Wrong Presence and Motion_Level in CSI Multinode Mesh mode

[Mehrab-Anwar]

I am trying to use the CSI Multinode Mesh mode using 3 ESP32 S3.

Built the firmware and provisioned each of the three nodes with a unique node ID (1-3), tdm-total (3) and tdm-slot (0-2). Then on this page: localhost:3000/api/v1/sensing/latest, I see the presence flag and motion level flags are constantly fluctuating. Staying in the room, staying still or moving, leaving the room—nothing makes any difference.
Most of the time presence is true, and the motion level is switching between still and moving. In all cases, the confidence is greater than 0.75. Sometimes it even becomes 1.00.

Health says: {"clients":2,"source":"esp32","status":"ok","tick":139269}

The placement of the ESP32s doesn't make any difference as well. I have tried keeping them next to each other and keeping them a few meters apart, forming a triangle.

The router is placed in the next room, on the wall between the two rooms.

What can be the possible solution for this scenario?

[Velociraptor85]

I have the same setup and I am at the same state essentially also I never get 3 clients although I see them reporting with the debug enabled I have tried nodeIDs starting at 0 and 1 non work the tdm setting also makes sense to me that why I set it but apparantly this also does not do anything.
Further more I have added to one node which i have on my desk an l2410 which recognized but I think I am now more or less at the point where learning would make sense but when I try that it says no traning data found although I have asked the server to do so

Well non the less I am really unsure if all of this really works.

[Kracozebr]

Anybody has working setup? I'm facing the same with 6 esp32-s3. It seems that I watch demo in observability tab not real data from sensors although I use real data:
In Observatory UI → Settings → DATA tab → set WS URL to ws://localhost:3000/ws/sensing → set source to "Live WebSocket"

[aldx8]

BūT I Thing C & K & 3 vRy vRy gooD likē This. 34CLK في الأربعاء، ١ أبريل ٢٠٢٦، ٠٧:٥٢ Kracozebr ***@***.***> كتب:

…

*Kracozebr* left a comment (ruvnet/RuView#348) <#348 (comment)> Anybody has working setup? I'm facing the same with 6 esp32-s3. It seems that I watch demo in observability tab not real data from sensors although I use real data: In Observatory UI → Settings → DATA tab → set WS URL to ws://localhost:3000/ws/sensing → set source to "Live WebSocket" — Reply to this email directly, view it on GitHub <#348?email_source=notifications&email_token=BQYU7LXBHY3HDNTQFWIVAWD4TSOB5A5CNFSNUABFM5UWIORPF5TWS5BNNB2WEL2JONZXKZKDN5WW2ZLOOQXTIMJWG42DIMZXGM2KM4TFMFZW63VKON2WE43DOJUWEZLEUVSXMZLOOSWGM33PORSXEX3DNRUWG2Y#issuecomment-4167443734>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BQYU7LVI3OMKRUP4KCNCDB34TSOB5AVCNFSM6AAAAACXIAGGH2VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHM2DCNRXGQ2DGNZTGQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[Velociraptor85]

I think the fusion tab in the webinterface is the best to see the CSI data and thus find out what you would expect also the lined paper https://ntu-aiot-lab.github.io/mm-fi and the related youtube video is I think an good indication of what one should expect from the data.
Also #333 seems like an honest representation of what could be possible and what is just random numbers. All in all I would say super interessting project and idea but I am unsure if the raised expectation about the pose and relative position in the room is actually available in the csi data 🤷