#spectraMesh has been working pretty solidly all day today. Much better than yesterday. The issues I fixed this morning around network sync have made a world of difference. Another change I added was the addition of the EU frequency ranges with their own channel definitions and hopping logic.
Next up is message retry logic if messages aren’t acked as accepted by the network. Acks and implicit acks (seeing another node relay your message) are already in the protocol.
A sneak preview of what I’ve worked on today…
If I drop the power lower and place intermediate nodes I can simulate longer distances and force relaying, which is helping with testing multi-hop.
It’s been fun to build this stuff - and I think after today I’m likely going to open this up for alpha testing. I’ve had some intrepid explorers offer to help and I plan on taking them up on that offer.
I do need to add in the UK/EU frequency bands to the TDMA logic first.
Well, after fighting with it all day.... I'm scrapping the firmware for T114. I've had NOTHING but problems with it - not just with SpectraMesh but with Meshtastic and Meshcore too.
I've fought with it all day, but... It just isn't stable.
SO - I'm switching gears and porting the Aurora firmware for the Lilygo TDeck ( which runs MeshCore) to run SpectraMesh.
I *might* also be enjoying an old fashioned. What could possibly go wrong?!
The ADR tweaks were necessary - otherwise you’d be paying a 2-400ms penalty for negotiation for no reason. In cases where multiple packets are needed (small file transfers - an experimental feature) both sides negotiate the best data rate settings they can maintain, switch settings, verify they can still talk, then data transfer happens before they drop back to the base mesh settings.
I did manage to get some additions to the #SpectraMesh protocol, specifically a route trace where each node that passes a message adds its hash to the packet headers, which allows us to do node discovery (find more neighbors), as well as build a routing table that helps us with knowing how to route unicast packets.
I’m also doing some more tweaks to adaptive data rate, ensuring we only do the handshakes and setup in situations where the benefit outweighs the impact of the negotiation itself.
Additionally, I think I’m going to try to port my Aurora firmware for the Lilygo T-Deck from MeshCore over to SpectraMesh. That’d bring the total number of supported devices to 4, which still isn’t a lot but is a start.
At this point I also need help testing this - I can test things all I want in my lab and around my house but more people and different environments will be helpful in tuning network performance.
If interested you can email me or DM me!
Today in “adventures with #SpectraMesh” - I’m going to test automatic location info with my GPS enabled T-114 from Heltec, as well as continue multi-hop testing. I think ADR might need some tuning / tweaking as I think it may be causing some dropped traffic in some scenarios when it shouldn’t be - I’ll have to test with it on and off to see if that’s the cause, then troubleshoot.
I also want to tweak the network join / sync logic to be slightly more resilient.
It took a while but I finally got #SpectraMesh firmware to work for the Heltec T114, to bring us to three boards that have working firmware - one of which has GPS!
I'm contemplating loading SpectraMesh onto the #hackersTownPager with a config setting to select between it and MeshCore for the mesh protocol.
I also want to make firmware for the Lilygo T-Deck.
Anyone with a spare Heltec V3, T114, or Xiao ESP32S3 WIO (ideally more than one...) want to help test this?
#SpectraMesh natively supports position data (manually entered or via GPS/GNSS), and the companion app has a map command that will give you your neighbor nodes and their position relative to your position.
This is helpful if you want to set up directional links and/or plot out your network deployment.
Location data is auto propagated to neighbors (IF you provide it, it is optional), so they need do nothing to make use of it.
There will be options for precise or “general area” fixes.
The Wrew (Pager Monger)