We present an angle-of-arrival (AoA) based RF
sequence recovery system that reconstructs emitter trajectories
from sparse, noisy observations. Our method leverages gridbased mobility graphs with beam search inference to recover
plausible paths under uncertainty. The approach discretizes the
surveillance area into a spatial grid where nodes represent candidate emitter positions and edges enforce mobility constraints. A
beam search algorithm maintains multiple trajectory hypotheses,
updating path probabilities as new AoA measurements arrive.
Experimental evaluation on synthetic trajectories demonstrates
robust reconstruction performance even when observation fractions are low (25%) and AoA measurements are corrupted by
significant noise (σθ = 10◦
). Compared to classical triangulation
methods, our graph-based approach improves median position
error by 25–40% specifically in low-observation-fraction (ρ < 0.5) and high-noise (σθ > 8
◦
) regimes, highlighting its utility
for passive geolocation in contested RF environments where
traditional methods fail.
