New tool detects satellite rendezvous in public catalogs
PairWatch uses satellite pairs and motion symmetry to find adversarial maneuvers.
Researchers built PairWatch to find adversarial rendezvous and proximity operations (RPO) in public satellite catalogs. This tool looks at satellite pairs instead of treating each satellite in isolation.
Existing detectors often mistake routine station-keeping for adversarial behavior. They also miss RPO when the target stays passive.
PairWatch is a pair-aware AnomalyTransformer. It uses an attention prior augmented with a sign-symmetric geometric similarity over relative-motion features. This uses the kinematic symmetry between approach and recede.
The researchers tested the tool using an open benchmark of 359 windows from public Space-Track data. This data includes 11 documented RPO events and 100 Clohessy-Wiltshire synthetic adversarial windows. It also includes a five-class hard-negative set including the TerraSAR-X / TanDEM-X helix formation.
PairWatch is the only model that succeeds using a single threshold to separate real RPO from operational negatives. This property held in 100% of bootstrap resamples. Other neural baselines scored between 0-19.5% and four physics-informed detectors scored under 0.1%.
Simpler baselines like LSTM-AD attained higher overall AUROC by shortcutting a synthetic-vs-formation contrast. This shortcut has no operational meaning and fails the security-relevant task.
Key facts
- PairWatch uses satellite pairs and kinematic symmetry to detect RPO.
- The tool succeeded in 100% of bootstrap resamples using a single threshold.
- The benchmark includes 11 documented RPO events and 100 Clohessy-Wiltshire synthetic windows.