A user in an other thread pointed out that wifi signals are complicated.
I have "Wi-Fi Sweetspots" and I've been trying to figure out how bad my wireless coverage is but it's a bit wonky. It also reports strength in Mbps.
I've found it hard to translate that into any guesses about where I should put APs or what the actual user experience will be in that location.
Have people used devices like the Olmlmo Emf Meter?
Does something like that actually tell you something useful about wifi coverage in the home and make it easier to plan an effective coverage network?
I use find two Android apps very helpful:
- WiFi analyzer; and
- Ubiquiti WiFiman.
The former is helpful for showing the strengths of SSIDs against channels. The latter is great at tracking transitions between access points in an 802.11r context - in one plot it will show signal strength vs time with a high update frequency and tracks transitions from AP to AP.
Thanks. How do you use those?
Do they allow you reliably predict where you'll need access points? Or is it more of a confirmation when you have other reason to believe that a signal is weak in a particular area?
it is more of a confirmation. Prediction is hard, unless you know exactly what is in your walls, you know already everything about the devices you plan to use, the layout of the rooms and any other factors. I heard even water can dampen wifi signals, so if you place a water bottle in between, you might get (slightly?) worse signal. Do you know radios with external atennas? If you place your hand close to the antenna you will be able to affect the signal and make it temporarily worse or better. I think prediction entails way too much research as compared to the easyness of confirming, if a location is actually good by looking at actual throughput or latency tests.
From what little I understand of radio waves and signal propagation it seems complicated. As near as I can tell we have really good physics and math models of how radio waves behave under ideal conditions. Once we get to the real world, conditions are typically so far from ideal that the actual calculations become computationally infeasible.
Some people do seem to have a knack for it though. I know the ARRL runs contests on that sort of thing and people can develop the skills to get good at that.
I also found https://www.ekahau.com/ That looks really nifty but I suspect that's only worth it if you're making adjustments to APs at the scale of campuses, convention centers or small cities.
It sounds like the best use of these apps is to do some spot checking. ie If my spouse or kids complain that the internet is bad in their room, I can quickly sweep around and see if there's a general problem or if it's likely a problem with the machine.
This app may be handy to visualize the wifi coverage https://www.netspotapp.com/
(I have no personal experience with it though)
just use standard network tools on your smartphone and then walk around and Wifi-benchmark relevant corners in your living space:
- check for ping timeouts with short (eg. 200ms) interval. Areas with ping timeout = signal is too low. Ping with short interval gives you more responsive feedback compared to iperf3, if signal is too low in certain areas.
- determine byte/sec throughput on different spots with any iperf3 client or similar bandwith test tool. Spots with low bandwith = low signal quality or low amplitude
In between make some simple heat-map-like notes on an improvised paper map.
depending on the result, try different AP locations and rinse and repeat. If you often have uploads, also test the reverse direction with iperf3. If you have different AP device types, test them individually. Test different antenna pointing directions, if you think this may help
I was going to mention that they existed and they are one of the few cases where they do attempt to do coverage prediction, but as you've pointed out - they are VERY expensive and so not really an option unless you're a large enterprise shop.
Our applications engineering team has a license and their Sidekick hardware, tbh I wasn't very impressed by the survey report I was provided, but that may have been user error by the apps engineer. The licensing is set up in such a way that I apparently can't run my own reports with the data he collected.
Edit: Also, it's probably overkill for your use case compared to WifiAnalyzer, but a Nuand BladeRF Micro + gr-fosphor is great for interference hunting. I recently used the setup to hunt down an oddball interference source on two 5 GHz channels - turned out to be the last of a set of wireless HDMI extenders that were purchased years ago and have been slowly phased out in favor of a product called ShareLink. Thing emits non-wifi compatible OFDM beacons when idle, and "goes hard" with an OFDM signal that spans two 5 GHz channels when the TV is on.