We propose a hybrid localization framework that
fuses soft angle-of-arrival (AoA) distributions with timedifference-of-arrival (TDoA) refinements for improved emitter
geolocation. Unlike classical triangulators that threshold AoA
estimates into hard lines-of-bearing, our method leverages full
AoA logits as probabilistic evidence, then applies TDoA-based
refinement to resolve ambiguity and reduce geometric dilution
of precision (GDOP). Monte Carlo simulations show consistent
accuracy gains, with up to 40% RMSE reduction compared to
AoA-only triangulators under noisy conditions. The approach
scales to multi-emitter settings and arbitrary sensor configurations, providing a flexible and computationally efficient fusion
mechanism suitable for radar, GNSS, and passive surveillance
applications.
