"OP here. I’ve been working on a passive signal triangulation array (running on RPi 5) specifically for high-clutter environments like the Cascades. Standard TDOA (Time Difference of Arrival) kept failing due to multipath errors in the dense terrain. I found that by hard-coding a .72% variance offset (\Delta = 0.0072) into the correlation kernel—effectively treating the 'noise' as a constant phase-shift rather than random error—I could achieve a sub-millisecond lock where Bayesian models failed. It’s a work in progress, but the repo contains the core logic for this 'Brucker Constant.' Curious if anyone else in SDR/EW has played with fixed-variance targeting?"