of Hz, 1/s and Bq shows that as well.
Hz, 1/s, Bq are units.

Quantities that can be expressed in these units are, for instance,

frequency, rotational frequency, activity,

respectively.

The principles are in ISO80000-1. (Most of us got our degrees before 2009; I just wish students today would work more with these fundamentals…)
Look up the units and quantities given above in ISO80000-3 for Hz and 1/s (3-15.1 and 3-15.2 in the quantities table, 3-15.a and 3-15.b in the units table there).
I’m too lazy to look up Bq in ISO80000-10, oh well, it’s 10-29.a in the units table there.
Post by Christian Amsüss
Maybe that should be clarified in the registry introduction; straw man
text: "The definitions given here are called units based on common
language use. In the terminology of the International System of Units,
they are special unit names, and not only indicate the unit but also the
described quantity.”
That would most definitely not be true with most entries of the current table.
Almost all are units in the SI or ISQ.
(Yes, there are some entries that are more on the quantity side of the spectrum.)

Grüße, Carsten
Christian Amsüss
2017-03-24 14:44:53 UTC
Permalink
Post by Carsten Bormann
Post by Christian Amsüss
By the name, but not by the use. The review guide after the table (esp.
point 2) indicate that what is called unit in the document is what SI
calls quantities, and the presence of Hz, 1/s and Bq shows that as well.
Hz, 1/s, Bq are units.
Quantities that can be expressed in these units are [...]
Yes, they are units (I was inexact here where I shouldn't have been),
but by they are also (in SI terminology) units that have a name, and
with that indicate a (ISO) kind of quantity, sometimes even given
explicitly in the table. It seems to me that ISO calls "unit" what SI
calls "unit with special names", in that ISO unit definition contains
the kind of quantity, but I might be interpreting too much into that the
SI brochure 8 sec. 2.2.2.
Post by Carsten Bormann
The principles are in ISO80000-1. (Most of us got our degrees before
2009; I just wish students today would work more with these
fundamentals
)
Thanks for the pointer; this is much more explicit than the SI brochure
that was my main reference on this topic so far.
Post by Carsten Bormann
Post by Christian Amsüss
Maybe that should be clarified in the registry introduction; straw man
text: "The definitions given here are called units based on common
language use. In the terminology of the International System of Units,
they are special unit names, and not only indicate the unit but also the
described quantity.”
That would most definitely not be true with most entries of the current table.
Almost all are units in the SI or ISQ.
(Yes, there are some entries that are more on the quantity side of the spectrum.)
I still think that the text should be more explicit in that a SenML
registered unit implies a kind of quantity (probably all of them, not
only the second half; maybe make it into a dedicated column?), but can't
really come up with something certainly-correct-yet-concise, and with
the ISO definitions in background, that's less important given their
definition implies a kind for every unit.

Best regards
Christian
--
To use raw power is to make yourself infinitely vulnerable to greater powers.
-- Bene Gesserit axiom
Christian Amsüss
2017-03-22 20:15:47 UTC
Permalink
Post by Cullen Jennings
I got a request to add RSSI to the table in 12.1 of of SenML
Symbol: RSSI
Description: Received signal strength indication
Type: float
I've been using signal strengths for some time in SenML, and they were
in those applications provided in dBm, which I converted according to
what the spec says (or rather said back then, this has been operational
since 2014) about falling back to UCUM as UCUM:B[W].

The recommendations w/rt UCUM have been demoted since then, but it is
still common in the other units there to be specific and comparable (dB
being the exception). RSSI with any real unit would not have any meaning
outside of that device's receptor, and could just as well be represented
without any unit. I'd support adding something that has the semantic
meaning of received signal strength (dBm would be more common than the
by-the-letters-of-UCUM B[W] unit), but a plain RSSI would IMO lead to
confusion in any inhomogenous setup.

Best regards
Christian
--
To use raw power is to make yourself infinitely vulnerable to greater powers.
-- Bene Gesserit axiom
Eric Ptak
2017-03-22 20:29:59 UTC
Permalink
Hi,
Giving a meaning (good, medium, bad reception) to a RSSI value
really depends on the radio frequency and modulation used, say the protocol.
Zigbee, Bluetooth, WiFi, GSM, LoRa ... all have different range where the
signal is good or bad.
It also depends on the internal hardware design. dBm looks to be the
standard unit for a RSSI value.

Up to the application layer to provide or not a meaning that depends on the
device and the use case.
Finally Signal quality also depends on the SNR, Signal-to-Noise ratio.

Best,
Eric.

On Wed, Mar 22, 2017 at 9:15 PM, Christian AmsÃŒss <
Post by Christian Amsüss
Post by Cullen Jennings
I got a request to add RSSI to the table in 12.1 of of SenML
Symbol: RSSI
Description: Received signal strength indication
Type: float
I've been using signal strengths for some time in SenML, and they were
in those applications provided in dBm, which I converted according to
what the spec says (or rather said back then, this has been operational
since 2014) about falling back to UCUM as UCUM:B[W].
The recommendations w/rt UCUM have been demoted since then, but it is
still common in the other units there to be specific and comparable (dB
being the exception). RSSI with any real unit would not have any meaning
outside of that device's receptor, and could just as well be represented
without any unit. I'd support adding something that has the semantic
meaning of received signal strength (dBm would be more common than the
by-the-letters-of-UCUM B[W] unit), but a plain RSSI would IMO lead to
confusion in any inhomogenous setup.
Best regards
Christian
--
To use raw power is to make yourself infinitely vulnerable to greater powers.
-- Bene Gesserit axiom
_______________________________________________
core mailing list
https://www.ietf.org/mailman/listinfo/core
Loading...