Antw.: [vt-dev] Re: reading and setting the transmission rate

[onel...@gmx.net]

Hi Carlos -

the rate control mechanism needs unicast traffic to each of its neighbor stations, in order to measure and subsequently determine the rate for each neighbor.

Of course there is a time out, so in order to keep the information updated this requires frequent traffic to all the individual neighbors.

This is rather impractical, since it would require the spectrum to be spammed with unicast traffic.

If there is no recent unicast traffic, the driver will report 36 megabit.

Cheers,
Elektra



---
I have ceased to talk with the voice in my head. The voice in my head has stopped talking to me. I am one with the universe. The mind dissolved in a little cloud of logic. Serenity.

----- Reply message -----
Von: "Carlos Rey-Moreno" <cr...@ehas.org>
An: "village-telco-dev" <village-...@googlegroups.com>
Betreff: [vt-dev] Re: reading and setting the transmission rate
Datum: Mo., Jul. 23, 2012 11:00


Hi all,

Now it's working I just needed to sleep a bit and think clearly. To do it, just in case somebody else want to do it in the future is modifying the /etc/config/wireless file adding the option rate including the number of the rate and the letter M. In case you want to change also the mode (in our case we want to fix to 11b 1Mbit to get the best sensitivity and optimize the links from there) you need to use the option hwmode. In the last case it would something like:

option 'hwmode' '11b'
option 'rate' '6M'

Once that is done, you need to reload the wireless interfaces with the command "wifi". After that the changes appear both in iwconfig and in wlanconfig ath0 list.

A nice thing to solve in the future is why when in hwmode auto, the changes in rate does not appear in any of the aforementioned commands

Best,

Carlos

On 07/22/2012 08:22 PM, Carlos Rey-Moreno wrote:
> Hi all,
>
> Finally we have decided installing RC3 and so far everything is
> working as expected. However, the issue I mentioned this morning
> regarding changing the rate manually of automatically still does not
> work, or I dont know how to do it properly, which is more likely, :-)
>
> We want to know which rate the MP is using in every moment as for us
> to know how much margin do we have with the received signal. However,
> "wlanconfig ath0 list" is not showing the changes and is all the time
> showing 36M. Similarly iwconfig says Rate 0 kb/s. "iwpriv ath0
> get_rate11g" gives 18 as a result, and if I change "iwpriv ath0
> rate11g" track the chage, although in "wlanconfig ath0 list" doesnt't
> appear the change.  If I try to change "iwpriv ath0 minrate" and
> iwpriv ath0 maxrate" let say to 6mbit both, the change does not
> reflect in any of the aforementioned commands.
>
> I make lal this changes through the commad line since the options to
> change them do not appear any more on the Web Interface.
>
> Does anybody knows how I can obtain the real value of the rate being
> used in every moment? and also how can I change the rate consistently?
>
> Probably we will use the auto mode but we had some issues wiht the
> previous version of the SECN firmware, that resulted in not updating
> the rate even when the RSSI of the link was 2 dB. Furhtermore as I
> said I want to know how much margin do we have in everylink but
> without knowing the rate I can't know the sensitivity and thus the
> real strength of the signal received (the parameter DBM in wlanconfig
> ath0 list does not upgrade consistently and uses always -95, that I
> guess is the one for 1Mbps, no matter what rate you're using).
>
> Thank you very much in advance,
>
> Carlos


--
You received this message because you are subscribed to the Google Groups "village-telco-dev" group.
To post to this group, send email to village-...@googlegroups.com.
To unsubscribe from this group, send email to village-telco-...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[Carlos Rey-Moreno]

Hi Elektra, and thank you very much for your answer. We'll try to use it
qhen continuing with the alignment of the antennas today. It could be
impractical, but up to know we didn't know about other mechanism to
align them but to relay on the wifi beacons and the readings we got in
wlanconfig or with the originator packets used by batmand-adv and shown
in batctl o. So maybe, when we get a link done we use ping to optimize
it (and would explain the improve on RSSI when succeeding on making a
call), although I have the feeling that by optimizing one we are going
to lose signal in others... I'd done many point-to-point long distance
links but I have to admit that these middle-distance
multipoint-to-multipoint ones are turning out more tricky than I
expected. Is there any other method that anyone from the list could
recommend?

However, the ping method needs a link, how can we do a better alignment
when the link is not yet good enough as for allowing ping? Making ping
packets smaller?

There is also one last thing that I don't fully understand, the
asymmetry in the links. Even when there is only a PtP link, the signal
reading in both ends is completely different. Maybe now with doing
carefully with the ping method, things change and the links are more
symmetrical, but up to know is weird that using the same TxPower in both
sides the signal reading is so different, which make me think that they
are using different sensitivities. Although as I couldn't track the rate
used with the auto method I wasn't this point. Is there anything else
that could be causing this issue?

By now we've managed to make calls in links over 1km but the links
around 2km show a weaker RSSI and a maximum of 110 in batctl o, which
doesn't allow us to connect to them.

Thanks a lot for in advance for any comment or recommendation we can get.

Best,

Carlos

[Carlos Rey-Moreno]

Hi Elektra and all,

Today we've been running some tests, and we still to arrive to some
conclusions, although there are some things in the air:
- When working in auto, the rate doesn't change from 36, no matter we
use ping or generate traffic with iperf going through the link.
- The values of the RSSI are very similar even though we fix the rate to
1M, 6M and 36M. We were expecting to see some changes due to the
different sensitivities but they keep in values very similar to each other.
- Those values are very different in both sides of a point to point
link. On the one side we have a MP out of the box (MP1), the other one
is one of the MPs we've built with the external antenna (MP2). We are
kind of sure that the external antennas are working fine because we get
beacons form nodes as far as 4 km. However, the asymmetry is too big.
The MP1 is showing a RSSI of around 30, whereas the other is showing a
RSSI of around 15.
- The txpower shown in iwconfig is 18 dBm although we've entered 20 in
the web interface. This is confirmed by the readings, that only increase
by 1 db (from 17 to 18).
- In all the links, even on those with extremely clear line of sight, we
are having around 12 dBs at least less than what we were expecting from
Radio Mobile. Surely 2 of them are from the txpower thing, but the other
10 I dont know why they are lost. Maybe by finding out the issue with
the symmetry we get closer values here.

Anybody can help with any of the above?

Thanks in advance,

Carlos

[Elektra]

Hi Carlos -

in order to adjust hi-gain antennas for wireless long shots I use a raw radio packet sniffer, in order to find the initial alignment. Many years ago I used Kismet, these days we have Horst in the package repository.

Ping with different packet sizes is also OK, once you have established a link. That is good for optimizing the alignment. And of course the SNR reading. You can do that with wlanconfig or iwspy in combination with iwconfig.

In ad-hoc mode you have to tell the driver which SNR to report via iwconfig. Since in ad-hoc mode you have multiple neighbors you must select the neighbor you want to measure. You do that with iwspy. You ping the node, learn the MAC address and tell the driver via iwspy to report the SNR to this neighbor.

A simple script (a while loop that greps for the signal strength reading) can show you updated information. It is recommended to ping the neighbor you want to measure in the background, so the driver keeps on updating the information in a timely manner.

> There is also one last thing that I don't fully understand, the
> asymmetry in the links. Even when there is only a PtP link, the signal
> reading in both ends is completely different. Maybe now with doing
> carefully with the ping method, things change and the links are more
> symmetrical, but up to know is weird that using the same TxPower in both
> sides the signal reading is so different, which make me think that they
> are using different sensitivities. Although as I couldn't track the rate
> used with the auto method I wasn't this point. Is there anything else
> that could be causing this issue?

Many issues are possible. Are your antennas and cables OK? Fix the alignment for optimum SNR and check the reading. The receiver sensitivity has some tolerance, as always in the real physical world. However, they should not be dramatic between devices of the same make.

The sensitivity of the radio is fixed and maximum sensitivity is measured at lowest bitrate, typically with 6% or 10% packet error rate. The driver will always report the maximum sensitivity. Of course, depending on the modulation and data rate the minimum SNR varies. Higher data rates require better SNR, naturally.

> By now we've managed to make calls in links over 1km but the links
> around 2km show a weaker RSSI and a maximum of 110 in batctl o, which
> doesn't allow us to connect to them.

Doubling the distance will reduce the signal by 6dB. If the margin is just enough to get the smallest packets through, it is a link that needs improvement ;)

Cheers,
Elektra

[Elektra]

Hi Carlos -

the txpower is up to 20dBm at bitrates up to 24Mbit, and goes down at higher bitrates. The driver reports the lowest tx power, which is 18dBm at 54 Mbit. So you are operating at 20dBm - at bitrates < 24Mbit

There is no wonder that the devices are reporting different SNRs if they have different antennas with different gain.

An antenna amplifies in transmit direction *and* in receiving direction.

So lets assume your external antenna has a gain of 15dBd. The internal MP has an antenna with 0dBd gain. The external antenna will boost the sensitivity when receiving packets by 15dB and it will amplify transmit power by 15dB when transmitting.

So when you have the situation that you have described, the difference in RSSI between both sides must be 15 dB - for a 15dB antenna.

What is the actual expected gain of your external antenna?

Receiving beacons from 4 km away is a indication that the external antenna is boosting the signal, but not really proof that it is performing well. Well, that depends on the expected gain. Beacons are short packets and sent at lowest bitrate and highest tx-power.

Note also that the default MP antenna system is not perfectly omnidirectional (two monopoles in parallel on the pcb, so they interfere with each other somewhat). You will likely see a difference in SNR if you rotate it.

Cheers,
Elektra

[Carlos Rey-Moreno]

Hi Elektra,

Thanks for your answer.

> in order to adjust hi-gain antennas for wireless long shots I use a raw radio packet sniffer, in order to find the initial alignment. Many years ago I used Kismet, these days we have Horst in the package repository.

Yesterday we were using Horst, and I like it more than wlanconfig,
specially because you can reset the readings and get the max/min snr for
a given position. It certainly helps. The only problem is not being able
to run it remotely, every time we tried it stopped the link, I guess
because the radio starts scanning and the wireless link through which
you arrive to the station disappears.

>> There is also one last thing that I don't fully understand, the
>> asymmetry in the links. Even when there is only a PtP link, the signal
>> reading in both ends is completely different. Maybe now with doing
>> carefully with the ping method, things change and the links are more
>> symmetrical, but up to know is weird that using the same TxPower in both
>> sides the signal reading is so different, which make me think that they

>> are using different sensitivities. Is there anything else

>> that could be causing this issue?
> Many issues are possible. Are your antennas and cables OK? Fix the alignment for optimum SNR and check the reading. The receiver sensitivity has some tolerance, as always in the real physical world. However, they should not be dramatic between devices of the same make.

The issue with the antennas is better answered in the next email.

>
>
> The sensitivity of the radio is fixed and maximum sensitivity is measured at lowest bitrate, typically with 6% or 10% packet error rate. The driver will always report the maximum sensitivity. Of course, depending on the modulation and data rate the minimum SNR varies. Higher data rates require better SNR, naturally.

The problem was that I was assuming that by wlanconfig reporting a rate
of 36M, the SNR reported was also based on the sensisitivity at that
rate, my mistake. Now with horst, that is a scanning tool, I've realized
that the SNR shown in wlanconfig is the SNR over the maximum sensitivity
of the radio, no matter the rate used. That is the value I wanted to use
for aligning, so another thing less to worry about, :-)

Best,

Carlos

[Carlos Rey-Moreno]

Hi Elektra,

Thanks for your answer again.

the txpower is up to 20dBm at bitrates up to 24Mbit, and goes down at higher bitrates. The driver reports the lowest tx power, which is 18dBm at 54 Mbit. So you are operating at 20dBm - at bitrates < 24Mbit

Ok, so the the TxPower is also understood now. Also corroborated by some other tests we did yesterday.

There is no wonder that the devices are reporting different SNRs if they have different antennas with different gain. 

An antenna amplifies in transmit direction *and* in receiving direction.

So lets assume your external antenna has a gain of 15dBd. The internal MP has an antenna with 0dBd gain. The external antenna will boost the sensitivity when receiving packets by 15dB and it will amplify transmit power by 15dB when transmitting.

So when you have the situation that you have described, the difference in RSSI between both sides must be 15 dB - for a 15dB antenna. 

I do not agree with this statement, although I might be wrong. If the link budget is:
- Received power = Tx Power + Antenna Gain on reception + Antenna Gain on Transmission - Free Space Loss + other losses. 

Received power depends on the maximum Sensitivity of the receiver, which in this case is -95dBm for both Mesh Potatoes the one with antenna and the one without.
- The Antenna Gain on reception is:
     a) 9 dBi when the MP with antenna is receiving
     b) 0 dBi when the MP without antenna is receiving
- The Antenna Gain on Transmission is:
     a) 0 dBi when the MP without antenna is transmitting
     b) 9 dBi when the MP with antenna is transmitting
- Free Space Loss + other losses remains constant no matter who receives and who transmits.
- TxPower is 20 dBm in both sides and so no matter who transmits.

In this case I don't see why it should be different. Checking in Radio Mobile for different systems with different gains support the idea that the SNR of the link should be symmetrical.

In our case the problem was that the MP without antenna was behind a window and so the in reception there was interference created by similar copies of the signal, than when transmitting didn't happen. Once we Putted the MP out with just one clear path in between the two, the link was symmetrical.

What is the actual expected gain of your external antenna? 

The antenna gain was supposed to be 9 dBi, but I I think is here where our problems are... since we substituted some of the ones with antenna for ones without, and in many cases the ones without (MPs out of the box) had similar results, when no better results, than the ones with external antenna. Which means that either:
a) I did the soldering wrong;
b) The waterproofing of the antenna is consuming some of the radiated / received power.
c) The antenna is not as good as they say it is;

For a), I've been rereading the emails we exchanged, and I did everything as suggested and when measuring the resistance "between the center pin of your RF cable attached to the MP and the conductor path going to the bridge" it was very close to 0 Ohm in all cases and was 1 (or infinite) in between the two pads of R404. Maybe the soldering of the two pads of R402 is a bit bigger than a professional would have done, but is connecting them. Could this increase causing this problem?

Another issue could be that when connecting the pigtail is causing the problem you mentioned "Just make sure it doesn't touch the copper paths of the RF section directly (the copper path leading to the UFL connector at the top of the board), because this can change the physical properties of the microstrip RF line". I'm not sure we checked this, :-S. If the braid is insulating the cable, this could still be causing the problems we're facing?

For b), we done the following. Using hot glue to fix the knee, using hot glue to waterproof the connection between the antenna and the pigtail (there was half a centimeter of the rod in the pigtail uncover once the antenna was fully connected to the pigtail).  Putting self amalgamating tape on top of the glue and the connectors sometimes up to 5 to 10 cm of the antenna. Using shrinking tube over the whole antenna to further protected, from bottom to top, and with 2 cm extra folded down, as you recommended. Is it possible that the self amalgamating tape and the plastic of the shrinking tube are blocking the signal?.

For c), we're going to run some tests the weekend for checking with some spare antennas and pigtails we have if we obtain the 9 dBis before waterproofing them. If not, I'm thinking on using the reflector you designed. Do you think it behaves like a 180º sectoral antenna? or is more a directional one on the 45º angle that you mentioned?

Thanks a lot in advance.

Best,

Carlos

[elektra]

Hi Carlos -

> In this case I don't see why it should be different. Checking in Radio
> Mobile for different systems with different gains support the idea that
> the SNR of the link should be symmetrical.

– you are right.

> > What is the actual expected gain of your external antenna?
> The antenna gain was supposed to be 9 dBi, but I I think is here where
> our problems are... since we substituted some of the ones with antenna
> for ones without, and in many cases the ones without (MPs out of the
> box) had similar results, when no better results, than the ones with
> external antenna. Which means that either:
> a) I did the soldering wrong;
> b) The waterproofing of the antenna is consuming some of the radiated /
> received power.
> c) The antenna is not as good as they say it is;

My guess is c).

>
> For a), I've been rereading the emails we exchanged, and I did
> everything as suggested and when measuring the resistance "between the
> center pin of your RF cable attached to the MP and the conductor path
> going to the bridge" it was very close to 0 Ohm in all cases and was 1
> (or infinite) in between the two pads of R404. Maybe the soldering of
> the two pads of R402 is a bit bigger than a professional would have
> done, but is connecting them. Could this increase causing this problem?

No, unlikely.

> Another issue could be that when connecting the pigtail is causing the
> problem you mentioned "Just make sure it doesn't touch the copper paths
> of the RF section directly (the copper path leading to the UFL connector
> at the top of the board), because this can change the physical
> properties of the microstrip RF line". I'm not sure we checked this,
> :-S. If the braid is insulating the cable, this could still be causing
> the problems we're facing?

The braid presents ground so it will interfere with the microstrip properties if it comes too close. I can not quantify the effect, but it seems reasonable to me to keep the cable away from it.

> For b), we done the following. Using hot glue to fix the knee, using hot
> glue to waterproof the connection between the antenna and the pigtail
> (there was half a centimeter of the rod in the pigtail uncover once the
> antenna was fully connected to the pigtail). Putting self amalgamating
> tape on top of the glue and the connectors sometimes up to 5 to 10 cm of
> the antenna. Using shrinking tube over the whole antenna to further
> protected, from bottom to top, and with 2 cm extra folded down, as you
> recommended. Is it possible that the self amalgamating tape and the
> plastic of the shrinking tube are blocking the signal?.

The added weather proofing will change the properties of the antenna somehow, because the dielectric constant of the additional material around it is different than free air. You can make a simple test to figure that out - compare the stock antenna with a weatherproofed one. I suppose it will not make much difference, if any.

> For c), we're going to run some tests the weekend for checking with some
> spare antennas and pigtails we have if we obtain the 9 dBis before
> waterproofing them. If not, I'm thinking on using the reflector you
> designed. Do you think it behaves like a 180º sectoral antenna? or is
> more a directional one on the 45º angle that you mentioned?

The gain of the external reflector is nowhere close to the claimed 9dBi. Rather 3-5dBi. But I'd love to get some feedback on the reflector.

Here is a interesting web page discussing the gain of collinear dipoles. In order to achive 9dbi / 7 dBd, a collinear dipole antenna requires 6 collinear dipoles of at least 0.1λ end-to-end spacing:

http://www.w8ji.com/stacking_broadside_collinear.htm

In order to provide that gain, the elements and the distance between elements alone adds up to a length of at least ~ 41cm for the antenna, without antenna socket. How long is the omni that you are using? How many elements are inside?

My experience and the experience of other people is that the gain and radiation patterns of antennas and particularly omnis are often wishful thinking. I have seen collinear dipoles radiating skywards rather than flat. Such antennas might have decent gain, but it is not aiming where you want it.

So a good omni has decent gain, good SWR and radiates flat.

Cheers,
Elektra

[Carlos Rey-Moreno]

Hi Elektra,

Thank you very much for your answer.

> The added weather proofing will change the properties of the antenna somehow, because the dielectric constant of the additional material around it is different than free air. You can make a simple test to figure that out - compare the stock antenna with a weatherproofed one. I suppose it will not make much difference, if any.

We did the tests (including the braid not touching anything in the RF
circuit), and the differences are not even perceptible with the tools we
have at hand (horst.sh, wlanconfig).

>> For c), we're going to run some tests the weekend for checking with some
>> spare antennas and pigtails we have if we obtain the 9 dBis before
>> waterproofing them. If not, I'm thinking on using the reflector you
>> designed. Do you think it behaves like a 180º sectoral antenna? or is
>> more a directional one on the 45º angle that you mentioned?
> The gain of the external reflector is nowhere close to the claimed 9dBi. Rather 3-5dBi. But I'd love to get some feedback on the reflector.

For us considering using the reflector I would like to know what is the
sector you're expecting to obtain. because if it is very narrow then I
don't think it will make a difference with the antennas we have (see
below).

>
>
> Here is a interesting web page discussing the gain of collinear dipoles. In order to achive 9dbi / 7 dBd, a collinear dipole antenna requires 6 collinear dipoles of at least 0.1λ end-to-end spacing:
>
> http://www.w8ji.com/stacking_broadside_collinear.htm
>
> In order to provide that gain, the elements and the distance between elements alone adds up to a length of at least ~ 41cm for the antenna, without antenna socket. How long is the omni that you are using? How many elements are inside?

The antenna we have is 39 cm long however it counts only with three λ/4
elements and the maximum-current spacing is 0.64λ in between two of the
them and 0.7λ in between the other two. According to the graphics on the
web page you sent us that should give us in between 4 and 5 dBi gain. In
the test we found this difference or even bigger, however is not at all
omnidirectional and that gain is obtained in a very narrow beam.

So, we're going to do our best with the tools we have (the network is
feasible still), although it seems clear that additional investments in
proper external antennas are needed to obtain the RSSI I was aiming with
these ones.

Best,

Carlos

[Carlos Rey-Moreno]

I forgot to attach a picture from the inner parts of the antenna in case
helps the discussion.

Best,

Carlos

[Elektra]

Hi Carlos -

I have a copy or clone of your antenna here. It was a gift that a Freifunk activist gave me. I gladly took it but was sceptical about the promised performance. I expected it to have at least some gain. I actually never used it so far, so it collected dust on my shelf. Until today. I took it apart, shot a photo and underwent the thingy a simple hands-on performance test.

This antenna type seems to be sold worldwide and offers fantastic gain, of course. Same story like these external computer speakers with 2 Watt power supply and the advertisement "200 Watt PMPO" printed on the package. I took a detailed picture of it, with a tape measure beside, so you can compare it.

The design doesn't make much sense to me. I can see some kind of sleeve element at the bottom of the antenna and a number of 6 radiating elements. The coils seem to count as radiating elements as well.

Each long coil is ~3.1cm long - that is, by strange coincidence, a quarter wavelength at 2.4GHz. With regards to RF the properties of a coil have basically not much to do with its geometrical length in wavelength. It is the number of windings, diameter of the coil, parasitic capacitance and physical length that accounts for the properties of inductance and self-resonance frequency. And these coils have an awful lot of windings for the operating frequency. They look like helical antenna elements for a far longer wavelength to me. They are probably self-resonant somewhere below 1 GHz.

The straight elements between the coils are 0.4 wavelegths long. Which is also strange.

So finally I put the antenna to a simple hands-on test. I used a simple omnidirectional antenna of a TP-Link AP, attached to a TP-Link MR3220, as a reference. The stock rubber ducky TP-Link antenna has a simple sleeve dipole inside. A well known and working design. It just doesn't feature an awful lot of gain - 0dBd.

Together with a friend, who is a HAM, we put the device on the metal roof of a truck and did 10 wireless scans of the site, with each antenna from the same vertical position.

At best the "high-gain omni" picked up 30 WiFi networks, while the stock rubber ducky picked up 42! (By yet another strange coincidence this number is the answer to the final question of the universe in Douglas Adams book series "The hitchhikers guide to the galaxy" ;)

Comparing the readings of signal strength, out of the stations that have been picked up by both antennas, most have several dB better readings with the simple rubber ducky - in the magnitude of 1-8dB better. At best the signal was equal or 1 dB better for the "high-gain" antenna.

So, to be more accurate, despite the larger size and hence larger apperture, the "high-gain" antenna should be advertised with a negative "gain".

Cheers,
Elektra

[Andrés Leopoldo Pacheco Sanfuentes]

Ernie would certainly have liked your conclusion, Elektra! Or is it
Bert the one with the rubber duckies?

I hope the Sesame Street joke doesn't get "lost in translation",
though, since it is a US thing, not sure how "far and wide" the reach
of it is, to put it in antenna-terms:D

Best Regards | Cordiales Saludos | Grato,

Andrés L. Pacheco Sanfuentes
<al...@acm.org>
+1 (817) 271-9619