We’re working on improved router functionality in Meshtastic 1.3, don’t have enough data to make an informed decision and are looking for community feedback to help out.
The new function is to reduce total use of airtime and by extension, allow more nodes to use some of the slower channels.
What I’m looking for is the SNR between various devices when a device receives a packet in various scenarios. Please use the stock antennas that came with your device. Not looking for any speciality configurations, just totally “stock”.
Devices sitting on the same table, right next to each other (at multiple radio configurations – LongSlow/LongFast … etc, whatever you can get)
Devices in the same home with one wall between the sender and receiver (at multiple radio configurations – LongSlow/LongFast … etc, whatever you can get)
Devices in the same neighborhood. One device in your home and another device 30m (100ft) outside your home (at multiple radio configurations – LongSlow/LongFast … etc, whatever you can get)
Really looking for as much data as we can get. Would really appreciate any help we can get here.
You can get this data from the range test plugin, python cli but the easiest way to see the rx_snr is in the android app:
We can get the SNR from the logs (if we added it) but how would we know if they’re in the same room or a room apart?
But you know … the spirit of this does have some merit. We could use data for long-range testing. Maybe we can get that by requesting people’s range test data rather than updating analytics.
What I was thinking of, is that one node will stay stationary and then in the app, you start this “test” wherein you press “table”, “one wall”, “outside” respectively to get a reading, and send it to your analytics thingy.
I really think optional anonymous analytics should be made a part of the apps. Since the apps are open source people can see what data is being sent exactly.
If the exact locations are stripped from the data, and instead, the distance is calculated on device and sent along with SNR, that would give you nice stats without violating privacy.
Interesting, is this doable on iOS and T-Echo, or Android and T-Echo? I was just thinking of a cheat-sheet of different settings and expected results in the form of a diagram with speed vs range. Where 100% range would be long-slow and a “milliseconds” axis for the ping time (or ping/2) that would go in the other dimension.
I can create a Python backend to collect and process this data.
What we need is to collect the device hardware versions, frequency, SNR, and modem settings. I don’t believe we can get a meaningful sample size by asking here on the forum. A dataset from at least 200 users would give nicely rounded numbers.
A few estimates for long range city connections, EU433, LongSlow, one antenna on the balcony/roof.
roof -> inside same building: +9 - +14dB few 100m (street/block) roof to car antenna +4 - +10 dB (non-LOS, part of building in between) 1-2km roof->roof+2 - +6 dB (non-LOS, several rooftops in between) 3-5km roof->roof+3 - +12 dB (LOS) 3-5km roof->roof-19 - -8 dB (non-LOS, part of hill in between)
Long-range LOS and indoor connections are almost distinguishable from the SNR.
One thing I also learned, the radios are pretty much deaf in high-noise urban environments, a cheap SAW chip bandfilter improved the SNR more then 3dB, although it has about 2.5dB insertion loss, a Procom cavity filter improved it around 4-5dB.
Props to everyone who provided this information. We have enough info and this has been used to create an updated routing algorithm and improve Meshtastic.
A YouTube video will be published soon with the results of how this has been used.
