adventures in wifi

[solarbird]

Yesterday, I got poking around the Lair’s wifi with some signal analysis tools. The interference and terrible signal to noise ratios I had to fight in the recording studio are just as bad in wifi, if not worse. It’s really terrible.

But check this out – I think I’ve sussed part of it. In the illustration below, the red bar is the giant cement retaining wall. The bar is not actually to scale, sorry about that – it should be thicker, because it contains a lot of rebar. The gradients of colour are wifi strength, from a nearby hotspot which is not ours. I picked this one because it shows the effect best, but it shows up in imagery of other transmitters as well.

Do you see what’s going on here? The rebar in the retaining wall appears to be acting as a crude parabolic reflector. This relatively-hot-spot is showing up in all of the maps, pretty clearly, except for the ones where that area shows up as a shadow of reduced strength. I think those are signals from transmitters from the other side of the wall.

I mean honestly, look at this. Am I wrong? This is so neat. And I’m wondering if this is the cause of some of our other interference problems as well, like possibly even the BBC-World-Service-on-the-house-mains issues.

All sorts of wireless things act very strangely here. Even AM/FM radio. And I’m starting to wonder if there’s a way to improve the grounding on the rebar. I can’t imagine how, but still.

---

If you’re looking for the Grammy Awards Long List nominees, thank you for listening, and for your consideration.

Mirrored from Crime and the Blog of Evil. Come check out our music at:
Bandcamp (full album streaming) | Videos | iTunes | Amazon | CD Baby

Comments

[vatine]

If the rebar is conductively tied in (most) of the wall, it should be enough to drill a small-ish hole through the concrete to the rebar and tie in another ground strap.

I suspect that isn't the case, though. And at that point, I am not sure what to do. Coat it with a grounded copper film?

[solarbird]

Exactly, right? I've no idea. I mean, not 1000% no idea, but... very little idea.

So what I am doing is realising that's a thing, and going around setting up to take advantage of it as a listening post putting ferrites or other noise filters on everything that plugs in, since hey, I may as well reduce the number of functioning antennas, and doing better/more in-depth wifi surveys to see if I can get the wifi part of the network under some sort of control.

I'm also hoping that having a bunch of ferrites on everything plugged in will reduce total noise picked up by the core wiring just by providing pseudogrounding for it. I don't actually think that's entirely wrong, but I do think it has a lot of wishful thinking.

[vatine]

But! Drill-probing a concrete wall and checking sounds like fun. And it may well be possible to check for galvanic bonding without probing (thinking aloud starts in next paragraph).

Have a suitable sinus wave, have it close to one rebar (rebar finders are a thing, right?), then probe for its presence elsewhere close to rebar at different height. If there's good coupling, that should somehow be detectable, I think.

Actually, drilling sounds easier. But, you have something to do Exciting!Science!Engineering! on.

[arethinn]

What tools were you using? Is it something I could use to do similar analysis in my own home or is it likely to be beyond my expertise?

[solarbird]

I'm not doing anything hugely difficult, I don't think. The mapping tool is NetSpot, which is free in the OS X app store. It's very good for mapping; it's terrible for live monitoring. It also pitches you a higher-expense full-featured version, as well as individual downloadable features, but does so in a non-aggrivating way.

I've also been using WiFi Scanner, which is much better for things like channel overlap and live monitoring, which means I can make changes and see what happens within a few seconds. That's also in the OS X app store. This app cost me $15, but I think it's money well spent. Between the two, I've done a lot of good for our LAN.

I don't know that it's beyond anybody. The hardest part might be just drawing maps of your building for NetSpot. I didn't even have to draw them, thanks to having the blueprints. NetSpot has a built-in editing tool to let you draw your own maps, but the tool is amazingly terrible, so don't try. Draw them in something else and save them as image files. Make sure you get it to scale. You'll also need to know the size of the area you're mapping, or at least a meaningful part of it, because once you've loaded a map, it will ask you to pick two points on it and tell it the distance between.

Then get a lot of sample points, let it make the maps, and check the individual points as well. Look at both total signal strength and signal-to-noise ratios. It's really pretty straightforward - the colour field is great for people who aren't good at numbers and gives you an immediate and not-entirely-misleading overall view. It's a nice tool.

One other thing I found was to look up the antenna positions in your router. I got 10db better on one end of the building (with no penalty anywhere else!) by tipping one of the master wireless hubs on its side. No, really. It was because of where the antennae were inside the hub. Big, reproducible, immediate difference. That's the application for WiFi Scanner (or something like it) - immediate feedback when testing changes.

So yeah, overall, definitely worth it, and worth the $15.

[arcanetrivia]

Eh, sounds like it's beyond my hardware, then. The only Apple device we have is a first-gen iPad. I suppose similar things probably exist for Windows though.

[solarbird]

None of this is taxing on hardware. You just need to find the Windows equivalents. They absolutely exist. Hell, there might be Windows versions of these specific apps.

[arethinn]

I just meant anything requiring OS X wasn't going to be possible, since we don't have anything that runs it. (Although it's also true that a lot of things in the app store also don't run on our iPad of very little brain, since it can't go higher than iOS 5.)