Acoustic triangulation of accelerometers in upper stage helped pinpoint the strut, and using only 0.893 seconds of data (unless the accels were in the part that kept transmitting). That means the streaming sample rate must be quite high!
assumong 330m/s for the speed of sound, you can get 1 inch of precision at a sample rate of just under 13kHz. With 3 channels of 12-bit accelerometer data that's ~ 58kBps with no compression which doesn't seem too out there. Of course you can get a pretty close approximation at much lower sample rates (/2, /4, /8, /16) using band-limited interpolation
helium is 972 m/s, aluminum 5100 m/s, steel 6100 m/s. depending on what they actually measure, correct triangulation sounds like a non-trivial problem; good job spacex for solving that. (unless your cad software does that for you :))
