Ask HN: Sound multilateration/TDOA with observer privacy
Long version: There's an increasing amount of illegal and heavy fireworks being thrown around here over the last years, usually starting somewhere early September and lasting through the winter. Loud bangs, scaring kids and pets and damaging public goods. A few years ago I whipped up a proof-of-concept mobile app that detects a sudden increase in audio levels (BANG!), syncs time over NTP and sends a message to a central server with location and timestamp. The server correlates incoming messages and does basic multilateration/TDOA to determine the location of the fireworks.
I'd like to deploy this on a larger scale, offering the app to others as well, as this gets only effective with a larger number of observers. However, I have concerns over disclosing the location of participant - the usual privacy reasons apply here, but also the fact that pro-fireworks-punks tend to threaten the anti-fireworks people and terrorize their neighborhoods with even more fireworks.
I found one paper describing a mechanism and protocol similar to this ("PrOLoc: Resilient Localization with Private Observers Using Partial Homomorphic Encryption"), but this feels pretty heavy and over my head. Also this assumes the observers know the absolute distance to the target location, instead of having a relative time from a common clock, so I am not even sure if this algorithm applies to my problem.
Any insights, theory or papers on this subject much appreciated!