OpenWrt Forum Archive

Topic: Update on Linksys WRT1900AC support

The content of this topic has been archived between 16 Sep 2014 and 7 May 2018. Unfortunately there are posts – most likely complete pages – missing.

honicky wrote:

Hi all,

I know this has been asked before, but I'm not sure what the current state of affairs is.

I am trying to control the tx power on my WRT 1900ac.  I'm doing this for some lab testing, and need to be able to set the tx power (for example using iw dev wlan1 set txpower fixed 10mBm).

When I use the above command and look at the output on the spectrum analyzer, I see no change in output power.

More alarmingly, I have built (for lab use) a special reg-domain build that has country codes that correspond to different max powers:

...

country 29:
        (2400 - 2483 @ 40), (29)
        (5140 - 5860 @ 160), (29)

country 30:
        (2400 - 2483 @ 40), (30)
        (5140 - 5860 @ 160), (30)

So, I can set the max txt power to 15dBm by simply doing

iw reg set 15

This also fails to change the TX power at all!  This implies to me that the 1900ac is ignoring the regulatory domain settings!

Am I doing something wrong here?  Is there support for controlling the TX power, or even properly setting the max power for the purposes of regulatory compliance???

Thanks a lot in advance...

With the FCC getting pissed off; I hope you plan to do your testing in the great plains of North America (Out west in the middle of a section of land (One square mile / 2.6 square kilometres - 640 acres / 260 hectares) and not in a city or near an airport but in a Faraday cage or a shielded room.

It would seem to me that in 2015/2016 and earlier there would be a requirement in firmware to SET the country by looking at the IP address of the provider. In North America it is really simple; Canada, the United States & Mexico. In Europe where the countries are small; it would be easier to "pretend that you are from a different country". A table of ITU (International Telecommunications Union) assigned IP address by country would not be very large.

In Southern Ontario, FCC radio interference open field testing is usually done between Kitchener, Ontario and London, Ontario where interference is minimal. I've driven 2 1/2 hours west of Toronto to be there for the tests.

You do realize that while your "statement" may be true; it may be that you are simply trying to get higher power that you are NOT entitled to use; unless precautions are in place.

This question may NOT get answered even if the answer is known.

Rick

Edit: IF you know how to circumvent the country where you live; I STRONGLY SUGGEST that you don't post.

IF the FCC bans open source firmware in the US; the world will lose.

We need a world with the US more like the EU OR an EU more like the US to get some balance. God help us if the EU were states and the President was from ???.

(Last edited by RickStep on 12 Jan 2016, 22:38)

Hi Rick,

Thanks for your response. I don't want to do anything to jeopardize openwrt's standing with the FCC, and I understand your concern.  I'm working in a RF lab where we develop 5GHz equipment.  I am operating with cabled links only, and through 30dB of attenuation, so the only emissions we have will be way below the noise floor.

In any event, I actually don't need to transmit at a higher power: I'm trying to figure out how to reduce the power being transmitted in a predictable way.   I can control the max power just by adjusting the attenuation on the wire, so what I'm looking for is a way to set the tx power below whatever the max regulatory power is.  I'm perfectly happy to have the max power be whatever is prescribed by US regulatory domain.

I would prefer not to fiddle with the regulatory domain settings.  That was just a hack I used to try to accomplish the above goal.  If I could just use the "iw" command in the standard way, I'd prefer that.

With all that said, I have seen before that the drivers for the 1900ac may or may not support setting the TX power.  Can you or anyone else please update me on the status of that?  And if it is possible, what is the mechanism?

Thanks again in advance!

honicky wrote:

Hi Rick,

Thanks for your response. I don't want to do anything to jeopardize openwrt's standing with the FCC, and I understand your concern.  I'm working in a RF lab where we develop 5GHz equipment.  I am operating with cabled links only, and through 30dB of attenuation, so the only emissions we have will be way below the noise floor.

In any event, I actually don't need to transmit at a higher power: I'm trying to figure out how to reduce the power being transmitted in a predictable way.   I can control the max power just by adjusting the attenuation on the wire, so what I'm looking for is a way to set the tx power below whatever the max regulatory power is.  I'm perfectly happy to have the max power be whatever is prescribed by US regulatory domain.

I would prefer not to fiddle with the regulatory domain settings.  That was just a hack I used to try to accomplish the above goal.  If I could just use the "iw" command in the standard way, I'd prefer that.

With all that said, I have seen before that the drivers for the 1900ac may or may not support setting the TX power.  Can you or anyone else please update me on the status of that?  And if it is possible, what is the mechanism?

Thanks again in advance!

Two things:

1. The FCC could/would likely grant you a licence, did you apply for one. Improving communication with less power is what will be needed going forward as more and more devices clutter the airwaves.
2. You could reach out to some on this board via the private messaging system to allow messages to go to individuals. They MAY ask you to prove where you are; university or other or company.

If you are working on a system to automatically adjust the TX power for reliable communication; by default that would most likely reduce the power, BUT could move the power up and out of range, which could be allowed with an FCC licence.  It needs to be known how often connections drop and wouldn't drop if there was more power.

The question then is IF the connection is poor; do you increase the power OR find another channel.

The FCC could allow emergency services to INCREASE POWER and us mere mortals to allow the system to find another channel or we get dumped.

Good luck,

Rick

(Last edited by RickStep on 12 Jan 2016, 23:00)

RickStep wrote:

With the FCC getting pissed off; I hope you plan to do your testing in the great plains of North America (Out west in the middle of a section of land (One square mile / 2.6 square kilometres - 640 acres / 260 hectares) and not in a city or near an airport but in a Faraday cage or a shielded room.

It would seem to me that in 2015/2016 and earlier there would be a requirement in firmware to SET the country by looking at the IP address of the provider. In North America it is really simple; Canada, the United States & Mexico. In Europe where the countries are small; it would be easier to "pretend that you are from a different country". A table of ITU (International Telecommunications Union) assigned IP address by country would not be very large.

so what country code is mandatory for 192.168 IP addresses? or 10. IP addresses?

RickStep wrote:

The question then is IF the connection is poor; do you increase the power OR find another channel.

Increasing power, like slowing down the data rate, is the right thing to do if the connection is poor because of a weak signal. But if the connection is poor because of interference from other signals, both are the wrong thing to do because they just cause more interference.

This is why the right thing to do if you need to support a lot of users in a given area is to turn DOWN the power and disable the slower data rates (and have more access points). https://www.usenix.org/conference/lisa1 … d_wireless

back to the topic at hand, I've heard before that the 1900/1200 driver always does it's broadcasts at full power, even if the power setting is lower for communications. You may want to double check if this explains that honicky is seeing.

dlang wrote:
RickStep wrote:

With the FCC getting pissed off; I hope you plan to do your testing in the great plains of North America (Out west in the middle of a section of land (One square mile / 2.6 square kilometres - 640 acres / 260 hectares) and not in a city or near an airport but in a Faraday cage or a shielded room.

It would seem to me that in 2015/2016 and earlier there would be a requirement in firmware to SET the country by looking at the IP address of the provider. In North America it is really simple; Canada, the United States & Mexico. In Europe where the countries are small; it would be easier to "pretend that you are from a different country". A table of ITU (International Telecommunications Union) assigned IP address by country would not be very large.

so what country code is mandatory for 192.168 IP addresses? or 10. IP addresses?

THOSE ARE NOT WAN addresses but LAN addresses.

MY post was about the addresses assigned to the Internet Provider.

Check this link:

http://www.nirsoft.net/countryip/

RickStep wrote:
dlang wrote:
RickStep wrote:

With the FCC getting pissed off; I hope you plan to do your testing in the great plains of North America (Out west in the middle of a section of land (One square mile / 2.6 square kilometres - 640 acres / 260 hectares) and not in a city or near an airport but in a Faraday cage or a shielded room.

It would seem to me that in 2015/2016 and earlier there would be a requirement in firmware to SET the country by looking at the IP address of the provider. In North America it is really simple; Canada, the United States & Mexico. In Europe where the countries are small; it would be easier to "pretend that you are from a different country". A table of ITU (International Telecommunications Union) assigned IP address by country would not be very large.

so what country code is mandatory for 192.168 IP addresses? or 10. IP addresses?

THOSE ARE NOT WAN addresses but LAN addresses.

MY post was about the addresses assigned to the Internet Provider.

Check this link:

http://www.nirsoft.net/countryip/

And if you are putting the AP inside the WAN connection you will still only get a private address.

(Last edited by dlang on 12 Jan 2016, 23:19)

dlang wrote:
RickStep wrote:

The question then is IF the connection is poor; do you increase the power OR find another channel.

Increasing power, like slowing down the data rate, is the right thing to do if the connection is poor because of a weak signal. But if the connection is poor because of interference from other signals, both are the wrong thing to do because they just cause more interference.

This is why the right thing to do if you need to support a lot of users in a given area is to turn DOWN the power and disable the slower data rates (and have more access points). https://www.usenix.org/conference/lisa1 … d_wireless

back to the topic at hand, I've heard before that the 1900/1200 driver always does it's broadcasts at full power, even if the power setting is lower for communications. You may want to double check if this explains that honicky is seeing.

I agree; but you need to BURN all of the old routers and replace them with new ones.

First you would need a new router that ONLY transmits at a power to KEEP the connection from dropping.  Today that device does NOT exist.
Second you have to REMOVE ALL of the old routers.

Microsoft with Windows 8.0, 8.1 and 10 consigned millions of printers to the scrap heap BECAUSE Brother, Canon, Dell, Epson, HP, Lexmark, XEROX and others REFUSED to update drivers.

Somehow we need a way to remove routers/modems that over power the transmitter and create havoc over many city blocks in crowed communities.

That would most likely kill the Linksys wrt54 group. Maybe it finally needs to be turned OFF.

Mandated software built into Windows, Apple and Unix (and derivatives) and ALL others could fix this by REFUSING to allow any connection to a banned device.

A reasonable conversation close to someone is nice (my wife); ALWAYS shouting to get a message 3 feet (1 metre) is absurd (over powered routers). My wife would move to the spare bedroom or leave home.

Responses are always allowed.

Rick

dlang wrote:
RickStep wrote:
dlang wrote:

so what country code is mandatory for 192.168 IP addresses? or 10. IP addresses?

THOSE ARE NOT WAN addresses but LAN addresses.

MY post was about the addresses assigned to the Internet Provider.

Check this link:

http://www.nirsoft.net/countryip/

And if you are putting the AP inside the WAN connection you will still only get a private address.

YOU ARE MISSING HOW THE INTERNET WORKS!

The WAN connection in the modem connects to the backbone via the INTERNET SERVICE PROVIDER'S IP address NOT any crap in house addresses (192.168.0.1, 192.168.1.1, 192.xxx.xxx,10.xxx.xxx.xxx)

For someone that posts here regularly AND has many smarts with Linux; YOU are way OFF BASE with the difference between WAN addresses and LAN addresses. I strongly suggest that you learn the difference.

You are heading toward making yourself an ASS.

WAN address allocation by the ITU WHICH are never reused (Cannot be reused by another Country) are NOT the same as a 100 million routers using 192.168.0.1 LAN addresses for local address.

You REALLY MISSED THIS!

Perhaps you were tired and missed this. I, now retired, made silly mistakes when sleep deprived.

Can you resend your post.

Rick

so...  if 4.4rc7 in now 4.4.0

what is the bleeding edge at?

Cheers

Before you fly off the handle Rick, I think he was implying the situation where the AP may be separated from the WAN (e.g. Purely facilitating WLAN). Without a WAN IP that is country oriented, how would you apply the appropriate country code in your aforementioned scheme?

Lantis wrote:

Before you fly off the handle Rick,

To late wink

It's okay Rick.... I fly off the handle when I see how people are going to create secure networks by using vlans wink

Gets me just about every time... lol

I think trunk is still at 4.1, have not seen a Makefile change come through.
I think dlang point is something along the lines of pulling your AP out of your suitcase while travelling and connecting to your hotel lan with a non-routable IP; but I could be wrong. Now, please tell me all about correction lines to go along with the Section description.

trunk got bumped to 4.1.15 with r48223, 4.4rc7 to 4.4.0 with r48203

Is anyone else seeing latency on either/both radios?  I can reproduce at will....

Cheers

Ya, I just meant trunk has not yet been bumped to 4.4. I have put an image with 4.4 final today r48221, but I am still playing with parameters in /etc/config/wireless.

dlang wrote:
RickStep wrote:

The question then is IF the connection is poor; do you increase the power OR find another channel.

Increasing power, like slowing down the data rate, is the right thing to do if the connection is poor because of a weak signal. But if the connection is poor because of interference from other signals, both are the wrong thing to do because they just cause more interference.

This is why the right thing to do if you need to support a lot of users in a given area is to turn DOWN the power and disable the slower data rates (and have more access points). https://www.usenix.org/conference/lisa1 … d_wireless

back to the topic at hand, I've heard before that the 1900/1200 driver always does it's broadcasts at full power, even if the power setting is lower for communications. You may want to double check if this explains that honicky is seeing.

@dlang thanks.  I am actually operating in a RF lab for the purposes of testing the equipment we are manufacturing for interoperability and other things like that.  I am generating wifi traffic on wired links using the WRT1900acs and need to be able to control the power for some of our test cases.  So, while proper power control *is* important in normal WiFi channels, my channel is not at all normal smile   @RickStep Yes, we have all the licensing taken care of.

Regarding broadcasts, I'm generating TCP traffic over the link, so very few of the MAC packets should be broadcasts.  I'm not familiar enough with WiFi to know what percentage of the PHY packets should be broadcasts, but I tested your hypothesis by looking at the time domain signal being transmitted.  I don't see different packets with different powers as I would expect to see if a few broadcast packets were transmitting at a high power, while the rest were at a lower power.

Is there someone on this forum who's an expert on the WRT1900ac drivers that I could PM talk with about this to figure out if setting the transmit power is possible, and if not how I could fix the driver so that we can do that?  Again, I'm not interested in violating the regulatory constraints, only in operating below the maximum regulatory power.

Thank!

Perhaps try irc #openwrt

@anomeome Thanks, I'll do that.

RickStep wrote:

I agree; but you need to BURN all of the old routers and replace them with new ones.

First you would need a new router that ONLY transmits at a power to KEEP the connection from dropping.  Today that device does NOT exist.
Second you have to REMOVE ALL of the old routers.

Microsoft with Windows 8.0, 8.1 and 10 consigned millions of printers to the scrap heap BECAUSE Brother, Canon, Dell, Epson, HP, Lexmark, XEROX and others REFUSED to update drivers.

Somehow we need a way to remove routers/modems that over power the transmitter and create havoc over many city blocks in crowed communities.

That would most likely kill the Linksys wrt54 group. Maybe it finally needs to be turned OFF.

Mandated software built into Windows, Apple and Unix (and derivatives) and ALL others could fix this by REFUSING to allow any connection to a banned device.

A reasonable conversation close to someone is nice (my wife); ALWAYS shouting to get a message 3 feet (1 metre) is absurd (over powered routers). My wife would move to the spare bedroom or leave home.

Responses are always allowed.

Rick

no, you don't need to burn all existing routers and make new operating systems not talk to perfectly functional devices.

bumping things up to higher power is like talking louder in a crowded room to be heard, it doesn't work well, but saying that you have to shoot everyone who's loud doesn't solve the problem either.

There are always going to be anti-social people who turn it up because they either don't know any better, or don't care about anyone else. You just have to live with it and not freak out.

You have to start by educating people that the right thing to do is to turn power down, but use more APs. This actually works pretty well even if you have noisy people in the area. You can then educate your neighbours and get them to turn power down as well. 5GHz and MU-MIMO will help (if done right).

2.4GHz wifi was never designed or engineered for the volume of people using it. I remember when the laptop PCMCIA cards were $1500 each. At that price, the number of people using it was very low, and the system worked well. With only three usable channels, there is just plain not enough bandwidth at 2.4GHz for what everyone is trying to use it for (a dozen people in a house streaming 4k video to their phones)

RickStep wrote:
dlang wrote:
RickStep wrote:

THOSE ARE NOT WAN addresses but LAN addresses.

MY post was about the addresses assigned to the Internet Provider.

Check this link:

http://www.nirsoft.net/countryip/

And if you are putting the AP inside the WAN connection you will still only get a private address.

YOU ARE MISSING HOW THE INTERNET WORKS!

The WAN connection in the modem connects to the backbone via the INTERNET SERVICE PROVIDER'S IP address NOT any crap in house addresses (192.168.0.1, 192.168.1.1, 192.xxx.xxx,10.xxx.xxx.xxx)

For someone that posts here regularly AND has many smarts with Linux; YOU are way OFF BASE with the difference between WAN addresses and LAN addresses. I strongly suggest that you learn the difference.

You are heading toward making yourself an ASS.

WAN address allocation by the ITU WHICH are never reused (Cannot be reused by another Country) are NOT the same as a 100 million routers using 192.168.0.1 LAN addresses for local address.

You REALLY MISSED THIS!

Perhaps you were tired and missed this. I, now retired, made silly mistakes when sleep deprived.

Can you resend your post.

Rick

Sorry @RickStep, but YOU are wrong.  Those addresses are perfectly valid WAN addresses from the point of view of a router.  I believe your error is based on two false assumptions:
#1 - That OpenWRT devices will always be plugged directly into a cable/dsl/whatever modem and will recieve an IP from the ISP
and
#2 - That all ISPs assign their clients publicly routable IPs.

In regards to #1, though I'm hard pressed to come up with a plausible reason why someone would use another router as the primary point of access to their network and use a nice fancy high powered unit the WRT1900AC as a simple access point, I'm sure someone else could...  The best I can think of is using a high end professionally managed security solution (think Barracuda or FortiNET) for a router with another router behind it as an access point (DHCP, NAT, firewall, etc turned off).  Granted, that's a bit of a stretch, especially for a home user, but it IS possible, and it would result in the OpenWRT router/AP getting a private IP.

In regards to #2, lots of ISPs don't necessarily give public IPs to their end users, and lots that do charge extra for it.  Around here, afaik, all cellular data connections (and yes, OpenWRT routers can be hooked up to use cellular as WAN) are served 10.* addresses and NAT'd to the wider internet.  Both of my parents have (rural) wireless internet.  My father is on satellite, and my mother uses a wireless ISP similar to cellular (but different, it uses a DOCSIS cable modem and dish style directional antenna to connect to the tower).  Two different ISPs, yet both are served 10.10 addresses and NAT'd to the internet.

I'll grant that in general your solution could work, but @dlang definitely has a valid point that it isn't a guarantee, or a cure-all.  Further, I'd wager that situation #2 becomes more and more common as we exhaust our supply of IPv4 addresses, given how slow ISPs and backbone providers are being upgrading their equipment to support IPv6.

RickStep wrote:

WAN address allocation by the ITU WHICH are never reused (Cannot be reused by another Country) are NOT the same as a 100 million routers using 192.168.0.1 LAN addresses for local address.

You REALLY MISSED THIS!

Perhaps you were tired and missed this. I, now retired, made silly mistakes when sleep deprived.

Can you resend your post.

Rick

Rick, I know how networks work. But unless you are the edge router, you don't have any way of knowing what the WAN address (or addresses in some cases) are for your network.

If I have a modem from my ISP that gives me a 192.168.3.x address that I then plug the WAN port of my AP into, the AP is going to see 192.168.3.x, not the WAN address of my network. In business settings, it's extremely common for the APs to be connected to an internal network instead of directly to the WAN.

There are also ISPs that give people 10.x IP addresses as the WAN addresses of their home network and then do NAT inside the ISP to connect to the rest of the world.

If you try to find out by connecting to a server on the Internet and asking it what address it sees you comeing from, you still don't know your WAN address, because you may be getting there through a VPN. It would be easy to setup a VPN on the edge router that directs your traffic to some other country so that your 'WAN address' shows up at that other country.

This is not a problem that technology can solve. Even if a GPS was installed in every AP, there would still be times when GPS signals are blocked and so a location can't be found.

It's also very possible to have layers of NAT so that you have an IP range inside your network that the wan port of your AP connects to that 'belongs' to some other country, and then gets NATed to a IP range that's legal for your country. Doing so would make it so that there are 4 IP addresses on the Internet that you can't reach, but it's easy to pick ones that you don't care about (and change them if you find you do care)

Also, the prohibition against re-useing addresses isn't as solid as you are thinking. For example, Google's 8.8.8.8 DNS address exists in many different datacenters, in many different countries (search for anycast addresses for how and why people make use of a 'loophole' in IP routing to legitimately use the same IP address multiple places in a network)

CalvinF wrote:

In regards to #1, though I'm hard pressed to come up with a plausible reason why someone would use another router as the primary point of access to their network and use a nice fancy high powered unit the WRT1900AC as a simple access point, I'm sure someone else could...  The best I can think of is using a high end professionally managed security solution (think Barracuda or FortiNET) for a router with another router behind it as an access point (DHCP, NAT, firewall, etc turned off).  Granted, that's a bit of a stretch, especially for a home user, but it IS possible, and it would result in the OpenWRT router/AP getting a private IP.

any time you need more than one AP to cover a building (think basement to attic in a house, or even end to end in a large house), one of the APs is going to be inside another router (and possibly both APs will be inside a third router)

There are many ISP's that route RFC1918 addresses on the network, and do assign those type of addresses to outside interfaces on CPE. However, those addresses are at some point converted to a "outside" address when crossing ISP egress. In these situations, the ISP is just another private LAN.

There are also situations where there are point to point circuits, on the internet, that have RFC1918 addresses, but those addresses are just valid between points. VPNs can also have private addresses between two points, but are within the tunnel, and not routed on the internet.

I don't agree on much of anything RickStep talks about, but do understand the spirit of what he said.

If you doubt that private addresses can not be routed on the internet, there's a simple test to prove that at some point it gets converted to a "outside" address.  >>   http://ipchicken.com/

You should never see a 192.168.x.x 10.x.x.x or 172.16.x.x - 172.31.x.x being displayed on the "outward facing" ipchicken smile

davidc502 wrote:

There are many ISP's that route RFC1918 addresses on the network, and do assign those type of addresses to outside interfaces on CPE. However, those addresses are at some point converted to a "outside" address when crossing ISP egress. In these situations, the ISP is just another private LAN.

There are also situations where there are point to point circuits, on the internet, that have RFC1918 addresses, but those addresses are just valid between points. VPNs can also have private addresses between two points, but are within the tunnel, and not routed on the internet.

I don't agree on much of anything RickStep talks about, but do understand the spirit of what he said.

If you doubt that private addresses can not be routed on the internet, there's a simple test to prove that at some point it gets converted to a "outside" address.  >>   http://ipchicken.com/

You should never see a 192.168.x.x 10.x.x.x or 172.16.x.x - 172.31.x.x being displayed on the "outward facing" ipchicken smile

Fair enough, I'll give you that, maybe I took the comments about the modem and router WAN IP too literally. wink 

A router or modem could use a service like ipchicken.com or http://icanhazip.com/ to determine the ultimate public ip, but that's a whole 'nother layer of complexity from looking at the WAN address of the router/modem.   Even then, such a system could be easily subverted by anyone looking to do so with some rather simple iptables trickery, (unless you're specifically talking about a modem, but that doesn't apply here at all as this is an OpenWRT discussion forum and OpenWRT doesn't support modems.  Even if the router part works, the modem itself won't).  It wouldn't be difficult to detect such requests and re-route them to an internal server to return a false ip.

And that still wouldn't address the further issue of air gapped networks, networks outside of local ISP coverage areas, or networks configured to look that way to a router/ap (using iptables again, in this case  to simply block the request).

Taken literally, though, I still say he's wrong about never having private addresses on WAN ports.  In fact, MY router currently has a 172.16.x.x IP on it's WAN port.  tongue

Either way, I completely disagree with using the WAN/Public IP to enforce wifi regulatory settings.  There are too many edge cases to take into account.  It would be extremely difficult to do properly, if it's possible at all, and it's way more complex that just looking at the WAN IP.  I also disagree with most of what else he's said, but others have adequately addressed those issues.

(Last edited by CalvinF on 13 Jan 2016, 04:37)

Sorry, posts 9551 to 9550 are missing from our archive.