From: Starlink [mailto:starlink-bounces@lists.bufferbloat.net] On Behalf Of dan via Starlink
Sent: Monday, November 13, 2023 3:50 PM
To: Dave Taht
Cc: Dave Taht via Starlink; Network Neutrality is back! Let´s make the technical aspects heard this time!
Subject: Re: [Starlink] bluetooth occupancy sensing

 

I also have bluetooth occupancy sensors lol.  We have a lab test going of triangulated bluetooth tag locating.  So you put a BLE tag on a device and *3* or more receivers in a space at different locations and heights which are documented and then BLE tags are triangulated based on RSSI. 

[RR] This only works if ALL the sensors are synchronized. Adding more sensors does not help either, because for every sensor you add that’s not synchronized, you need to estimate a time bias/offset.  And if the oscillators are not synchronized (in frequency), then you have another parameter to estimate … the frequency offset.  It can get out of hand really fast! The way this is “normally” done is to “calibrate” the area around the sensors.  What that really means is that “array manifold vectors” are collected during a calibration phase then used during the measurement phase to “locate the transmitters” and even that is a “tricky problem”.  Check out MUSIC!!! It’s a half-century old!  

Having sensors at various heights allows for tracking even the 'z' axis.  These tags are very cheap, you can buy complete tags for a couple of bucks, don't even have to build your own, and you can get them built into cutable (or non-cuttable) wristbands.  You can also do short-term tracking of cell phone beacons, though privacy mode means that you only get a short 'session' with a phone (because of privacy mode on newer phones) that isn't paired with something but if you have a phone with a bluetooth headset, the 'locks' the bluetooth mac address and now you can track the phone anywhere that the bluetooth headsets follow.  You can also track cars which don't scramble the mac, but you get cars with wifi mac as well.

We can get bluetooth to within inches accurate when it's line of site. 

[RR] This  requires sub-nanosecond synchronization … remember it’s a nanosecond/foot (the inverse of the speed of light that is!) JJ

 In a pocket or something it's about a meter because bodies/clothes etc reduce RSSI unevenly.  The purpose of this is a couple of things, 'patient tracking' in any sort of a facility like nursing home or hospital, and device tracking, again in a facility with shared hardware like portable EKGs and handheld XRays etc that get 'misplaced' and staff has to go on a hunt for.  It's also much cheaper than lorawan as BLE transmits many times a second and runs for years while lora is built for more range and only transmits intermittently, usually 10-60 minutes to preserve battery.

We're testing mainly on dragino and milesight devices.  I'm also having decent enough luck with mikrotik's knot which can track BLE beacons with high enough precision.  Mikrotik has their own somewhat expensive BLE beacons also but these are basically universal.

[RR] If you are getting inch accuracy without addressing the synchronization problem, I’d love to hear how, especially when there are let’s say 10 BT transmitters to “locate simultaneously”. JJJ