Editing Openai/693e48dd-ce3c-800a-b636-9b3efde0fdec (section)

=== Assistant: Great — this “phone-in-hand” run is actually more boring than the dresser test, and that’s a useful result. ===
Great — this “phone-in-hand” run is actually more boring than the dresser test, and that’s a useful result.

==== ### ====

At discovery, everything is weak and clustered:
* 4210…: -91
* C36F…: -94
* 46B9…: -96 → -98
* EC90…: -96
* 57A8… (at 22:24:53): -94

That’s a classic “ambient devices at the edge of range” cluster.

===== - 4210…: -92, then -89, then back to -91 =====
* 57A8…: -93

So across ~2 minutes, your main target only moved ~3 dB (-92 → -89), and then drifted back.

==== If you walked from dresser → back door → back while holding the receiver, a location-bound transmitter (like a Wi-Fi Light) often produces a bigger RSSI swing (commonly 8–20 dB) because you’re changing distance and obstructions relative to that transmitter. ====

Here, 4210… stayed basically flat.

That usually means one of these is true:
# It’s far enough away that your in-house movement doesn’t change the path loss much (through walls / neighboring unit / outside).
# Multipath dominates (reflections) so the net RSSI looks jittery but not strongly distance-correlated.
# It’s a device with intermittent advertising and you’re just catching it sporadically; the few samples you got aren’t enough to show the full curve.

What it does not look like:
* a device “on you” (you’d expect stronger RSSI at least sometimes)
* a device you walked closer to (you’d expect a clearer rise than 3 dB)

==== Earlier, when the phone was stationary and you returned, you saw 4210… jump into the high -70s briefly. In this run, you never see anything like -77. ====

That suggests the earlier spike was likely:
* a transient favorable orientation/reflection,
* or you were briefly much closer to the transmitter than during this run,
* or the device changed its transmit/advertise behavior.

==== For the device you care about (4210…), you need more frequent RSSI points while you move. ====

Simple method:
* Keep scanning
* Walk slowly for ~20 seconds
* Pause 20 seconds
* Walk back 20 seconds
* Pause 20 seconds

Then copy/paste the 4210… RSSI lines during the whole interval (even if it’s repetitive). With 15–30 samples, we can see if there’s a real movement curve or if it’s just noise.

==== - 4210… behaved like an ambient, not-close device (mostly -89 to -92). ====
* That pattern is consistent with something not in your immediate personal bubble.

If you tell me one thing, I can tighten the inference:
* When you walked to the back door, did you go toward an exterior wall / window, or deeper into the interior?